CN120409649A - Adaptive knowledge graph construction and in-depth analysis method supporting multiple types of files - Google Patents
Adaptive knowledge graph construction and in-depth analysis method supporting multiple types of filesInfo
- Publication number
- CN120409649A CN120409649A CN202510574039.2A CN202510574039A CN120409649A CN 120409649 A CN120409649 A CN 120409649A CN 202510574039 A CN202510574039 A CN 202510574039A CN 120409649 A CN120409649 A CN 120409649A
- Authority
- CN
- China
- Prior art keywords
- processing
- item
- matters
- files
- derivative
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/166—Editing, e.g. inserting or deleting
- G06F40/186—Templates
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Animal Behavior & Ethology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to a self-adaptive knowledge graph construction and depth analysis method supporting multiple types of files, which belongs to the technical field of file processing and comprises the steps of receiving processing instructions provided by a user, analyzing all files contained in the processing instructions, at least extracting file entities, matching processing matters for the processing instructions and processing templates corresponding to the processing matters based on analysis results, wherein the processing templates contain processing contents for processing the entities in the files, determining and establishing association relations among all files contained in the processing instructions based on a preset entity alignment technology, executing all processing contents through a preset processing engine according to the determined association relations among the processing templates and the files, and feeding back processing results to the user.
Description
Technical Field
The application relates to the technical field of file processing, in particular to a self-adaptive knowledge graph construction and depth analysis method supporting multiple types of files.
Background
With the development of informatization, the number of documents that enterprises and individuals need to process and review increases exponentially, and in particular, in the fields of law, finance, scientific research and government, the need for multiple document inspection is increasingly prominent. The conventional document processing review mode mainly relies on manual document content review and comparison, and when the number of the document materials to be reviewed and compared is more numerous, the manual processing mode is low in efficiency and is easy to cause the situation of wrong review or left review points due to human factors, so that improvement is needed.
Disclosure of Invention
In order to realize high-efficiency and high-accuracy processing of multiple file materials, the application provides a self-adaptive knowledge graph construction and depth analysis method supporting multiple types of files.
In a first aspect, the present application provides a method for constructing and deeply analyzing an adaptive knowledge graph supporting multiple types of files, which adopts the following technical scheme:
receiving a processing instruction proposed by a user, and analyzing all files contained in the processing instruction, wherein the analysis operation at least comprises extraction of file entities;
Matching processing matters with processing templates corresponding to the processing matters for the processing instructions based on analysis results, wherein the processing templates comprise processing contents for processing entities in the files;
based on a preset entity alignment technology, determining and establishing an association relation among all files contained in the processing instruction;
and executing all processing contents through a preset processing engine according to the determined association relation between the processing template and the file, and feeding back a processing result to a user.
By adopting the technical scheme, the method and the device automatically analyze and extract the key content (entity) in each file according to the technical scheme, match specific processing matters based on the key content of the file, so that the processing requirements of the user are adaptively obtained by analyzing the file content under the condition that the specific processing purpose of a processing instruction is not explicitly triggered by the user, establish the association relation among files based on the entity alignment technology and the extracted file entity, and realize the mutual correspondence between the association relation and the processing content by taking the file entity as a tie between the association relation and the processing content, so that a processing engine can quickly lock the file to be processed and the specific entity in the file according to the association relation when executing the processing content, thereby helping to improve the processing efficiency, and solve the problem that the processing accuracy is influenced by manual errors by replacing manual processing with the intelligent processing mode.
Optionally, the method further comprises:
And after each processing item is obtained by matching, determining whether a derivative requirement meeting a preset derivative condition exists according to the processing item, if so, determining the derivative item corresponding to the derivative requirement, and generating and feeding back a derivative item processing guide to a user according to a processing template corresponding to the derivative item.
By adopting the technical scheme, the processing obtained through direct matching is realized, the derived demand is automatically mined, the upgrading from 'passive response' to 'active service' is realized, the user demand is mined through deep long-term analysis, the processing scheme of user planning and guiding matters is optimized from the long-term aspect, the service intellectualization and automation level is improved, and the user experience is improved.
Optionally, the determining whether there is a derivative requirement meeting a preset derivative condition, if yes, determining a derivative item corresponding to the derivative requirement, including:
Determining whether a relation chain containing the processing matters exists or not through a pre-constructed matter relation graph, wherein the relation chain also contains other matters except the processing matters, and the relation chain has an association relation between adjacent matters and corresponds to an association requirement for describing the association relation;
And calculating the association strength of each item in the association chain and the processing item based on all files contained in the processing instruction, wherein the association strength at least comprises the processing completion rate of processing the item by utilizing all files contained in the processing instruction, wherein the item of which the association strength meets the preset derivative condition is taken as the derivative item, and the corresponding association requirement is taken as the derivative requirement.
By adopting the technical scheme, the item relation graph for describing the association relation between different items is preset, all items in the item relation graph are bridged through association requirements, all items associated with the current processing item are captured based on the relation graph, so that jump from single-point implementation matching to intelligent overall item mining is realized, long-term requirements of users are more comprehensively mined, accurate capturing and intelligent screening of the derivative items are further realized by calculating the association strength of the items and the processing items, and when the association relation is calculated, the processing completion rate of the current existing file for executing the derivative items is fully considered, so that the screening accuracy of the derivative items is improved, and the screened derivative items are more fit with actual objective conditions.
Optionally, the parsing operation further includes performing semantic recognition on the file and extracting keywords;
The matching, based on the analysis result, the processing instruction with the processing item and the processing template corresponding to the processing implementation, including:
Based on a data set corresponding to each item stored in a preset structured database, comparing the similarity between keywords in the analysis result and the data set, and determining all candidate items and the similarity corresponding to each candidate item;
The candidate item with the highest similarity is taken as a processing item, a corresponding processing template is determined, and all the candidate items except the processing item are taken as derivative items.
By adopting the technical scheme, the scheme specifically discloses a matching logic for matching processing matters, and when a plurality of matters (namely, alternative matters) similar to file contents (namely, keywords) appear, the application proposes that the matters with highest similarity are selected as the processing matters, namely, the processing matters are limited to be unique, and meanwhile, in order to ensure that the matters which a user actually wants to complete can be more efficiently completed, all the alternative matters which are not selected as the processing matters are selected as the derivative matters, so that the application is connected with the follow-up processing scheme of the related derivative matters.
Optionally, the method further comprises:
Determining whether a conflict problem exists between a processing item and all derivative items corresponding to the processing item, if so, determining a conflict resolution scheme corresponding to the conflict type based on a conflict type corresponding to the conflict problem, and adding the conflict problem and the conflict resolution scheme corresponding to the conflict problem to the derivative item processing guide;
Wherein, the conflict problem refers to the situation that mutual contradiction occurs when entities shared by the processing items and the derivative items are processed according to the corresponding processing content;
the conflict type at least comprises that the processing matters and the derivative matters are inconsistent with each other due to the fact that the specific contents of the common entities are inconsistent when corresponding processing contents are executed on the common entities.
By adopting the technical scheme, the application further analyzes the processing matters and the derivative matters to pre-judge whether the processing matters and the derivative matters have conflict problems or not in the execution process, the essence of the conflict problems is that the specific processing matters contained in the processing templates corresponding to the processing matters are contradicted, for example, a file A, a business license (containing an address a) and a file B, a new lease contract (containing an address B) are contained in the file uploaded by a user, the finally selected processing matters are business license annual inspection and comprise the derivative matters, an address change record exists in the common entity, the address is required to be locked when the processing matters are executed, the original address (namely the address a) is required to be covered when the processing matters are processed, and therefore, the mutual conflict operation occurs when the processing matters and the derivative matters are processed.
Optionally, the method further comprises:
periodically based on the historical discovered conflict problems, taking matters related to the conflict problems as target matters;
Resolving and reorganizing all processing contents contained in the processing template corresponding to the target item, wherein the conflict contents exist in the reorganized processing contents, are the processing contents only containing corresponding conflict problems, and establish and store conflict relations among the conflict contents, the conflict problems and the corresponding items;
The conflict resolution scheme includes at least prioritizing processing content other than conflicting content.
By adopting the technical scheme, the detailed disassembly and reassembling of the specific processing contents of each item are realized according to the conflict problem among the items at regular intervals, so that when the conflict problem occurs, the processing contents of the items are preferentially executed, and the processing contents except the conflict contents, among the processing contents contained in the items, can improve the conflict detection granularity on one hand, and can realize parallel processing of the processing contents among the items without the conflict problem on the other hand, thereby improving the processing progress of each item.
Optionally, the method further comprises:
each time a processing item is obtained by matching, determining and storing a matching relation between the processing item and a corresponding key file, wherein the key file plays a role in determining the processing item in the process of matching;
periodically analyzing the sensitivity of each key file based on the matching relation stored in the historical period, calibrating the key files with the sensitivity higher than a preset sensitivity threshold as high-sensitivity files, and determining a transaction set for each high-sensitivity file, wherein the transaction set comprises transactions with the matching relation with the corresponding high-sensitivity file;
When processing instructions match processing matters, if the high-sensitivity files exist in all files contained in the processing instructions, the matters in the matters set corresponding to the high-sensitivity files are preferentially called to execute matching operation.
By adopting the technical scheme, the relation between the file and the processing matters is analyzed, the file sensitivity is determined, the file sensitivity can be used for representing the scarcity of the file, such as the file which can be used only when certain special matters are handled (namely the high-sensitivity file), the high-sensitivity file is calibrated and used in the subsequent matching operation of the processing matters, and the matching operation is carried out by calling the matters corresponding to the high-sensitivity file, so that the matching efficiency of the processing matters can be improved.
In a second aspect, the present application provides an adaptive knowledge graph construction and depth analysis system supporting multiple types of files, including:
The file analysis module is used for receiving a processing instruction proposed by a user, analyzing all files contained in the processing instruction, and the analysis operation at least comprises extraction of file entities;
The item matching module is used for matching the processing item with a processing template corresponding to the processing item for the processing instruction based on the analysis result, wherein the processing template comprises processing content for processing the entity in the file;
The cross-file association module is used for determining and establishing association relations among all files contained in the processing instruction based on a preset entity alignment technology;
and the item processing module is used for executing all processing contents through a preset processing engine according to the determined association relation between the processing template and the file and feeding back the processing result to the user.
In a third aspect, the present application provides an adaptive knowledge graph construction and depth analysis apparatus supporting multiple types of files, comprising a memory and a processor, the memory having stored thereon a computer program capable of being loaded by the processor and performing the method according to any of the first aspects.
In a fourth aspect, the present application provides a computer readable storage medium storing a computer program capable of being loaded by a processor and performing the method according to any one of the first aspects.
In summary, the present application includes at least one of the following beneficial technical effects:
1. According to the method, the device and the system, the plurality of files are obtained, automatic analysis of the files is achieved, specific processing matters are matched based on analysis results, and therefore the processing requirements of users are adaptively obtained through analysis of file content under the condition that the specific processing purpose of the processing instructions triggered by the users is not clear;
2. Furthermore, the application establishes the association relation of the cross files based on the knowledge graph data and the entity alignment technology, and finally realizes the efficient processing of the processing matters based on the association relation and the processing template corresponding to the processing matters. In summary, the method helps to improve the processing efficiency by efficiently extracting the association relationship between the induced files from the complicated multiple files, and the intelligent processing mode replaces manual processing, so that the problem that the processing accuracy is affected due to manual errors can be solved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a method for constructing and deeply analyzing an adaptive knowledge graph supporting multiple types of files according to an embodiment of the present application.
Fig. 2 is a block diagram of a system for constructing and analyzing depth of an adaptive knowledge graph supporting multiple types of files according to an embodiment of the present application.
Reference numerals illustrate 201, a file analysis module, 202, a transaction matching module, 203, a cross-file association module, 204 and a transaction processing module.
Detailed Description
The application is described in further detail below with reference to fig. 1-2.
The embodiment of the application discloses a self-adaptive knowledge graph construction and depth analysis method (hereinafter referred to as an analysis method) supporting multiple types of files, which aims to efficiently analyze multiple source files to obtain user demands and then realize response and processing on the user demands. The implementation subject of the analysis method is an adaptive knowledge graph construction and depth analysis system (hereinafter referred to as analysis system for short) supporting multiple types of files, and the specific implementation steps of the analysis method by the analysis system will be specifically described below with reference to fig. 1.
S101, receiving a processing instruction proposed by a user, and analyzing all files contained in the processing instruction, wherein the analysis operation at least comprises extraction of file entities.
In an implementation, a user may access the analysis system in the form of a web page and upload multiple source files to trigger processing instructions, where the multiple source files refer to multiple files, and file formats (e.g., PDF, word/image, etc.), file contents may be different from each other. After receiving all the files contained in the processing instruction, the analysis system analyzes the files, wherein the analysis operation specifically comprises the steps of extracting texts and tables in PDF and Word formats by using an open source library (such as Apache PDFBox and pdfplumber/docx library of Python), recognizing the files in the images and the scanned parts by using an OCR technology (such as TESSERACT and an Alicloud OCR API), and extracting entities (such as names, addresses and dates) from the extracted contents by using a pre-training model (such as SpaCy and an Alicloud NLP).
S102, matching processing matters with processing templates corresponding to the processing matters for the processing instructions based on the analysis result, wherein the processing templates comprise processing contents for processing the entities in the file.
Wherein S102 specifically includes the following sub-steps:
And comparing the keyword in the analysis result with the data set in a similarity mode based on the data set corresponding to each item stored in the preset structured database, determining all the candidate items and the similarity corresponding to each candidate item, taking the candidate item with the highest similarity as a processing item, and determining a corresponding processing template.
In practice, the processing matters are used for characterizing the matters required to be processed by the user, and the analysis system analyzes and obtains the processing matters required to be processed by the user, namely the matching processing matters according to all files recorded when the user triggers the processing instructions and according to analysis results of file contents. Correspondingly, the analysis system is pre-constructed with a structured knowledge base, and the structured knowledge base can be specifically a graph database (Neo 4 j) or a relational database, so as to be used for storing specific file contents required by each processing item and realize the corresponding relation of the processing item-required file contents. If the processing item is "store opening", it corresponds to the need to provide files such as "business license", "lease contract", "sanitary license" and the like.
The analysis system compares the specific file content stored in the knowledge base corresponding to all the processing matters with the analysis result in turn and calculates the similarity, for example, the similarity is calculated by using an NLP model (such as BERT). Exemplary, the analysis result comprises keywords of leasing and shops, and the keywords correspond to processing matters with matching content of opening.
If the comparison process involves a plurality of matters, if a fire check list appears in the analysis result, the corresponding matching content is the processing matters of sanitary permission, and on the contrary, the application further provides the matching score calculated by combining the weight of the TF-IDF key words and the embedding similarity of the BERT sentences, and the matters with the highest matching score are selected as the processing matters, namely the matters with the highest matching degree are selected as the only processing matters.
Further, each processing item corresponds to a processing template, the processing template includes a specific processing content for processing the corresponding processing item, the processing content can be expressed in the form of "entity+processing operation", that is, the specific processing operation performed on the entity is specifically described, for example, when the processing item is "store opening", the corresponding processing template includes at least one processing content that is "verifying the consistency of the business license address and the lease contract address". The corresponding entity in the content is an 'address', and the 'business license address' and the 'lease contract address' are specific description modes for describing the 'address' entity and are derived from different files.
S103, determining and establishing an association relationship among all files contained in the processing instruction based on a preset entity alignment technology;
S104, executing all processing contents through a preset processing engine according to the determined association relation between the processing template and the file, and feeding back the processing result to the user.
In implementation, the analysis system pre-builds a knowledge graph related to each entity, wherein the knowledge graph comprises all description contents for describing the same entity and files to which each description content belongs, so that the analysis system can firstly build association between different description contents of the same entity through an entity alignment technology and the knowledge graph, and then build association relationship between files according to the files to which the description contents belong. And finally, executing the processing content in the processing template based on the association relation until the processing of all the processing content in the processing template is completed, and feeding back a processing result to the user, wherein the processing result specifically comprises processing completion and processing failure. And when the file contained in the processing instruction is missing and the processing content which cannot be processed further exists due to the fact that the file is not related, the processing result corresponding to the feedback is processing failure, and the analysis system is used for feeding back the failure reason, namely outputting the specific processing content of the processing failure.
Optionally, the analysis method further comprises the steps of:
And S105, after each time of matching to obtain the processing matters, taking all the alternative matters except the processing matters as the derivative matters, determining whether the derivative requirement meeting the preset derivative condition exists according to the processing matters, if so, determining the derivative matters corresponding to the derivative requirement, and generating and feeding back the derivative matters processing guidance to the user according to the processing templates corresponding to the derivative matters.
The determining whether the derivative requirement satisfying the preset derivative condition exists in S105, and if so, determining the derivative item corresponding to the derivative requirement specifically includes:
Determining whether a relation chain containing processing matters exists or not through a pre-constructed matter relation graph, wherein the relation chain also contains other matters except the processing matters, and the adjacent matters in the relation chain have association relations and correspond to association requirements for describing the association relations;
And calculating the association strength of each item in the association chain and the processing item based on all the files contained in the processing instruction, taking the item of which the association strength meets the preset derivative condition as the derivative item and the corresponding association requirement as the derivative requirement, wherein the association strength at least comprises the processing completion rate of processing the belonging item by utilizing all the files contained in the processing instruction.
In practice, in conjunction with the foregoing, the present application matches the unique treatment, and then defines the treatment as a derivative for other alternatives than treatment.
In addition, the application further provides a method for screening other derivative matters related to the treatment matters based on the treatment matters and a preset matter relation map, and further obtains a final derivative matter set by taking a union set of all the derivative matters obtained by the two determination screening, and generates corresponding derivative matter treatment guide for each derivative matter according to a treatment template corresponding to the derivative matters and feeds back the corresponding derivative matter treatment guide to a user, wherein the derivative matter treatment guide is used for describing files and specific treatment contents required by executing the corresponding derivative matters. Based on the treatment items and the preset item relation map, the specific method for further screening other derivative items related to the treatment items is as follows:
The item relation map can be embodied in the form of a topological graph, items are used as nodes, items with association relation are connected by arrows, and association requirements (such as the precondition that item A is item B) are annotated at a line section, and the association requirements are used for describing the association relation. When the processing item is "restaurant operation registration", other items associated with the processing item include, in particular, "food operation license", "fire check", "sanitation license handling", and the like, and when the processing is implemented as "company registration", it corresponds to an item associated with "tax registration", and since the "tax registration" implementation itself is a precondition of another item "invoice claim", it corresponds to generation of a relationship chain of "company registration", "tax registration", "invoice claim".
After determining all matters and relation chains related to the processing matters, the analysis system is used for analyzing the association strength between each matter and the processing matters, the calculation rule of the association strength can be predefined by people, and a calculation method for calculating the association strength is exemplarily described below:
The analysis system is pre-defined with a plurality of association types and association strength values corresponding to each association type, wherein the association types comprise legal mandatory association types (namely, government stipulates that certain implementations must be combined and transacted, such as 'restaurant business registration' and 'food business license', 'fire check', 'health license' are legal association matters), business logic dependency types (namely, non-legal requirements and existence of preconditions or dependency relationships among matters, such as 'enterprise loan application' and 'withholding registration', 'credit inquiry', and the like), user potential requirement types (through historical data mining, most users derive transacted matters after transacting the matters, such as 'tax registration', 'social security account' are further transacted after the user transacts 'individual business license').
The analysis system analyzes the association type of the association relation between each item and the processing item, determines the association strength value, compares the association strength value with all file materials contained in the current processing instruction according to the file materials required by handling each item, obtains a similarity value, and characterizes the processing completion rate by using the similarity value. And finally, based on a preset weight value, carrying out weighted summation on the association strength value and the similarity value to obtain a total score, and if the total score is higher than the preset score, considering that the preset derivative condition is met, and taking the corresponding item as the derivative item of the treatment item.
Optionally, the analysis method further comprises the steps of:
determining whether conflict problems exist between the processing items and all the derivative items corresponding to the processing items, if so, determining a conflict resolution scheme corresponding to the conflict types based on the conflict types corresponding to the conflict problems, and adding the conflict problems and the conflict resolution scheme corresponding to the conflict problems to the derivative item processing guide;
wherein, the conflict problem refers to the situation that mutual contradiction occurs when entities shared by the processing items and the derivative items are processed according to the corresponding processing content;
the conflict type at least comprises that the processing matters and the derivative matters are mutually contradicted due to the fact that the specific contents of the common entities are inconsistent when corresponding processing contents are executed on the common entities.
In implementation, the conflict types include conflict at the data level, that is, when the processing matters and the derivative matters execute corresponding processing contents on the shared entity, the conflict is caused by inconsistent specific contents of the shared entity, that is, specific numerical values of fields of the same entity in different files are inconsistent, such as file A business license (including address: XX way 1), file B new lease contract (including address: XX way 2), and based on the scheme, the processing matters are already obtained by matching the current case, namely, business license annual inspection, then the alternative implementation "address change record" obtained by matching file B becomes the derivative matters, at this time, the situation that specific numerical values of the same entity in different files are in conflict occurs, and the two files correspond to different matters, so that a conflict problem is caused between the processing matters and the derivative matters, the original address (namely, XX way 1) needs to be locked when the processing matters are executed (namely, XX way 1) and the original address needs to be covered by the new address (namely, XX way 2) when the processing derivative matters are executed (namely, the processing business license is processed).
The conflict type may also include a conflict at the rule level, i.e., business rules prohibit certain implementation combinations (e.g., "cancellation" and "social security"), so that a conflict problem exists between the transaction "company cancellation" and the derivative "social security payment".
The conflict type may also include a state-level conflict, i.e., both transaction A and transaction B require use of resource X (e.g., file X), but the two differ in the desired state for resource X. For example, the user submits a document C, a loan application form, a corresponding match to obtain the processing matters of "loan management", a document D, a corresponding match to obtain the derivative matters of "loan management", and a property certificate (i.e. resource X) needs to be pressed when "loan management" is handled, but the property certificate needs to be used when "loan management" is handled, i.e. the property certificate needs to be required to be in an "idle" state when "loan management" is handled, but the property certificate is in an "occupied" state when "loan management" is handled, thus causing a conflict problem between the "loan management" and the "loan management".
The analysis system can define conflict types in advance and conflict cases corresponding to each conflict type, wherein the conflict cases comprise specific matters which are mutually conflicting, so that the analysis system can conveniently analyze whether the conflict problem exists after deriving derivative matters each time. In addition, for each conflict type, the analysis system also pre-stores a corresponding conflict resolution scheme, which may include, by way of example, a priority override policy, legal mandatory matters in preference to user active matters in preference to system recommended derivative matters. Such as address changes (legal matters) are processed in preference to business license annual checks (processing matters proposed by users), and can also comprise interaction with users, so that the users can select feedback solutions by themselves and conflict resolution can be realized based on the solutions of the user feedback.
Optionally, the analysis method further comprises:
periodically based on the historical discovered conflict problems, taking matters related to the conflict problems as target matters;
Resolving and reorganizing all processing contents contained in the processing template corresponding to the target item, wherein the conflict contents exist in the reorganized processing contents, are the processing contents only containing corresponding conflict problems, and establish and store conflict relations among the conflict contents, the conflict problems and the corresponding items;
The conflict resolution scheme includes at least prioritizing processing content other than conflicting content.
In implementation, according to the conflict problem generated in the historical period and the found frequency of the conflict problem, when the frequency is higher than the preset frequency, the corresponding matters are taken as target matters, the processing contents contained in the target matters are disassembled and recombined, namely, each processing content is finally disassembled into processing steps which cannot be further disassembled, the processing steps are minimum business processing rules which are independent of other processing steps and can be independently executed, the conflict contents are identified from the disassembled processing contents, other processing contents of the non-conflict contents can be recombined again, and then the non-conflict contents contained in the processing matters can be processed preferentially every time the conflict problem related to the conflict contents is encountered, so that the classification of the processing contents is realized in a disassembled mode, the processing contents which can be processed in parallel are conveniently found, and the processing progress of the corresponding matters is facilitated.
Optionally, the analysis method further comprises the steps of:
each time a processing item is obtained by matching, determining and storing a matching relation between the processing item and a corresponding key file, wherein the key file is a file playing a role in determining the processing item in the process of matching;
Periodically analyzing the sensitivity of each key file based on the matching relation stored in the historical period, calibrating the key files with the sensitivity higher than a preset sensitivity threshold as high-sensitive files, and determining a transaction set for each high-sensitive file, wherein the transaction set comprises transactions with the matching relation with the corresponding high-sensitive file;
when the processing instruction matches the processing items, if the high-sensitivity files exist in all the files contained in the processing instruction, the items in the item set corresponding to the high-sensitivity files are preferentially called to execute the matching operation.
In the implementation, in combination with the specific matching process of the related matching processing matters, the method is used for comparing the file analysis result with the content of the specific file corresponding to the matters stored in the knowledge base in a similarity manner, so that in the process, the analysis system establishes a matching relationship of the matters-key files, and the analysis result extracted from the key files can be considered to be consistent with the content of the specific file stored by the corresponding matters, that is, all the key files corresponding to the matters are matched together to obtain the corresponding matters.
After the matching relation is determined, the analysis system determines the occurrence times of each key file in all the matching relations as the sensitivity of the corresponding key file, and considers that the smaller the occurrence times are, the higher the corresponding sensitivity is, and when the sensitivity is higher than a preset sensitivity threshold, the corresponding key file is marked as a high-sensitivity file, correspondingly, when the processing instruction is required to be matched for processing matters each time later, whether the processing instruction contains the high-sensitivity file is preferentially determined, and if the processing instruction exists, the matters with the matching relation with the high-sensitivity file are preferentially called to execute the matching operation, so that the matching sequence with all matters is limited.
The embodiment of the application also discloses a system for constructing the self-adaptive knowledge graph and analyzing the depth, which supports the multi-type files. Referring to fig. 2, comprising:
the file analysis module 201 is configured to receive a processing instruction set by a user, analyze all files included in the processing instruction, and perform an analysis operation at least including extracting a file entity;
The item matching module 202 is configured to match a processing item with a processing template corresponding to the processing item for the processing instruction based on the analysis result, where the processing template includes processing content for processing an entity in the file;
the cross-file association module 203 is configured to determine and establish an association relationship between all files included in the processing instruction based on a preset entity alignment technique;
and the item processing module 204 is used for executing all processing contents through a preset processing engine according to the determined association relationship between the processing template and the file and feeding back the processing result to the user.
Optionally, the system further comprises a derivative requirement mining module, wherein the derivative requirement mining module is used for determining whether a derivative requirement meeting a preset derivative condition exists according to the processing matters after the processing matters are obtained by matching, if so, determining the derivative matters corresponding to the derivative requirement, and generating and feeding back the derivative matter processing guidance to the user according to the processing templates corresponding to the derivative matters.
Optionally, the derived demand mining module is further configured to determine, through a pre-constructed item relationship graph, whether a relationship chain including processing items exists, wherein the relationship chain further includes other items except the processing items, and an association relationship exists between adjacent items in the relationship chain and corresponds to an association demand for describing the association relationship, where the item relationship graph is used to characterize the association relationship between different items, and further is configured to calculate, based on all files included in the processing instruction, association strength between each item in the relationship chain and the processing item, respectively, and use an item whose association strength satisfies a preset derived condition as the derived item and use the corresponding association demand as the derived demand, where the association strength at least includes a processing completion rate of processing the belonging item by using all files included in the processing instruction.
Optionally, the item matching module 202 is further configured to compare the keyword in the analysis result with the data set according to the data set corresponding to each item stored in the preset structured database, determine all the alternatives, and the similarity corresponding to each alternative, and determine the alternative with the highest similarity as the processing item and the corresponding processing template, and use all the alternatives except the processing item as the derivative item.
Optionally, the system further comprises a conflict resolution module, wherein the conflict resolution module is used for determining whether a conflict problem exists between the processing items and all the derivative items corresponding to the processing items, if so, based on a conflict type corresponding to the conflict problem, a conflict resolution scheme corresponding to the conflict type is determined, and the conflict problem and the conflict resolution scheme corresponding to the conflict problem are added to the derivative item processing guide, wherein the conflict problem refers to the situation that the processing items and the entity shared by the derivative items are mutually contradicted when the processing items and the entity shared by the derivative items are processed according to corresponding processing content, and the conflict type at least comprises that the processing items and the derivative items are mutually contradicted due to the fact that the specific content of the shared entity is inconsistent when the corresponding processing content is executed on the shared entity.
Optionally, the system further comprises a processing content classification module, wherein the processing content classification module is used for regularly taking matters related to the conflict problems as target matters based on the conflict problems found by history, and resolving and recombining all processing contents contained in the processing templates corresponding to the target matters, wherein the conflict contents exist in the recombined processing contents, are the processing contents only containing the corresponding conflict problems, and establish and store conflict relations among the conflict contents, the conflict problems and the corresponding matters, and the conflict resolution scheme at least comprises the processing contents except the conflict contents.
Optionally, the system further comprises a sensitivity matching module, wherein the sensitivity matching module is used for determining and storing matching relations between the processing matters and key files corresponding to the processing matters each time when the processing matters are obtained by matching, the key files are files which play a role in determining the processing matters in a matching mode, the sensitivity of each key file is analyzed on the basis of the matching relations stored in a historical period, key files with sensitivity higher than a preset sensitivity threshold are calibrated to be high-sensitive files, an event set is determined for each high-sensitive file, the event set contains matters with matching relations with the corresponding high-sensitive files, and when the processing matters are matched for the processing instructions, if the high-sensitive files exist in all files contained by the processing instructions, the events in the event set corresponding to the high-sensitive files are preferentially called to execute matching operation.
The embodiment of the application also discloses a device for constructing and deeply analyzing the self-adaptive knowledge graph supporting the multi-type file, which comprises a memory and a processor, wherein the memory stores a computer program which can be loaded by the processor and execute the method for constructing and deeply analyzing the self-adaptive knowledge graph supporting the multi-type file.
The embodiment of the application also discloses a computer readable storage medium which stores a computer program capable of being loaded by a processor and executing the self-adaptive knowledge graph construction and depth analysis method supporting the multi-type files, wherein the computer readable storage medium comprises various media capable of storing program codes, such as a U disk, a mobile hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, a RAM), a magnetic disk or an optical disk.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.
The above embodiments are only for illustrating the technical solution of the present application, and not for limiting the scope of application. It will be apparent that the described embodiments are merely some, but not all, embodiments of the application. Based on these embodiments, all other embodiments that may be obtained by one of ordinary skill in the art without inventive effort are within the scope of the application.
Claims (10)
1. A method for constructing and deeply analyzing an adaptive knowledge graph supporting multiple types of files is characterized by comprising the following steps:
receiving a processing instruction proposed by a user, and analyzing all files contained in the processing instruction, wherein the analysis operation at least comprises extraction of file entities;
Matching processing matters with processing templates corresponding to the processing matters for the processing instructions based on analysis results, wherein the processing templates comprise processing contents for processing entities in the files;
based on a preset entity alignment technology, determining and establishing an association relation among all files contained in the processing instruction;
and executing all processing contents through a preset processing engine according to the determined association relation between the processing template and the file, and feeding back a processing result to a user.
2. The method for adaptive knowledge graph construction and depth analysis supporting multiple types of files according to claim 1, wherein the method further comprises:
And after each processing item is obtained by matching, determining whether a derivative requirement meeting a preset derivative condition exists according to the processing item, if so, determining the derivative item corresponding to the derivative requirement, and generating and feeding back a derivative item processing guide to a user according to a processing template corresponding to the derivative item.
3. The method for constructing and analyzing the depth of the adaptive knowledge graph supporting the multi-type document according to claim 2, wherein the determining whether there is a derivative requirement satisfying a preset derivative condition, if so, determining a derivative item corresponding to the derivative requirement, includes:
Determining whether a relation chain containing the processing matters exists or not through a pre-constructed matter relation graph, wherein the relation chain also contains other matters except the processing matters, and the relation chain has an association relation between adjacent matters and corresponds to an association requirement for describing the association relation;
And calculating the association strength of each item in the association chain and the processing item based on all files contained in the processing instruction, wherein the association strength at least comprises the processing completion rate of processing the item by utilizing all files contained in the processing instruction, wherein the item of which the association strength meets the preset derivative condition is taken as the derivative item, and the corresponding association requirement is taken as the derivative requirement.
4. The method for constructing and deeply analyzing an adaptive knowledge graph supporting multiple types of files according to claim 3, wherein the parsing operation further comprises semantic recognition of the files and keyword extraction;
The matching, based on the analysis result, the processing instruction with the processing item and the processing template corresponding to the processing implementation, including:
Based on a data set corresponding to each item stored in a preset structured database, comparing the similarity between keywords in the analysis result and the data set, and determining all candidate items and the similarity corresponding to each candidate item;
The candidate item with the highest similarity is taken as a processing item, a corresponding processing template is determined, and all the candidate items except the processing item are taken as derivative items.
5. The method for adaptive knowledge graph construction and depth analysis supporting multiple types of documents according to claim 4, wherein the method further comprises:
Determining whether a conflict problem exists between a processing item and all derivative items corresponding to the processing item, if so, determining a conflict resolution scheme corresponding to the conflict type based on a conflict type corresponding to the conflict problem, and adding the conflict problem and the conflict resolution scheme corresponding to the conflict problem to the derivative item processing guide;
Wherein, the conflict problem refers to the situation that mutual contradiction occurs when entities shared by the processing items and the derivative items are processed according to the corresponding processing content;
the conflict type at least comprises that the processing matters and the derivative matters are inconsistent with each other due to the fact that the specific contents of the common entities are inconsistent when corresponding processing contents are executed on the common entities.
6. The method for adaptive knowledge graph construction and depth analysis supporting multiple types of documents according to claim 4, wherein the method further comprises:
periodically based on the historical discovered conflict problems, taking matters related to the conflict problems as target matters;
Resolving and reorganizing all processing contents contained in the processing template corresponding to the target item, wherein the conflict contents exist in the reorganized processing contents, are the processing contents only containing corresponding conflict problems, and establish and store conflict relations among the conflict contents, the conflict problems and the corresponding items;
The conflict resolution scheme includes at least prioritizing processing content other than conflicting content.
7. The method for adaptive knowledge graph construction and depth analysis supporting multiple types of files according to claim 1, wherein the method further comprises:
each time a processing item is obtained by matching, determining and storing a matching relation between the processing item and a corresponding key file, wherein the key file plays a role in determining the processing item in the process of matching;
periodically analyzing the sensitivity of each key file based on the matching relation stored in the historical period, calibrating the key files with the sensitivity higher than a preset sensitivity threshold as high-sensitivity files, and determining a transaction set for each high-sensitivity file, wherein the transaction set comprises transactions with the matching relation with the corresponding high-sensitivity file;
When processing instructions match processing matters, if the high-sensitivity files exist in all files contained in the processing instructions, the matters in the matters set corresponding to the high-sensitivity files are preferentially called to execute matching operation.
8. A self-adaptive knowledge graph construction and depth analysis system supporting multiple types of files is characterized by comprising,
The file analysis module (201) is used for receiving a processing instruction proposed by a user, analyzing all files contained in the processing instruction, and the analysis operation at least comprises extraction of file entities;
The item matching module (202) is used for matching the processing item with a processing template corresponding to the processing item for the processing instruction based on the analysis result, wherein the processing template comprises processing content for processing the entity in the file;
a cross-file association module (203) for determining and establishing association relationships among all files contained in the processing instruction based on a preset entity alignment technology;
And the item processing module (204) is used for executing all processing contents through a preset processing engine according to the determined association relation between the processing template and the file and feeding back the processing result to the user.
9. An adaptive knowledge graph construction and depth analysis device supporting multiple types of files, comprising a memory and a processor, wherein the memory has stored thereon a computer program that can be loaded by the processor and that performs the method according to any of claims 1 to 7.
10. A computer readable storage medium, characterized in that a computer program is stored which can be loaded by a processor and which performs the method according to any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510574039.2A CN120409649A (en) | 2025-05-06 | 2025-05-06 | Adaptive knowledge graph construction and in-depth analysis method supporting multiple types of files |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510574039.2A CN120409649A (en) | 2025-05-06 | 2025-05-06 | Adaptive knowledge graph construction and in-depth analysis method supporting multiple types of files |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN120409649A true CN120409649A (en) | 2025-08-01 |
Family
ID=96520323
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202510574039.2A Pending CN120409649A (en) | 2025-05-06 | 2025-05-06 | Adaptive knowledge graph construction and in-depth analysis method supporting multiple types of files |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN120409649A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113157947A (en) * | 2021-05-20 | 2021-07-23 | 中国工商银行股份有限公司 | Knowledge graph construction method, tool, device and server |
| CN115238945A (en) * | 2022-04-11 | 2022-10-25 | 平安国际智慧城市科技股份有限公司 | Travel guide generation method, device and equipment and storage medium |
| CN115472256A (en) * | 2022-09-16 | 2022-12-13 | 西南医科大学附属医院 | Medical laboratory report analysis system and method based on clinical laboratory medical big data |
| CN119149757A (en) * | 2024-11-13 | 2024-12-17 | 北京奇虎科技有限公司 | Code map construction method, device, electronic equipment and computer storage medium |
| CN119441451A (en) * | 2023-08-03 | 2025-02-14 | 宁德时代新能源科技股份有限公司 | Document parsing method, device and equipment |
| CN119719308A (en) * | 2024-12-25 | 2025-03-28 | 中国工商银行股份有限公司 | Business information query method, device, electronic device, storage medium and program product |
-
2025
- 2025-05-06 CN CN202510574039.2A patent/CN120409649A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113157947A (en) * | 2021-05-20 | 2021-07-23 | 中国工商银行股份有限公司 | Knowledge graph construction method, tool, device and server |
| CN115238945A (en) * | 2022-04-11 | 2022-10-25 | 平安国际智慧城市科技股份有限公司 | Travel guide generation method, device and equipment and storage medium |
| CN115472256A (en) * | 2022-09-16 | 2022-12-13 | 西南医科大学附属医院 | Medical laboratory report analysis system and method based on clinical laboratory medical big data |
| CN119441451A (en) * | 2023-08-03 | 2025-02-14 | 宁德时代新能源科技股份有限公司 | Document parsing method, device and equipment |
| CN119149757A (en) * | 2024-11-13 | 2024-12-17 | 北京奇虎科技有限公司 | Code map construction method, device, electronic equipment and computer storage medium |
| CN119719308A (en) * | 2024-12-25 | 2025-03-28 | 中国工商银行股份有限公司 | Business information query method, device, electronic device, storage medium and program product |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10579651B1 (en) | Method, system, and program for evaluating intellectual property right | |
| US20150032645A1 (en) | Computer-implemented systems and methods of performing contract review | |
| Jallan et al. | Text mining of the securities and exchange commission financial filings of publicly traded construction firms using deep learning to identify and assess risk | |
| Breitung et al. | A canonical correlation approach for selecting the number of dynamic factors | |
| US20220374401A1 (en) | Determining domain and matching algorithms for data systems | |
| WO2022271431A1 (en) | System and method that rank businesses in environmental, social and governance (esg) | |
| Fazzinga et al. | Online and offline classification of traces of event logs on the basis of security risks | |
| CN119205338A (en) | Contract risk identification method, device, equipment, storage medium and program product | |
| Dey et al. | Which pull requests get accepted and why? a study of popular npm packages | |
| US11494850B1 (en) | Applied artificial intelligence technology for detecting anomalies in payroll data | |
| Haryono et al. | Aspect-based sentiment analysis of financial headlines and microblogs using semantic similarity and bidirectional long short-term memory | |
| Jagdish et al. | Identification of End‐User Economical Relationship Graph Using Lightweight Blockchain‐Based BERT Model | |
| CN120410761A (en) | Intelligent financial risk early warning method based on multimodal data fusion | |
| CN114723542A (en) | Detection method, device, equipment and storage medium for tax abnormal binding behavior | |
| CN114139539A (en) | Enterprise social responsibility index quantification method, system and application | |
| Jones et al. | Identifying essential nonprofits with a novel NLP method | |
| Li et al. | Research on Efficiency in Credit Risk Prediction Using Logistic‐SBM Model | |
| Castro Fernandez | What is the value of data? a theory and systematization | |
| Huang et al. | An intelligent mechanism to automatically discover emerging technology trends: Exploring regulatory technology | |
| CN120409649A (en) | Adaptive knowledge graph construction and in-depth analysis method supporting multiple types of files | |
| Stasinos | Uncovering Smart Contract Vulnerabilities: A Systematic Literature Review and a Deep Learning Approach to Predict Known and Unknown Threats. | |
| Rattan et al. | Detecting high level similarities in source code and beyond | |
| US20200334595A1 (en) | Company size estimation system | |
| Wang et al. | Diminishing downsides of data mining | |
| CN119444445B (en) | Strategy customization method, device, electronic device and medium based on evaluation model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |