CN113987296B

CN113987296B - Solution detection method and device for application questions

Info

Publication number: CN113987296B
Application number: CN202111389102.3A
Authority: CN
Inventors: 生士东
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2021-11-22
Filing date: 2021-11-22
Publication date: 2024-06-11
Anticipated expiration: 2041-11-22
Also published as: CN113987296A

Abstract

The application provides a solution detection method and device for application questions; the method comprises the following steps: acquiring the stem information of the application questions and the answer information of the application questions; the stem information at least comprises: at least one reference value for solving the application questions, the answer information at least comprising: an operation formula composed of at least two actual values and operators, and an operation result of the operation formula; deconstructing the operation formula to obtain each actual value in the operation formula; comparing each actual value with a reference value to obtain a comparison result; when the actual numerical value in the comparison result characterization operation expression agrees with the stem information, obtaining a standard answer of the application question; and detecting the operation process and the operation result of the operation formula based on the standard answer to obtain a detection result for representing the correctness of the answer information. The application can improve the accuracy of automatic correction of the application questions.

Description

Solution detection method and device for application questions

Technical Field

The present application relates to computer technology, and more particularly, to a method and apparatus for detecting solutions of application questions.

Background

With the development of modern information technology and the demand of education market, automatic correction is becoming an emerging technology, and is being continuously popularized and popularized. The automatic correction brings great convenience to teachers and students, and compared with the manual correction, the automatic correction has the advantages of simple flow, greatly improved efficiency, stable accuracy and saving manpower, material resources and educational resources. However, the existing automatic correction scheme only can judge the correct errors of all the calculation results in the student responses, cannot identify the response content which is randomly carried out by the student and is irrelevant to the questions, and cannot judge the middle process of the student responses, so that the accuracy of the results of automatic correction of the questions is low.

Disclosure of Invention

The embodiment of the application provides a solution detection method and device for application questions, a computer readable storage medium and a computer program product, which can improve the accuracy of automatic correction of the application questions.

The technical scheme of the embodiment of the application is realized as follows:

The embodiment of the application provides a solution detection method for application questions, which comprises the following steps:

Acquiring the stem information of an application question and the answer information of the application question;

Wherein, the stem information at least includes: at least one reference value for solving the application questions, the answer information at least comprising: an operation formula composed of at least two actual values and operators, and an operation result of the operation formula;

deconstructing the operation formulas to obtain each actual numerical value in the operation formulas;

comparing each actual value with the reference value to obtain a comparison result;

When the comparison result represents that the actual numerical value in the operation formula is matched with the stem information, obtaining a standard answer of the application question;

And detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for representing the correctness of the answer information.

The embodiment of the application provides a solution detection device for application questions, which comprises:

The first acquisition module is used for acquiring the stem information of the application questions and the answer information of the application questions; wherein, the stem information at least includes: at least one reference value for solving the application questions, the answer information at least comprising: an operation formula composed of at least two actual values and operators, and an operation result of the operation formula;

the deconstructing module is used for deconstructing the operation formulas to obtain the actual numerical values in the operation formulas;

the comparison module is used for comparing each actual value with the reference value to obtain a comparison result;

The second obtaining module is used for obtaining a standard answer of the application question when the comparison result represents that the actual numerical value in the operation formula is matched with the question stem information;

and the detection module is used for detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for representing the correctness of the answer information.

In the above scheme, the comparison module is further configured to compare the actual values with the reference values respectively, so as to obtain an intermediate comparison result; and when the intermediate comparison result represents that the actual numerical value exists in the stem information, generating the comparison result representing that the actual numerical value is matched with the stem information.

In the above scheme, the comparison module is further configured to generate a detection result for characterizing the answer information error when the intermediate comparison result characterizes that the actual numerical value does not exist in the stem information; acquiring the number of the operation formulas; outputting a manual verification prompt message when the number of the operation formulas is at least two and the actual numerical value which does not exist in the stem information is the intermediate result of the answer information; the manual verification prompt information is used for prompting the detection result to be manually verified.

In the above solution, the second obtaining module is further configured to obtain, when the answer types of the application questions include at least two answer types, standard answers of the at least two answer types, where the standard answers include an answer process and a standard operation result; the step of detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for representing the correctness of the answer information comprises the following steps: determining a solution type corresponding to the operation formula; extracting target standard answers corresponding to the answer types of the operation formulas from the standard answers of the at least two answer types; and detecting the operation process of the operation formula and the operation result based on the target standard answer to obtain a detection result for representing the correctness of the answer information.

In the above scheme, the second obtaining module is further configured to determine an operation result symbol and an unknown number in the operation formula; deconstructing the operation formulas based on the operation result symbols to obtain at least two operation parts; and counting the number of the operation result symbols, the number of the unknowns and the number of the operation parts, and determining the answer type corresponding to the operation formula based on the counting result.

In the above scheme, the second obtaining module is further configured to match a standard operation result included in the target standard answer with the operation result to obtain a first matching result, and match a solution process included in the target standard answer with an operation process of the operation formula to obtain a second matching result; and generating a detection result for representing the correctness of the answer information by combining the first matching result and the second matching result.

In the above scheme, the detection module is further configured to determine a solution type corresponding to the operation formula when the standard answer includes a standard operation result and the solution types of the application questions include at least two solutions; when the standard operation result is the same as the operation result, according to the answer type corresponding to the operation formula, carrying out accounting on the operation process of the operation formula to obtain an accounting result; and determining a detection result used for representing the correctness of the answer information based on the accounting result.

In the above scheme, the device further comprises a matching module, wherein the matching module is used for acquiring first text reply information for responding to the application questions and second text reply information in the answer information; and matching the first text reply information with the second text reply information, and determining the correctness of the second text reply information based on a matching result.

In the above scheme, the device further comprises a model module, wherein the model module comprises a feature extraction layer, a feature processing layer and an output layer; the model module is used for acquiring the stem information of the application questions and the answer information of the application questions through the feature extraction layer; deconstructing the operation formula through the feature processing layer to obtain each actual value in the operation formula, and comparing each actual value with the reference value to obtain a comparison result; and when the comparison result represents that the actual numerical value in the operation formula is matched with the stem information, acquiring a standard answer of the application question, and detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for representing the correctness of the answer information.

An embodiment of the present application provides an electronic device, including:

A memory for storing executable instructions;

And the processor is used for realizing the solution detection method of the application questions provided by the embodiment of the application when executing the executable instructions stored in the memory.

The embodiment of the application provides a computer readable storage medium which stores executable instructions for realizing the solution detection method of the application questions provided by the embodiment of the application when being executed by a processor.

Embodiments of the present application provide a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the electronic device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the electronic device executes the solution detection method of the application questions provided by the embodiment of the application.

The embodiment of the application has the following beneficial effects:

In the process of answering and detecting the application questions, whether the actual values contained in the answer information of the application questions are matched with the answer information of the application questions or not is determined by comparing the reference values contained in the answer information of the application questions with the actual values contained in the answer information of the application questions, and when the actual values contained in the answer information of the application questions are matched with the answer information of the application questions, the answer information of the application questions is detected based on the acquired standard answers of the application questions, so that a detection result for representing the correctness of the answer information is obtained. Therefore, by judging whether answer information of the application questions is wedged or not, answer content which is randomly carried out by students and is irrelevant to the questions is identified, and accuracy of automatic correction of the application questions is improved.

Drawings

FIG. 1 is a schematic diagram of an alternative architecture of a solution detection system 100 for application questions provided in an embodiment of the present application;

fig. 2 is a schematic structural diagram of an electronic device according to an embodiment of the present application;

FIG. 3 is an alternative schematic diagram of providing a standard answer to an application question according to an embodiment of the present application;

FIG. 4 is an alternative schematic diagram of student responses to an application provided by an embodiment of the present application;

FIG. 5 is an alternative schematic diagram of an automatic correction process for application questions provided by an embodiment of the present application;

FIG. 6 is a flow chart of a solution detection method for application questions provided by an embodiment of the present application;

FIG. 7 is an alternative schematic diagram of a solution detection process of an application question provided in an embodiment of the present application;

FIG. 8 is a schematic flow chart of an alternative method for obtaining a test result for characterizing the correctness of answer information according to an embodiment of the present application;

FIG. 9 is an alternative schematic diagram of a horizontal discriminating process according to an embodiment of the present application;

FIG. 10 is a schematic diagram of an alternative embodiment of the detachment determination process;

FIG. 11 is an alternative schematic diagram of the solution equation discrimination process according to the embodiment of the present application;

FIG. 12 is a schematic flow chart of an alternative method for obtaining a test result for characterizing the correctness of answer information according to an embodiment of the present application;

FIG. 13 is an alternative schematic diagram of a horizontal solution detection method according to an embodiment of the present application;

FIG. 14 is an alternative schematic diagram of a solution-off detection method according to an embodiment of the present application;

FIG. 15 is an alternative schematic diagram of a solution equation solution detection method according to an embodiment of the present application;

FIG. 16 is an alternative schematic diagram of a solution detection method for solutions provided by embodiments of the present application;

FIG. 17 is a schematic flow chart of an alternative method for training a solution detection model of an application problem according to an embodiment of the present application;

FIG. 18 is a schematic diagram of an alternative structure of a solution detection model for application questions provided in an embodiment of the present application;

fig. 19 is a schematic flow chart of an alternative answer detection method for application questions according to an embodiment of the present application.

Detailed Description

The present application will be further described in detail with reference to the accompanying drawings, for the purpose of making the objects, technical solutions and advantages of the present application more apparent, and the described embodiments should not be construed as limiting the present application, and all other embodiments obtained by those skilled in the art without making any inventive effort are within the scope of the present application.

In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is to be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with one another without conflict.

In the following description, the terms "first", "second", "third" and the like are merely used to distinguish similar objects and do not represent a specific ordering of the objects, it being understood that the "first", "second", "third" may be interchanged with a specific order or sequence, as permitted, to enable embodiments of the application described herein to be practiced otherwise than as illustrated or described herein.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the application only and is not intended to be limiting of the application.

Before describing embodiments of the present application in further detail, the terms and terminology involved in the embodiments of the present application will be described, and the terms and terminology involved in the embodiments of the present application will be used in the following explanation.

1) Four arithmetic tools: the tool to derive 6 from the output expression, such as 1+2+3, is currently done based on the inverse Polish expression.

2) Marking a platform: the labeling platform is mainly used for labeling stems, correct answers and corresponding coordinate areas of teaching auxiliary questions, and comprises a stem area and an answer area.

3) And (5) calculating: i.e., equation calculation, the operation of completely writing out the calculation process,

The following describes an exemplary application of the answer detection device for application questions provided by the embodiments of the present application, where the answer detection device for application questions provided by the embodiments of the present application may be implemented as various types of user terminals such as a notebook computer, a tablet computer, a desktop computer, a set-top box, a mobile device (for example, a mobile phone, a portable music player, a personal digital assistant, a dedicated messaging device, a portable game device), a voice interaction device, an intelligent home appliance, a vehicle-mounted terminal, and the like, and may also be implemented as a server.

Referring to fig. 1, fig. 1 is a schematic diagram of an optional architecture of an answer detection system 100 for application questions provided in an embodiment of the present application, in order to implement an application scenario for answer detection of the application questions (for example, the application scenario for answer detection of the application questions may be an application scenario in which APP automatically modifies student homework, for example, student homework is modified by APP after being submitted, and a determination is made as to whether the student homework is correct or not), a terminal (a terminal 400 is illustrated in an exemplary embodiment) is connected to a server 200 through a network 300, and the network 300 may be a wide area network or a local area network, or a combination of both.

The terminal 400 is configured for a user to use the client 401 and display on a display interface 401-1 (the display interface 401-1 is shown as an example). The terminal 400 and the server 200 are connected to each other through a wired or wireless network.

The server 200 is configured to obtain stem information of an application question and answer information of the application question; here, the stem information includes at least: at least one reference value for solving the application questions, the answer information at least comprising: an operation formula composed of at least two actual values and operators, and an operation result of the operation formula; deconstructing the operation formula to obtain each actual value in the operation formula; comparing each actual value with a reference value to obtain a comparison result; when the actual numerical value in the comparison result characterization operation expression agrees with the stem information, obtaining a standard answer of the application question; based on the standard answers, detecting the operation process and the operation result of the operation formula to obtain a detection result for representing the correctness of the answer information; transmitting a detection result for characterizing the correctness of the answer information to the terminal 400;

the terminal 400 is further configured to display, in the graphical interface 401-1, a received detection result for characterizing the correctness of the answer information.

In some embodiments, the server 200 may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, and basic cloud computing services such as big data and artificial intelligence platforms. The terminal 400 may be, but is not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, a set-top box, and a mobile device (e.g., a mobile phone, a portable music player, a personal digital assistant, a dedicated messaging device, a portable game device, a smart speaker, and a smart watch), etc. The terminal device and the server may be directly or indirectly connected through wired or wireless communication, which is not limited in the embodiment of the present application.

Referring to fig. 2, fig. 2 is a schematic structural diagram of an electronic device according to an embodiment of the present application, in an actual application, the electronic device may be the server 200 or the terminal 400 shown in fig. 1, referring to fig. 2, and the electronic device shown in fig. 2 includes: at least one processor 410, a memory 450, at least one network interface 420, and a user interface 430. The various components in terminal 400 are coupled together by a bus system 440. It is understood that the bus system 440 is used to enable connected communication between these components. The bus system 440 includes a power bus, a control bus, and a status signal bus in addition to the data bus. But for clarity of illustration the various buses are labeled in fig. 2 as bus system 440.

The Processor 410 may be an integrated circuit chip having signal processing capabilities such as a general purpose Processor, such as a microprocessor or any conventional Processor, a digital signal Processor (DSP, digital Signal Processor), or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like.

The user interface 430 includes one or more output devices 431, including one or more speakers and/or one or more visual displays, that enable presentation of the media content. The user interface 430 also includes one or more input devices 432, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

Memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid state memory, hard drives, optical drives, and the like. Memory 450 optionally includes one or more storage devices physically remote from processor 410.

Memory 450 includes volatile memory or nonvolatile memory, and may also include both volatile and nonvolatile memory. The non-volatile memory may be read only memory (ROM, read Only Me mory) and the volatile memory may be random access memory (RAM, random Access Memor y). The memory 450 described in embodiments of the present application is intended to comprise any suitable type of memory.

In some embodiments, memory 450 is capable of storing data to support various operations, examples of which include programs, modules and data structures, or subsets or supersets thereof, as exemplified below.

An operating system 451 including system programs, e.g., framework layer, core library layer, driver layer, etc., for handling various basic system services and performing hardware-related tasks, for implementing various basic services and handling hardware-based tasks;

Network communication module 452 for reaching other computing devices via one or more (wired or wireless) network interfaces 420, exemplary network interfaces 420 include: bluetooth, wireless compatibility authentication (WiFi), and universal serial bus (USB, universal Serial Bus), etc.;

A presentation module 453 for enabling presentation of information (e.g., a user interface for operating peripheral devices and displaying content and information) via one or more output devices 431 (e.g., a display screen, speakers, etc.) associated with the user interface 430;

An input processing module 454 for detecting one or more user inputs or interactions from one of the one or more input devices 432 and translating the detected inputs or interactions.

In some embodiments, the solution detecting device for application questions provided in the embodiments of the present application may be implemented in software, and fig. 2 shows the solution detecting device 455 for application questions stored in the memory 450, which may be software in the form of a program, a plug-in, etc., and includes the following software modules: the first acquisition module 4551, the deconstructing module 4552, the comparison module 4553, the second acquisition module 4554 and the detection module 4555 are logical, and thus may be arbitrarily combined or further split according to the functions implemented.

In other embodiments, the solution detecting device for Application questions provided in the embodiments of the present application may be implemented in hardware, and as an example, the solution detecting device for Application questions provided in the embodiments of the present application may be a processor in the form of a hardware decoding processor, which is programmed to perform the solution detecting method for Application questions provided in the embodiments of the present application, for example, the processor in the form of a hardware decoding processor may use one or more Application specific integrated circuits (ASICs, applications SPECIFIC INTEGRATED circuits), DSPs, programmable logic devices (PLDs, programmable Logic Device), complex Programmable logic devices (CPLDs, complex Programmable Logic Device), field Programmable gate arrays (FPGAs, fi eld-Programmable GATE ARRAY) or other electronic components.

In some embodiments, the terminal or the server may implement the solution detection method for the application questions provided by the embodiments of the present application by running a computer program. For example, the computer program may be a native program or a software module in an operating system; the Application program can be a local (Native) Application program (APP), namely a program which can be installed in an operating system to run, such as an instant messaging APP and a web browser APP; the method can also be an applet, namely a program which can be run only by being downloaded into a browser environment; but also an applet that can be embedded in any APP. In general, the computer programs described above may be any form of application, module or plug-in.

In the prior art, the process of automatically correcting the application questions by the correction software mainly aims at the calculation formula in the application questions, carries out four arithmetic processing, then judges whether the current formula is correct, and judges pairs if the current formula is correct; if the error occurs, the error is judged, as shown in fig. 3, fig. 3 is an optional schematic diagram of a standard answer of an application question provided by the embodiment of the present application, and for the application question, the student makes multiple answers, exemplarily, referring to fig. 4, fig. 4 is an optional schematic diagram of a student answer of an application question provided by the embodiment of the present application, based on the student answer, automatic correction software will correct the student answer accordingly, referring to fig. 5, fig. 5 is an optional schematic diagram of an automatic correction process of an application question provided by the embodiment of the present application, based on fig. 5, it may be determined that the existing automatic correction software has the following problems in the automatic correction process: and1, judging the formula, wherein the judgment of the wedge problem is not carried out, namely, whether the number in the formula appears in the stem is not judged. For example, "9*8 =72" in fig. 5, an erroneous expression is apparent; problem 2, when the formula is judged, the judgment of the intermediate process is not performed, that is, only the formula and the result are judged, whether the intermediate process is written correctly or not is not judged, for example, in fig. 5, "8+2+5=10+4=15", and it is obvious that the intermediate process of "10+4" does not conform to the expectation of the intermediate process and should be erroneous; question 3, no correction of the answer, such as "answer" in fig. 5: the area of the square is 64 square centimeters "without giving an altered result.

Based on this, according to the above description of the answer detection system and the electronic device for the application questions provided by the embodiment of the application, the answer detection method for the application questions provided by the embodiment of the application is described below. In practical implementation, the solution detection method of the application questions provided by the embodiment of the application may be implemented by a terminal or a server alone or by the terminal and the server cooperatively, and the solution detection method of the application questions provided by the embodiment of the application is illustrated by the server 200 in fig. 1 alone. Referring to fig. 6, fig. 6 is a flowchart of a solution detection method for an application problem according to an embodiment of the present application, and will be described with reference to the steps shown in fig. 6.

Step 101, a server acquires the stem information of an application question and the answer information of the application question; the stem information at least comprises: at least one reference value for solving the application questions, the answer information at least comprising: an operation formula composed of at least two actual numerical values and operators, and an operation result of the operation formula.

In some embodiments, after the students do the questions, the questions are uploaded to the server through the client, the paper-made questions and the answer scanning or photographing for the paper-made questions can be uploaded, or the pictures or documents containing the electronic-made questions and the on-line answer for the electronic-made questions can be uploaded, and after the server receives the pictures or documents, the server uses an optical character recognition (OCR, optical Character Recognition) technology to recognize the uploaded questions and the answer of the application questions to be modified, namely, the question stem information of the application questions and the answer information of the application questions are obtained.

It should be noted that, the stem information herein includes at least one reference value for solving the application question, where the at least one reference value for solving the application question includes data that appears in the stem and numbers that do not appear in the stem (i.e., implicit numbers in the stem and numbers that are used for solving the application question but are irrelevant to the stem).

As an example, for the title "maximum three digits plus maximum two digits, the resulting sum is how many times 16", where the maximum three digits 999 and the maximum two digits 99 can be considered as implicit digits in the stem; still alternatively, "the annual revenue for a year of Ming is xx yuan, the daily revenue and the monthly revenue are more or less red", 365 (i.e., 365 days a year) and 12 (i.e., 12 months a year) can be regarded as implicit numbers in the stem.

For the above example, since the maximum three-digit number and the maximum two-digit number have special meanings, that is, the maximum three-digit number may be regarded as the minimum four-digit number minus one, the maximum three-digit number may be regarded as the minimum three-digit number minus one, and the minimum four-digit number 1000 and the minimum three-digit number 100 have the ease of calculation and the high frequency of occurrence, 1000, 100,1 may be regarded as an implicit number in the stem, that is, the maximum three-digit number corresponds to "1000-1", and the maximum two-digit number may be regarded as "100-1".

And 102, deconstructing the operation formula to obtain each actual numerical value in the operation formula.

In actual implementation, after an operation expression composed of at least two actual values and operators is acquired, each actual value constituting the operation expression is obtained by performing deconstructing processing on the operation expression.

Taking the above example, for the question "how many times the sum is 16 is the maximum three digits plus the maximum two digits", for this question, the student's answer may be "(999+99)/(16"), or may be "(999+98)/(16"), or may be "{ (1000-1) + (100-1) } ++16", here, see fig. 7, fig. 7 is an alternative schematic diagram of the answer detection process of the application question provided by the embodiment of the present application, as shown in fig. 7, the calculation formula of the student's answer may be deconstructed for the student's answer 1, to obtain each actual numerical value 999, 99 and 16 in the calculation formula; for student answer 2, the calculation formula of the student answer can be deconstructed to obtain actual values 1000, 1, 100, 1 and 16 in the calculation formula; for student answer 3, the calculation of the student answer may be deconstructed to obtain actual values 999, 98, and 16 in the calculation.

And 103, comparing each actual value with a reference value to obtain a comparison result.

In actual implementation, two types of comparison results exist when each actual value is compared with the reference value, namely, the actual value of the comparison result representation exists in the stem information, and the actual value of the comparison result representation does not exist in the stem information, and then two different types of comparison results are described.

In some embodiments, the intermediate comparison results are obtained by comparing each actual value with the reference value, and when each intermediate comparison result represents that the actual value exists in the stem information, a final result is generated, that is, a comparison result representing that the actual value agrees with the stem information is generated.

Continuing the above example, with continued reference to fig. 7, for student answer 1, comparing the deconstructed values 999, 99, and 16 with the reference values 999, 99, and 16 to obtain 3 intermediate comparison results, wherein the 3 intermediate comparison results all characterize the actual values to exist in the stem information, thus generating a comparison result of the actual value-fit stem information; for student response 2, comparing the deconstructed values 1000, 1, 100, 1 and 16 with the reference values 1000, 1, 100, 1 and 16 to obtain 5 intermediate comparison results, wherein the 5 intermediate comparison results represent that the actual values exist in the stem information, so that the comparison result of the actual value fit stem information is generated.

In some embodiments, the intermediate comparison results are obtained by comparing each actual value with the reference value, and when at least one intermediate comparison result indicates that the actual value does not exist in the stem information, a final result is generated, that is, a detection result for indicating that the answer information is wrong is generated.

Continuing with the above example, referring to fig. 7, for student answer 3, the deconstructed values 999, 98 and 16 are compared with the reference values 999, 98, 1 and 16 to obtain 4 comparison results, where the third comparison result characterizes that the actual value does not exist in the stem information, and a detection result for characterizing that the answer information is wrong is generated. Here, 98 is a number for solving the application problem but is irrelevant to the problem stem.

It should be noted that, "equation (999+98+1)/16" in fig. 7 is a badcase input during model training, and since the present application is directed to answer detection of an application question, the information in the question has practical meaning, for example, the maximum two-digit number is 99, based on the calculation convenience and occurrence of high frequency, only a form of a difference between 1000 and 1 can be written, but a form of a sum of 98 and 1 cannot be written, so that for the equation "(999+98+1)/16", even if the calculation result is correct, the evaluation is wrong, that is, "equation (999+98+1)/16" is a badcase input during model training, 98 is a reference value which cannot exist in the stem information, and therefore, it is a student answer "(999+98+1)/16", and when the comparison of the practical value 98 and the reference value 98 is successful, the indicated intermediate comparison result indicates that the practical value is not exist in the stem information. While for student answer 3, even if "solution (999+98+1)/16" is badcase input during model training, since the correct result of this question is "(999+99)/16", the calculation result of "(999+98+1)/16" is correct, and when at least one intermediate comparison result representing actual values of solution values 999, 98 and 16 of student answer 3 and reference values 999, 98, 1 and 16 is not present in the stem information, a detection result for representing an answer information error is generated.

In practical implementation, after a detection result for representing an answer information error is generated, the number of the operation formulas is first obtained, and different subsequent processing modes exist based on the obtained number of the operation formulas, and then two different subsequent processing modes are described.

In some embodiments, when the number of the obtained operation formulas is at least two and the actual numerical value which does not exist in the stem information is the intermediate result of the answer information, outputting a manual verification prompt message aiming at the detection result representing the error of the answer information, wherein the manual verification prompt message is used for prompting the manual verification of the detection result.

In the above example, when the student answers "999+99=1098, 1098++16", the deconstructed values obtained by the deconstructed expressions at this time are 999, 99, 1098, and 16, 1098 are actual values which are not present in the stem information, but since there are two expressions and 1098 is an intermediate result of the answer information, the manual verification prompt information is output for the detection result representing the answer information error.

In some embodiments, when the obtained operation formula is one, or when the number of obtained operation formulas is at least two and the actual numerical value not existing in the stem information is not an intermediate result of the answer information, a detection result characterizing the answer information error is directly output.

Taking the above example, the scenario of answer 3 is the student at this time, and therefore, the detection result characterizing the answer information error is directly output; or when the student answers "999+99=1097, 1098/16", the deconstructed values obtained by the deconstructed operation formulas at this time are 999, 99, 1097, 1098 and 16, 1097 and 1098 are actual values which are not present in the stem information, and two operation formulas exist at the same time, but since 1097 and 1098 are not intermediate results of the answer information, the detection result representing the answer information error is directly output.

And 104, when comparing the actual numerical value contract question stem information in the result characterization operation formula, acquiring a standard answer of the application question.

In practical implementation, when comparing actual numerical value contract question stem information in the result characterization operation formula, obtaining standard answers of the application questions, specifically, when the answer types of the application questions comprise at least two, obtaining standard answers of at least two answer types, wherein the standard answers comprise an answer process and a standard operation result. It should be noted that the answer types of the application questions at least include horizontal, discrete, solution equation and answer.

As an example, when the question is "which number is the sum of 3 and 6", the answer type corresponding to the question is horizontal, the solution equation, and the answer, and the obtained standard answer is "x+3=6, x=3: the sum of 3 and 3 is 6"," 6-3=3, answer: the sum of 3 and 3 is 6", and" answer: the sum of 3 and 3 is 6".

It should be noted that, when there are a plurality of solutions of different application questions, that is, the solution types of the application questions may be only one, and for example, when the question is "which number is the sum of 3 and 6, please answer with the solution equation", the obtained standard answer is only "x+3=6, x=3, answer: the sum of 3 and 3 is 6".

Step 105, based on the standard answer, detecting the operation process and the operation result of the operation formula to obtain a detection result for representing the correctness of the answer information.

In practical implementation, after obtaining at least two standard answers of answer type, based on the standard answers, the operation process and the operation result of the operation formula are detected, and the process of obtaining the detection result for characterizing the correctness of the answer information is referred to fig. 8, and fig. 8 is a schematic flowchart of an alternative process for obtaining the detection result for characterizing the correctness of the answer information according to the embodiment of the present application, based on fig. 6, step 105 may be further implemented as follows:

In step 1051, the solution type corresponding to the operation formula is determined.

In actual implementation, since different answer types exist in the application questions, the answer types corresponding to the operation formulas in the acquired answer information are determined firstly, and specifically, operation result symbols and unknowns in the operation formulas are determined; deconstructing the operation formulas based on the operation result symbols to obtain at least two operation parts; and counting the number of operation result symbols, the number of unknowns and the number of operation parts, and determining the answer type corresponding to the operation formula based on the counting result.

It should be noted that, for different answer types, there are different determination manners, and a process of determining an answer type corresponding to an operation expression is described next based on the different answer types.

In some embodiments, since the transverse expression is an operation expression with only one equal sign in the operation expressions, that is, the transverse expression is an operation expression including one equal sign, whether the operation expression is the transverse expression is determined by judging whether the operation expression has only one equal sign and how many numerical operation expressions on both sides of the equal sign, specifically, the operation expressions on both sides are split according to the equal sign, that is, deconstructing is performed on the operation expressions to obtain at least two operation parts, and when the operation parts obtained by deconstructing are two and the operation parts are composed of numbers and logic characters, the operation expression is determined to be the transverse expression, otherwise, the operation expression does not belong to the transverse expression.

As an example, referring to fig. 9, fig. 9 is an optional schematic diagram of a horizontal type discriminating process provided in the embodiment of the present application, based on fig. 9, whether the operation formula 1 is a horizontal type is determined, first, the left and right sides are split according to the equal sign to obtain two operation parts of 1+2 and 3, and since only two operation parts are obtained, and both operation parts are composed of numbers and logic characters, the operation formula 1 is determined to be a horizontal type; then judging whether the operation formula 2 is a horizontal type or not, firstly splitting the left side and the right side according to an equal sign to obtain four operation parts of 1+2+3, 6, 3+3 and 6, and because the obtained operation parts are more than two, the operation formula 2 is not a horizontal type; finally, judging whether the operation formula 3 is a horizontal type or not, firstly, dividing the left side and the right side according to an equal sign to obtain four operation parts of x+3, 6, x and 6, and because the obtained operation parts are more than two, and the obtained operation parts have unknowns, the operation parts do not consist of numbers and logic characters, and therefore the operation formula 3 is not a horizontal type.

In some embodiments, since the disjunction is composed of one or more rows of operation formulas, that is, a plurality of operation formulas are displayed in series, and a plurality of equal signs are included in the operation formulas, whether the operation formulas are disjunction is determined by determining whether a plurality of equal signs and the number of numerical operation formulas on both sides of the equal signs, specifically, first, the operation formulas on both sides of the equal signs are split, that is, deconstructing is performed on the operation formulas to obtain at least three operation parts, and when the operation parts obtained by deconstructing are greater than two and each operation part is composed of numbers and logic characters, the operation formulas are determined to be disjunction, otherwise, the operation formulas are not disjunction.

As an example, referring to fig. 10, fig. 10 is an optional schematic diagram of a discriminant determination process provided in an embodiment of the present application, based on fig. 10, whether the operation formula 1 is a discriminant is determined, first, the left and right sides are split according to the equal sign to obtain four operation parts of 1+2+3, 6, 3+3 and 6, and since the obtained operation parts are greater than two, a plurality of equal signs are described, and the operation part is composed of numbers and logic characters, the operation formula 1 is a discriminant; then judging whether the operation formula 2 is in a disjunction form or not, firstly splitting the left side and the right side according to an equal sign to obtain two operation parts of 1+2 and 3, and because only two operation parts are obtained, only one equal sign is described, the operation formula 2 is not in the disjunction form; finally, whether the operation formula 3 is a disjunction is judged, first, the left and right sides are split according to the equal sign, and four operation parts of x+3, 6, x and 6 are obtained, and although the obtained operation parts are four, a plurality of equal signs are described, the operation part has unknown numbers and is not composed of numbers and logic characters, so the operation formula 3 is not a disjunction.

In some embodiments, since the solution equation is not only a pure numerical expression but also includes letters of an unknown number, whether the solution equation is determined by determining whether the solution equation includes the unknown number, specifically, first splitting the solution equation according to the equal number, that is, performing a deconstructing process on the solution equation to obtain at least four operation portions, and when at least two operation portions in the at least four operation portions include the unknown number, determining that the solution equation is not a solution equation, otherwise.

As an example, referring to fig. 11, fig. 11 is an optional schematic diagram of a solution equation determining process provided in the embodiment of the present application, based on fig. 11, determining whether the solution equation is given by the operation formula 1, first splitting the left and right sides according to the equal sign to obtain four operation parts x+3, 6, x and 6, where the operation parts x+3 and x have unknowns, so that the operation formula 3 belongs to the solution equation; then judging whether the operation formula 2 is a solution formula or not, firstly splitting the left side and the right side according to an equal sign to obtain two operation parts of 1+2 and 3, wherein the two operation parts are composed of numbers and logic characters, and no unknown number exists, so that the operation part does not belong to the solution formula.

It should be noted that, after the judgment of the horizontal and the discrete equations, if the answer type of the operation equation is not the above three types, the answer type of the application question is an answer.

Step 1052, extracting target standard answers corresponding to the answer types of the operation formulas from the standard answers of at least two answer types.

In actual implementation, after the answer type of the operation formula is judged, extracting a target standard answer corresponding to the answer type of the operation formula from the obtained standard answers of at least two answer types.

For the above example, for the term "which number and 3 sum to 6", if the formula is horizontal, answer: the sum of 3 and 3 is 6"," 6-3=3, answer: the sum of 3 and 3 is 6", and" answer: the sum of 3 and 3 is 6 "extract" 6-3=3, answer: and the sum of 3 and 3 is 6' which is the target standard answer.

It should be noted that, based on the requirement of the application question, when only one answer type exists in the obtained standard answer, the obtained standard answer is the target standard answer.

Step 1053, based on the target standard answer, detecting the operation process and the operation result of the operation formula to obtain a detection result for representing the correctness of the answer information.

In actual implementation, after the target standard answer is obtained, detecting an operation process and an operation result of an operation formula in the obtained answer information based on the target standard answer to obtain a detection result used for representing the correctness of the answer information, specifically, matching a standard operation result included in the target standard answer with the operation result to obtain a first matching result, and matching a solution process included in the target standard answer with the operation process of the operation formula to obtain a second matching result; and combining the first matching result and the second matching result to generate a detection result for representing the correctness of the answer information.

In practical implementation, the successful matching of the first matching result is that the standard operation result included in the target standard answer is completely consistent with the operation result, and as the student may have the condition of omitting steps in the process of answering the application questions, the successful matching of the second matching result does not need the complete consistency of the answering process included in the target standard answer and the operation process of the operation formula, and the second matching result can be considered to be successful when the similarity between the answering process included in the target standard answer and the operation process of the operation formula exceeds the preset similarity threshold by setting the similarity threshold.

In actual implementation, only when the first matching result is successfully matched and the second matching result is successfully matched, a detection result which represents that the answer information is correct is generated, otherwise, a detection result which represents that the answer information is incorrect is generated.

In other embodiments, when the obtained standard answer does not include the answering process and only includes the standard operation result, the operation result of the operation formula is detected based on the standard answer, and the process of obtaining the detection result for characterizing the correctness of the answer information is referred to as fig. 12, and fig. 12 is a schematic flowchart of an alternative process for obtaining the detection result for characterizing the correctness of the answer information according to the embodiment of the present application, based on fig. 6, step 105 may be further implemented as follows:

In step 201, when the standard answer includes a standard operation result and the answer types of the application questions include at least two, the answer types corresponding to the operation formulas are determined.

In actual implementation, the answer type is determined for the operation formula, and here, the process of determining the answer type for the operation formula is specifically referred to step 1051.

In practical implementation, after the answer type corresponding to the operation formula is determined, the result of the student answer, namely the operation result, is also determined based on the operation part obtained by the solution in the process of determining the answer type. Next, a process of determining the result of the student's answer will be described.

In some embodiments, when the answer type corresponding to the operation formula is horizontal, two operation parts obtained by the solution are obtained, and when the operation parts are only numbers and have no logical characters (such as +, -, /), the operation parts are answers, i.e. operation results.

As an example, referring to fig. 13, fig. 13 is an optional schematic diagram of a horizontal answer detection method provided by an embodiment of the present application, when the answer type corresponding to the operation formula is horizontal, 1+2 and 3 obtained by the solution are obtained, and since 3 is only a number, and no logical character exists, 3 is an answer, that is, an operation result.

In some embodiments, when the answer type corresponding to the operation formula is a disjunction formula, at least three operation parts obtained by the solution are obtained, and when the operation parts are only numbers and have no logical characters (such as +, -, /), the operation parts are answers, i.e. operation results.

As an example, referring to fig. 14, fig. 14 is an optional schematic diagram of a solution-separating detection method provided by an embodiment of the present application, when a solution type corresponding to an operation formula is a solution-separating type, 8+2+5, 10+5, and 15 obtained by solution are obtained, and since 15 is only a number, no logical character exists in the operation part, and 15 is an answer, that is, an operation result.

In some embodiments, when the solution type corresponding to the operation formula is a solution formula, at least four operation parts obtained by the solution are obtained, and when two operation parts corresponding to the left and right sides of the equal sign do not have any logic characters (such as +, -, /), the two operation parts are the answer formula of the unknown number, that is, the operation part of the two operation parts that is a number is the operation result.

As an example, referring to fig. 15, fig. 15 is an optional schematic diagram of a solution equation solution detection method provided by the embodiment of the present application, when a solution type corresponding to an operation formula is a solution equation, x+2+3, 6, x+5, 6, x and 1 obtained by solution are obtained, and since x and 1 are two operation parts corresponding to the left and right sides of an equal sign and have no logical character, x=1 is an answer formula of an unknown number, where 1 is a digital operation part, and therefore 1 is an answer, that is, an operation result.

In actual implementation, after the operation result is obtained, comparing the standard operation result with the operation result, and continuing to step 202 when the standard operation result is the same as the operation result; when the standard operation result is different from the operation result, representing the current student answer error, generating a detection result for representing the answer information error and outputting the detection result.

And 202, when the standard operation result is the same as the operation result, according to the answer type corresponding to the operation formula, carrying out accounting on the operation process of the operation formula to obtain an accounting result.

In practical implementation, when the standard operation result is the same as the operation result, in order to prevent the situation of plagiarism results possibly occurring to students, the operation process of the operation formula is calculated according to the answer type corresponding to the operation formula, specifically, the answer is brought into the operation formula, and then the operation parts corresponding to the two sides of the equal sign are subtracted, so as to obtain at least one calculation result. Next, a process of calculating the operation procedure of the operation expression according to the answer type corresponding to the operation expression to obtain the calculation result will be described.

In some embodiments, when the operation formula is a horizontal formula, the operation result is brought into the operation formula, an operation formula is formed by subtracting operation parts on the left and right sides of the equal sign, and an operation result is determined by calculating the numerical value of the operation formula.

As an example, with continued reference to fig. 13, when the operation formula is a horizontal formula, 3 is brought into the operation formula, the operation parts 1+2 and 3 on both sides of the equal sign are subtracted to constitute the accounting operation formula 1+2-3, and the accounting result is determined to be 0 by inputting the accounting operation formula to the four-rule arithmetic calculator.

As another example, referring still to 13, when the operation formula is a horizontal formula, 4 is brought into the operation formula, the operation parts 1+2 and 4 on both sides of the equal sign are subtracted to constitute the accounting operation formula 1+2-4, and the accounting result is determined to be-1 by inputting the accounting operation formula to the four-rule operation calculator.

In some embodiments, when the operation formula is a disjunctive formula, the operation results are brought into the operation formula, at least two calculation formulas are formed by subtracting operation parts on the left and right sides of the equal sign, and at least two calculation results are determined by calculating the numerical values of the calculation formulas.

As an example, with continued reference to fig. 14, when the expression is the disjunctive expression, 15 is brought into the expression, two calculation expressions 8+2+5- (10+5) and 10+5-15 are formed by subtracting the operation parts 8+2+5, 10+5 and 10+5, 15 on both sides of the equal sign, and then the two calculation expressions are input to the four-rule calculation calculator, and it is determined that both calculation results are 0.

As another example, referring still to fig. 14, when the expression is the disjunctive expression, 15 is brought into the expression, two accounting expressions 8+2+5- (11+5) and 11+5-15 are formed by subtracting the operation parts 8+2+5, 11+5 and 11+5, 15 on both sides of the equal sign, and then the two accounting expressions are input to the four arithmetic calculator, and it is determined that the two accounting results are-1 and 1, respectively.

In some embodiments, when the operation formula is a solution formula, the operation result is brought into the operation formula to replace the unknown number with an answer, at least two calculation formulas are formed by subtracting operation parts on the left and right sides of the equal sign, and at least two calculation results are determined by calculating the numerical value of the calculation formulas.

As an example, with continued reference to fig. 15, when the expression is a solution equation, 1 is brought into the expression, two calculation expressions 1+2+3-6 and 1+5-6 are formed by subtracting the operation parts 1+2+3, 6 and 1+5, 6 on both sides of the equal sign, and then the two calculation expressions are input to the four calculation calculator, and it is determined that both the calculation results are 0.

As another example, referring still to 15, when the operation formula is a solution formula, 1 is brought into the operation formula, two calculation formulas 1+2+3-6 and 1+3-6 are formed by subtracting the operation parts 1+2+3, 6 and 1+3, 6 on both sides of the equal sign, and then the two calculation formulas are input to the four calculation calculator, and it is determined that the two calculation results are 0 and-2, respectively.

Step 203, determining a detection result for characterizing the correctness of the answer information based on the accounting result.

In actual implementation, after at least one accounting result is obtained, whether the accounting result is 0 is judged, only when each accounting result is 0, a detection result used for representing that answer information is correct is generated, otherwise, a detection result used for representing that answer information is incorrect is generated.

In some embodiments, when the operation formula is horizontal, and referring to fig. 13, when the obtained accounting result is 0, a detection result for representing that the answer information is correct, that is, correction is correct is generated; and when the obtained accounting result is-1, generating a detection result for representing the answer information error, namely, correcting the error.

In some embodiments, when the operation formula is the take-off type, and referring to fig. 14, when the obtained two accounting results are both 0, a detection result for representing that the answer information is correct, that is, correction is correct is generated; when the two obtained accounting results are-1 and-1, a detection result used for representing the answer information error, namely, the correction error is generated.

In some embodiments, when the operation formula is the solution equation, and referring to fig. 15, when the two obtained accounting results are both 0, a detection result for representing that the answer information is correct, that is, correction is correct is generated; when the two obtained accounting results are 0 and-2, a detection result for representing the answer information error, namely the correction error, is generated.

When the obtained standard answer comprises the standard operation result, the operation process of the operation formula is calculated according to the answer type corresponding to the operation formula, the calculation result is obtained, and the correctness of the answer information is further determined based on the calculation result. Here, through the accounting process, the situation that the student plagiarizes the result is avoided, and the correction accuracy of the student answer is improved.

It should be noted that, in some embodiments, for different answer types of the application questions, model training of the answer detection model for the application questions in various answer types may be performed through a neural network, so that classification for the different answer types of the application questions is completed, and recall for badcase may be performed, so that accuracy of classification of the answer types is improved.

In actual implementation, aiming at the condition that the answer type is answer, first acquiring first text reply information (namely standard answer) for responding to the application questions and second text reply information (namely student answer) in answer information; and matching the first text reply information with the second text reply information, and determining the correctness of the second text reply information based on a matching result.

It should be noted that, for the way of matching the first text reply message with the second text reply message, it may be to compare numbers of the student answer content with numbers of the standard answer content first, then to extract comparison words, non-words, nouns, etc. of the student answer content and the standard answer content according to different comparison results, and then to determine distances between numbers and name phrases in the student answer content and the standard answer content, so as to determine whether the student answer content is correct based on the extracted comparison words, non-words, and nouns.

Next, a process of judging whether the contents of the student responses are correct or not with respect to the different comparison results will be described.

In some embodiments, when the student answer content is the same as the number in the standard answer content, extracting a comparison word, a non-word, a noun and the like of the student answer content and the standard answer content, determining each name phrase and the number corresponding to the name phrase by comparing the distance between the number in the student answer content and the standard answer content, so as to compare each name phrase and the number corresponding to the name phrase in the student answer content with the number corresponding to the name phrase based on the extracted comparison word, non-word and noun, and judging whether the answer is correct or not according to the fact that the comparison word, the non-word and the noun are consistent with each name phrase and the number corresponding to the name phrase in the standard answer content; otherwise, the error is generated.

As an example, referring to fig. 16, fig. 16 is an optional schematic diagram of a solution detection method for solutions provided by an embodiment of the present application, where a first text reply message responding to an application question, that is, standard answer content, is "answer: there were 6 chickens for the Xiaoming and 5 chickens for the Xiaohong. "and the second text reply message in the answer information has three kinds, namely, the student answers 1: there were 6 chickens for the Xiaoming and 5 chickens for the Xiaohong. ", student answer 2": xiaoming is 1 chicken more than Xiaohong and student answers 3: xiaoming is more than Xiaohong chicken. Firstly judging whether the numbers of the student answer content and the standard answer content are the same or not, comparing to determine that the number of the student answer 1 and the standard answer content are the same, extracting comparison words, non-words, nouns and the like of the student answer content and the standard answer content, namely determining that the nouns in the student answer 1 are 'small bright', 'small red' and 'chicken', and the non-words are 'have', and then determining that the small bright corresponds to 6 chickens in the student answer 1, the small red corresponds to 5 chickens through comparing the distances between the name phrases 'small bright', 'small red' and the numbers '6' and '5', and determining the standard answer content is the same as the above process. Thus, the standard answer is completely consistent with the student answer, so student answer 1 is correct.

In some embodiments, when the number in the student answer content is different from the number in the standard answer content, extracting a comparison word, a non-word, a noun and the like of the student answer content and the standard answer content, and determining each name phrase and a number corresponding to each name phrase by comparing the distances between the number in the student answer content and the name phrase in the standard answer content, so that based on the extracted comparison word, the non-word and the noun, whether the meaning of the student answer content is consistent with the meaning of the standard answer content is compared, and if so, judging that the answer is correct; otherwise, the error is generated.

Continuing with the above example, referring to fig. 16, it is first determined whether the numbers of the student answer content and the standard answer content are the same, by comparison, it can be determined that the numbers of the student answer 2,3 are different from the numbers of the standard answer, for the student answer 2, first the comparison word, the non-word, the noun, etc. of the student answer content and the standard answer content are extracted, that is, the noun in the student answer 2 is determined to be "small bright", "small red" and "chicken", the comparison word is more ", then by comparing the distances between the name phrases" small bright "," small red "and the number" 1", the meaning of the student answer 2 is determined to be 1 chicken more than small red, and the meaning of the standard answer is that small bright corresponds to 6 chickens, small red corresponds to 5 chickens, here, by the numerical value given by the standard answer, the difference between 6 and 5 is calculated to be 1, so that the meaning of the student answer content is consistent with the meaning of the standard answer content, and therefore the student answer 2 is correct.

In some embodiments, when no number exists in the student answer content or the standard answer content, firstly extracting a comparison word, a non-word, a noun and the like of the student answer content and the standard answer content, determining each name phrase and a number corresponding to the name phrase through the distance between the number and the name phrase in answer information comprising the number, and then comparing whether the meaning of the student answer content is consistent with the meaning of the standard answer content or not based on the extracted comparison word, the non-word, the noun and the like, and if so, judging that the answer is correct; otherwise, the error is generated.

In the above example, first, the comparison words, non-words, nouns, and the like of the student answer content and the standard answer content are extracted, that is, the nouns in the student answer 3 are determined to be "small", "small red", and "chicken", the comparison words are determined to be "small", and since the number is not included in the student answer content, it is determined that the meaning of the student answer 3 is more than the small red chicken based on the comparison words, and the standard answer content includes the number, and therefore, by comparing the distance between the number and the name phrase in the standard answer content, the meaning of the standard answer content is determined to be small to correspond to 6 chickens, small red corresponds to 5 chickens, and here, by comparing the magnitudes of 6 and 5, it can be determined that 6 is greater than 5, and therefore, the small bright chickens are more than the small red chickens, and therefore, whether the meaning of the student answer content is consistent with the meaning of the standard answer content, and therefore the student answer 3 is correct.

It should be noted that, for the correction of the answer of the application question, the model training of the answer detection model of the application question of the stem and the answer content can be performed, so as to judge whether the student text answer is correct, thus improving the accuracy of judging the student text answer.

In the process of solving and detecting the application questions, the embodiment of the application determines whether the actual values contained in the answer information of the application questions are matched with the question stem information of the application questions or not by comparing the reference values contained in the question stem information of the application questions with the actual values contained in the answer information of the application questions, and detects the answer information of the application questions based on the acquired standard answers of the application questions when the actual values contained in the answer information of the application questions are matched with the question stem information of the application questions, thereby obtaining a detection result for representing the correctness of the answer information. Therefore, by judging whether answer information of the application questions is wedged or not, answer content which is randomly carried out by students and is irrelevant to the questions is identified, and accuracy of automatic correction of the application questions is improved.

In some embodiments, referring to fig. 17, fig. 17 is a schematic flow chart of an alternative method for training a solution detection model of an application problem provided by the embodiment of the present application, based on fig. 6, before step 101, training is further required to be performed on a solution detection model of an application problem, where it is to be noted that the solution detection model of an application problem includes a feature extraction layer, a feature processing layer, and an output layer, and here, referring to fig. 18, fig. 18 is a schematic flow chart of an alternative method for training a solution detection model of an application problem provided by the embodiment of the present application, and next, the steps shown in fig. 17 and fig. 18 are described in connection.

In step 301, the server obtains a training application question sample and a corresponding correct correction result.

In practical implementation, the training application question sample and the corresponding correct correction result obtained by the server may be uploaded by the user through the client, for example, the user may upload a paper edition question and a response scan or photograph performed on the paper edition question to the server, or upload a picture or document including an electronic edition question and an online response performed on the electronic edition question to the server.

Step 302, obtaining the stem information of the training application question sample and the answer information of the training application question sample through a feature extraction layer of the answer detection model of the application question.

In practical implementation, the acquired training application question sample comprises question stem information and answer information, wherein the question stem information at least comprises at least one reference value for solving the application question, and the at least one reference value for solving the application question comprises data appearing in the question stem and numbers not appearing in the question stem (namely, implicit numbers in the question stem and numbers which are used for solving the application question and are irrelevant to the question stem); the answer information at least comprises an operation formula formed by at least two actual numerical values and operators and an operation result of the operation formula.

It should be noted that, for the process of obtaining the stem information of the training application question sample, the OCR technology may be used to identify numbers related to answer calculation and hidden numbers in the application question stem, where, for the manner of identifying hidden numbers in the question stem, a high-frequency hidden number table may be constructed by counting numbers with higher occurrence frequency or numbers with specific practical significance (such as 365 days a year, 24 hours a day, etc.) in the training application question sample. Then constructing a digital list related to answer calculation based on the numbers in the answer (namely the numbers related to answer calculation and the implicit numbers in the answer) in the answer stem; based on the numbers which do not appear in the answer in the stem and the numbers which are used for solving the application problem and are irrelevant to the stem (such as the high-frequency hidden numbers selected randomly), a number list irrelevant to the answer calculation is constructed.

In practical implementation, the process of acquiring the answer information of the training application question sample may be the operation result of recognizing the operation formula and the operation formula in the answer information by utilizing the OCR technology.

In actual implementation, after the number list related to the answer calculation and the number list unrelated to the answer calculation are acquired, the construction of the operation formulas of the corresponding application questions is constructed based on the number list related to the answer calculation, and meanwhile, the construction of the operation formulas of the corresponding application questions is constructed based on the number list unrelated to the answer calculation.

In practical implementation, the reference value in the stem information is the deconstructed number in the calculated expression of the corresponding application question, and in the above example, based on the number list related to the answer calculation, the calculated expression of the corresponding application question constructed can be "(999+99) +.16" and "{ (1000-1) + (100-1) } +.16", and the corresponding deconstructed numbers are 999, 99, 16 and 1000, 1, 100, 1, 16; and based on the number list irrelevant to answer calculation, the calculated expression of the constructed corresponding application question can be "(999+98+1)/(16"), and the corresponding deconstructed numbers are 999, 98, 1 and 16.

Step 303, deconstructing the operation formulas in the answer information through the feature processing layer of the answer detection model of the application questions to obtain each actual numerical value in the operation formulas.

And step 304, comparing each actual value with a reference value in the stem information through a feature processing layer of a solution detection model of the application question to obtain a comparison result.

In actual implementation, based on the similarity, a construction operation formula with the highest similarity with the operation formula in the answer information is found from the operation formulas of the corresponding application questions obtained through construction, so that the deconstructed number of the construction operation formula with the highest similarity, namely the reference value in the stem information, is obtained, and then the actual values are compared with the reference value in the stem information, so that a comparison result is obtained.

Step 305, based on the comparison result, determining whether the actual numerical value in the operation formula matches the stem information through the output layer of the solution detection model of the application question.

In step 306a, when the actual numerical value in the operation formula agrees with the stem information, the standard answer of the application question is obtained through the output layer of the answer detection model of the application question, and the operation process and the operation result of the operation formula are detected based on the standard answer, so as to obtain the detection result for representing the correctness of the answer information.

And 306b, when the actual numerical value in the operation formula does not agree with the stem information, obtaining a detection result for representing the answer information error through an output layer of a solution detection model of the application question.

Step 307, updating model parameters of the answer detection model of the application question based on the difference between the detection result for the answer information and the correct correction result.

Therefore, through training of the answer detection model of the application questions, accuracy of detection results of answer information for the application questions, which are obtained based on the answer detection model of the application questions, is improved.

Next, continuing to describe the solution detection method for the application questions provided by the embodiment of the present application, fig. 19 is a schematic flow chart of an alternative solution detection method for the application questions provided by the embodiment of the present application, and referring to fig. 19, the solution detection method for the application questions provided by the embodiment of the present application is cooperatively implemented by a client and a server.

In step 401, the client responds to the uploading operation for the training application question samples, and obtains a plurality of training application question samples and corresponding correct correction results.

In practical implementation, the client may be a solution detection client of the application questions set in the terminal, the training application question samples may be triggered by the user based on the man-machine interaction interface of the client, so that the client presents the training application question sample selection interface on the man-machine interaction interface, and the user uploads the training application question samples locally from the terminal based on the training application question sample selection interface, thereby enabling the client to obtain the uploaded training application question samples.

Step 402, the client sends the training application question sample and the corresponding correct correction result to the server.

Step 403, the server inputs the received training application question sample to the answer detection model of the application question.

Step 404, outputting the detection result of the answer information in the test application question samples.

Step 405, updating model parameters of the answer detection model of the application question based on the difference between the detection result for the answer information and the correct correction result.

In practical implementation, the server completes training of the solution detection model of the application problem by iterating the training process until the loss function reaches convergence.

In step 406, the server generates a prompt message for completion of the solution detection model training of the application questions.

In step 407, the server sends a prompt message to the client.

In step 408, the client responds to the uploading operation for the application questions and the corresponding answer information, and obtains the application questions and the corresponding answer information.

In practical implementation, the application questions and the corresponding answer information can be obtained through shooting by a camera in communication connection with the terminal, and after the camera shoots the application questions and the corresponding answer information, the application questions and the corresponding answer information are transmitted to the terminal and automatically uploaded to the client by the terminal. Or after the user answers the electronic questions on the terminal, the picture containing the electronic questions and the on-line answer to the electronic questions is obtained through screenshot, or the document containing the electronic questions and the on-line answer to the electronic questions is stored, so that the picture or the document is uploaded to the client, and the application questions and the corresponding answer information are uploaded to the client.

In step 409, the client sends the application questions and corresponding answer information to the server in response to the answer detection instruction for the application questions.

In practical implementation, the answer detection instruction of the application question may be automatically generated by the client under a certain triggering condition, for example, after the client acquires the application question and the corresponding answer information, the answer detection instruction for the application question is automatically generated, or may be sent to the client by other devices connected with the terminal in a communication manner, or may be generated after the user triggers the corresponding answer detection function item based on a man-machine interaction interface of the client.

Step 410, the server inputs the received application questions and corresponding answer information to the answer detection model of the application questions, so that the answer detection model of the application questions carries out answer detection on the application questions and the corresponding answer information, and a detection result of the answer information for the application questions is obtained.

In step 411, the server sends the detection result to the client.

In step 412, the client outputs the detection result.

In practical implementation, the client may present a detection result of answer information for the application question in a man-machine interaction interface of the client, or store the detection result to the local terminal, or send the detection result to other devices connected with the terminal in a communication manner.

In the following, an exemplary application of the embodiment of the present application in a practical application scenario will be described.

The embodiment of the application can be applied to the following application scenes, for example, in a correction scene aiming at primary school application questions. In the prior art, a correction scheme aiming at application questions of primary school aims at an application question middle-formula, four arithmetic processing is carried out, whether the current formula is correct or not is judged, and if the current formula is correct, the correction is judged; if so, the error is judged. Here, referring to fig. 3 to 5, it can be determined that the existing automatic correction software has the following problems in the automatic correction process: and 1, judging the formula, wherein the judgment of the wedge problem is not carried out, namely, whether the number in the formula appears in the stem is not judged. For example, "9*8 =72" in fig. 5, an erroneous expression is apparent; problem 2, when the formula is judged, the judgment of the intermediate process is not performed, that is, only the formula and the result are judged, whether the intermediate process is written correctly or not is not judged, for example, in fig. 5, "8+2+5=10+4=15", and it is obvious that the intermediate process of "10+4" does not conform to the expectation of the intermediate process and should be erroneous; question 3, no correction of the answer, such as "answer" in fig. 5: the square area is 64 square centimeters "without giving correction results, so that the correction effect of the application problem is not ideal. The solution detection scheme of the application questions provided by the embodiment of the application can be widely applied to correction scenes aiming at the application questions of primary schools, can normally judge all formulas of the answers of students in the application questions, can correctly indicate errors even if the students answer randomly, and can normally judge errors aiming at parts of the answers of the application questions.

The answer detection method for the application questions provided by the embodiment of the application improves the accuracy of the correction result of the application questions by judging whether the application questions are wedged or not through answer of the application questions. According to the embodiment of the application, the stem and the arithmetic training are constructed through the neural network, and specifically, the number in the stem is determined: num_0, num_1, num_2, …, num_m, and numbers that do not appear in the stem (implicit numbers in the stem), the process of determining the numbers that do not appear in the stem may be to construct a high-frequency implicit number table based on the implicit numbers in the stem, such as 1, 3.14, 365, from Top-K numbers in the 20w pieces of data of the statistical training set. Then, based on the numbers in the answer (i.e. the numbers appearing in the stem and the implicit numbers in the stem), constructing a number list (positive example) related to the answer calculation; based on the numbers in the stem that do not appear in the answer, and the randomly selected high frequency implicit numbers, a list of numbers that are not relevant to the answer calculation is constructed (counterexample). When the application problem is corrected, referring specifically to fig. 7, the numbers in the formulas answered by the students are disassembled, and based on the positive examples and the negative examples, whether the numbers are returned is judged to be Yes, if Yes, the current calculation formula is represented by the wedge problem, if not, the current calculation formula is represented by the wedge problem, and the correction error of the problem can be directly judged.

The solution detection method of the application questions provided by the embodiment of the application classifies the question types of the application questions aiming at the question types in the form of 4 large categories in the current application questions, and particularly, aiming at a transverse formula, the middle has only one equal-sign one-step formula. Such a formula feature is currently mainly a formula comprising an equal sign. Judging whether the equation has one equal sign or not, wherein the two sides of the equal sign have digital arithmetic forms; for the take-off, it is possible for such an expression to consist of one or more rows. At present, the sub-features of the formula are shown by a plurality of arithmetic expressions, including a plurality of equal signs. Judging by judging that a plurality of equal signs and digital arithmetic components exist; for solving equations, such equations are now characterized by not only purely numerical equations, but also letters of unknown numbers. The unknown number can be judged by the judgment formula; for the answers, after the classification of the above formulas, when the answer is not in any one of the three formulas, the answer is judged.

The answer detection method of the application questions provided by the embodiment of the application carries out correction of corresponding question types aiming at answers of different question types after classifying the question types of the application questions.

For the horizontal type, the left and right sides are split according to "="; e.g. 1+2=3, after resolution: 1+2, 3; then judging the result of the student answer, judging the rule, judging that the character split in the first step is only a number, judging that no logical character exists, and then judging that the character is an answer, for example, the answer of the operation formula is 3, then comparing the 3 with a labeling platform labeling correct answer, checking whether the answer of the student answer is correct, and if so, continuing to carry out the next step aiming at the correction of the intermediate step. If not, the current horizontal answer is not correct, and an error result is returned. And finally, subtracting the right arithmetic expression from the left arithmetic expression to obtain a new arithmetic expression, and then sending the new arithmetic expression into a four-way arithmetic calculator to judge whether the result of the new arithmetic expression is 0, if the result is equal to 0, the representation is correct, and if the result is not equal to 0, the representation is incorrect.

For the dechucking, first, according to "=", a plurality of formulas, for example, 1+2+3=3+3=6 are disassembled, and the disassembling is completed as follows: 1+2+3, 3+3, 6; and then judging the result of the student answer, judging the rule, judging that the character split in the previous step is only a number, judging that the character is not any logic character, and then judging that the character is an answer, for example, the answer of the operation formula is 6, then comparing the 6 with a labeling platform label correct answer, checking whether the answer of the student answer is correct, and if the answer is correct, continuing to carry out the next step aiming at the correction of the intermediate step. If not, the current take-off answer is not correct, and an error result is returned. Finally, subtracting the right formula from the left formula to obtain a new formula, for example, 1+2+3=3+3=6, and converting the new formula into a subtraction formula, namely: 1+2+3- (3+3), 3+3-6, send two formulas into four arithmetic calculators, judge whether all equal to 0, if all equal to 0, the current take-off formula is correct, otherwise, the current take-off formula is wrong.

For solving the equation, firstly obtaining the answer of the unknown number, judging whether the left and right sides of all the formulas are free of redundant operators by 'disassembling the left and right sides of all the formulas', if not, representing the answer formula of the unknown number, for example, x+3-6=3, x-3=3, x=6, obtaining x=6, then judging whether the student answer is correct by comparing x=6 with the correct answer marked by the question, if the answer is consistent, judging that the result of the student answer is correct, and then carrying out the next step, namely judging the final result by judging the middle process; if the answer is inconsistent with the correct answer, the student answers as errors, and a correction error result is returned; finally, correcting the middle process, and replacing the unknown number with an answer in a carrying-in mode, namely: x+3-6=3 corresponds to 6+3-6=3, and x-3=3 corresponds to 6-3=3. Then, the new formulas, namely 6+3-6-3 and 6-3-3, are obtained by subtracting the left side and the right side of the equal sign, and are sent to a four-rule operation calculator, if all the formulas are equal to 0, the current formula is correct, and otherwise, the current formula is incorrect.

For the answer, referring to fig. 16, first, the comparison words, non-words, nouns and the like of the answer contents of the students and the standard answer contents output by the labeling platform are extracted at the same time, and then the distances between the numbers and the name phrases are judged to compare whether the answer contents of the students are consistent with the standard answer contents output by the labeling platform.

When the answer detection method for the application questions provided by the embodiment of the application is used for classifying the questions, the questions can be trained through a neural network to complete classification of the questions by various question form models, so that recall can be carried out for badcase, and the accuracy of the classification of the questions is improved; meanwhile, aiming at the answer and correction of the application questions, the follow-up training of the questions and the answer content can be performed based on the model, and whether the student text answers are correct or not can be judged.

The solution detection method of the application questions provided by the embodiment of the application mainly comprises the steps of carving questions, classifying the questions (horizontal, vertical, solving equations and answering), and correcting the middle process and results of each question. Finally, automatic correction of the application questions is completed, so that the correction task of teachers aiming at the application questions can be reduced in class; under the lesson, the method can help the pupil to automatically check the correctness of the application questions and help the parents of the pupil to reduce the burden of correcting the application questions.

Continuing with the description below of an exemplary structure of the solution detection apparatus 455 for application questions provided in the embodiment of the present application implemented as a software module, in some embodiments, as shown in fig. 2, the software module stored in the solution detection apparatus 455 for application questions of the memory 440 may include:

The first obtaining module 4551 is configured to obtain stem information of an application question and answer information of the application question; wherein, the stem information at least includes: at least one reference value for solving the application questions, the answer information at least comprising: an operation formula composed of at least two actual values and operators, and an operation result of the operation formula;

the deconstructing module 4552 is configured to deconstruct the operation formula to obtain each actual value in the operation formula;

The comparison module 4553 is configured to compare each actual value with the reference value to obtain a comparison result;

The second obtaining module 4554 is configured to obtain a standard answer of the application question when the comparison result characterizes that the actual numerical value in the operation formula matches the stem information;

And the detection module 4555 is configured to detect an operation process of the operation formula and the operation result based on the standard answer, so as to obtain a detection result for characterizing the correctness of the answer information.

In some embodiments, the comparison module 4553 is further configured to compare the actual value with each of the reference values respectively to obtain an intermediate comparison result; and when the intermediate comparison result represents that the actual numerical value exists in the stem information, generating the comparison result representing that the actual numerical value is matched with the stem information.

In some embodiments, the comparison module 4553 is further configured to generate a detection result for characterizing the answer information error when the intermediate comparison result characterizes that the actual numerical value does not exist in the stem information; acquiring the number of the operation formulas; outputting a manual verification prompt message when the number of the operation formulas is at least two and the actual numerical value which does not exist in the stem information is the intermediate result of the answer information; the manual verification prompt information is used for prompting the detection result to be manually verified.

In some embodiments, the second obtaining module 4554 is further configured to obtain, when the answer types of the application questions include at least two answer types, standard answers of the at least two answer types, where the standard answers include an answer process and a standard operation result; the step of detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for representing the correctness of the answer information comprises the following steps: determining a solution type corresponding to the operation formula; extracting target standard answers corresponding to the answer types of the operation formulas from the standard answers of the at least two answer types; and detecting the operation process of the operation formula and the operation result based on the target standard answer to obtain a detection result for representing the correctness of the answer information.

In some embodiments, the second obtaining module 4554 is further configured to determine an operation result symbol and an unknown number in the operation formula; deconstructing the operation formulas based on the operation result symbols to obtain at least two operation parts; and counting the number of the operation result symbols, the number of the unknowns and the number of the operation parts, and determining the answer type corresponding to the operation formula based on the counting result.

In some embodiments, the second obtaining module 4554 is further configured to match a standard operation result included in the target standard answer with the operation result to obtain a first matching result, and match a solution process included in the target standard answer with an operation process of the operation formula to obtain a second matching result; and generating a detection result for representing the correctness of the answer information by combining the first matching result and the second matching result.

In some embodiments, the detection module 4555 is further configured to determine a solution type corresponding to the operation formula when the standard answer includes a standard operation result and the solution types of the application questions include at least two solutions; when the standard operation result is the same as the operation result, according to the answer type corresponding to the operation formula, carrying out accounting on the operation process of the operation formula to obtain an accounting result; and determining a detection result used for representing the correctness of the answer information based on the accounting result.

In some embodiments, the device further includes a matching module, where the matching module is configured to obtain first text reply information for responding to the application question, and second text reply information in the answer information; and matching the first text reply information with the second text reply information, and determining the correctness of the second text reply information based on a matching result.

In some embodiments, the apparatus further comprises a model module comprising a feature extraction layer, a feature processing layer, an output layer; the model module is used for acquiring the stem information of the application questions and the answer information of the application questions through the feature extraction layer; deconstructing the operation formula through the feature processing layer to obtain each actual value in the operation formula, and comparing each actual value with the reference value to obtain a comparison result; and when the comparison result represents that the actual numerical value in the operation formula is matched with the stem information, acquiring a standard answer of the application question, and detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for representing the correctness of the answer information.

Embodiments of the present application provide a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes the solution detection method for the application questions according to the embodiment of the application.

An embodiment of the present application provides a computer-readable storage medium storing executable instructions, in which the executable instructions are stored, which when executed by a processor, cause the processor to perform the solution detection method for an application question provided by the embodiment of the present application, for example, the solution detection method for an application question as shown in fig. 6.

In some embodiments, the computer readable storage medium may be FRAM, ROM, PROM, EP ROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; but may be a variety of devices including one or any combination of the above memories.

In some embodiments, the executable instructions may be in the form of programs, software modules, scripts, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and they may be deployed in any form, including as stand-alone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

As an example, executable instructions may, but need not, correspond to files in a file system, may be stored as part of a file that holds other programs or data, such as in one or more scripts in a hypertext markup language (HTML, hyper Text Markup Language) document, in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub-programs, or portions of code).

As an example, executable instructions may be deployed to be executed on one computing device or on multiple computing devices located at one site or distributed across multiple sites and interconnected by a communication network.

In summary, the embodiment of the application has the following beneficial effects:

(1) In the process of answering and detecting the application questions, whether answer information of the application questions is wedged or not is judged, so that answering contents which are randomly carried out by students and are irrelevant to the questions are identified, and the accuracy of automatic correction of the application questions is improved.

(2) When the obtained standard answer comprises a standard operation result, according to the answer type corresponding to the operation formula, the operation process of the operation formula is calculated to obtain a calculation result, and the correctness of the answer information is further determined based on the calculation result. Here, through the accounting process, the situation that the student plagiarizes the result is avoided, and the correction accuracy of the student answer is improved.

(3) Aiming at different answer types of the application questions, model training of answer detection models of the application questions in various answer types can be carried out through a neural network, so that classification of the different answer types of the application questions is completed, recall can be carried out aiming at badcase, and accuracy of classification of the answer types is improved.

(4) Aiming at the correction of the answers of the application questions, the model training of the answer detection model of the application questions aiming at the stem and the answer content can be also carried out, so that whether the student character answers are correct or not is judged, and the accuracy of judging the student character answers is improved

(5) By training the answer detection model of the application questions, the accuracy of detection results of answer information for the application questions, which is obtained based on the answer detection model of the application questions, is improved.

The foregoing is merely exemplary embodiments of the present application and is not intended to limit the scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and scope of the present application are included in the protection scope of the present application.

Claims

1. The solution detection method for the application questions is characterized by comprising the following steps:

2. The method of claim 1, wherein comparing each of the actual values with the reference value to obtain a comparison result comprises:

Comparing the actual values with the reference values respectively to obtain an intermediate comparison result;

and when the intermediate comparison result represents that the actual numerical value exists in the stem information, generating the comparison result representing that the actual numerical value is matched with the stem information.

3. The method according to claim 2, wherein the method further comprises:

When the intermediate comparison result represents that the actual numerical value does not exist in the stem information, a detection result for representing the answer information error is generated;

Acquiring the number of the operation formulas;

outputting a manual verification prompt message when the number of the operation formulas is at least two and the actual numerical value which does not exist in the stem information is the intermediate result of the answer information;

the manual verification prompt information is used for prompting the detection result to be manually verified.

4. The method of claim 1, wherein the obtaining the standard answer to the application question comprises:

When the answer types of the application questions comprise at least two answer types, standard answers of the at least two answer types are obtained, wherein the standard answers comprise an answer process and a standard operation result;

the step of detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for representing the correctness of the answer information comprises the following steps:

determining a solution type corresponding to the operation formula;

extracting target standard answers corresponding to the answer types of the operation formulas from the standard answers of the at least two answer types;

And detecting the operation process of the operation formula and the operation result based on the target standard answer to obtain a detection result for representing the correctness of the answer information.

5. The method of claim 4, wherein the determining the solution type corresponding to the operation formula comprises:

determining an operation result symbol and an unknown number in the operation formula;

Deconstructing the operation formulas based on the operation result symbols to obtain at least two operation parts;

And counting the number of the operation result symbols, the number of the unknowns and the number of the operation parts, and determining the answer type corresponding to the operation formula based on the counting result.

6. The method of claim 4, wherein detecting the operation process of the operation formula and the operation result based on the target standard answer to obtain a detection result for characterizing the correctness of the answer information comprises:

matching the standard operation result included in the target standard answer with the operation result to obtain a first matching result, and matching the solution process included in the target standard answer with the operation process of the operation formula to obtain a second matching result;

And generating a detection result for representing the correctness of the answer information by combining the first matching result and the second matching result.

7. The method according to claim 1, wherein detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for characterizing the correctness of the answer information comprises:

when the standard answer comprises a standard operation result and the answer types of the application questions comprise at least two, determining the answer types corresponding to the operation formulas;

When the standard operation result is the same as the operation result, according to the answer type corresponding to the operation formula, carrying out accounting on the operation process of the operation formula to obtain an accounting result;

and determining a detection result used for representing the correctness of the answer information based on the accounting result.

8. The method according to claim 1, wherein the method further comprises:

acquiring first text reply information for responding to the application questions and second text reply information in the answer information;

and matching the first text reply information with the second text reply information, and determining the correctness of the second text reply information based on a matching result.

9. The method of claim 1, wherein the method is implemented by a solution detection model of an application question, the solution detection model of the application question comprises a feature extraction layer, a feature processing layer and an output layer,

The obtaining the stem information of the application questions and the answer information of the application questions comprises the following steps:

acquiring the stem information of the application questions and the answer information of the application questions through the feature extraction layer;

The deconstructing process is performed on the operation formula to obtain each actual value in the operation formula, and each actual value is compared with the reference value to obtain a comparison result, which comprises the following steps:

Deconstructing the operation formula through the feature processing layer to obtain each actual value in the operation formula, and comparing each actual value with the reference value to obtain a comparison result;

When the comparison result represents that the actual numerical value in the operation formula matches the stem information, a standard answer of the application question is obtained, and the operation process of the operation formula and the operation result are detected based on the standard answer to obtain a detection result for representing the correctness of the answer information, wherein the detection result comprises the following steps:

And when the comparison result represents that the actual numerical value in the operation formula is matched with the stem information, acquiring a standard answer of the application question, and detecting the operation process of the operation formula and the operation result based on the standard answer to obtain a detection result for representing the correctness of the answer information.

10. An answer detection device for application questions, the device comprising: