CN120806449B - A Smart Process Task Allocation Method and System Based on Activiti and Machine Learning - Google Patents
A Smart Process Task Allocation Method and System Based on Activiti and Machine LearningInfo
- Publication number
- CN120806449B CN120806449B CN202510874800.4A CN202510874800A CN120806449B CN 120806449 B CN120806449 B CN 120806449B CN 202510874800 A CN202510874800 A CN 202510874800A CN 120806449 B CN120806449 B CN 120806449B
- Authority
- CN
- China
- Prior art keywords
- task
- approval
- approver
- time
- model
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
- G06Q10/063114—Status monitoring or status determination for a person or group
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides an intelligent flow task distribution method and system based on the combination of Activiti and machine learning, and relates to the technical field of machine learning; the method comprises the steps of processing and calculating multi-dimensional road conference flow data, extracting road conference type characteristics, task emergency characteristics and task load characteristics of compliance approvers, constructing a task distribution model, training the task distribution model based on the extracted characteristics, enabling the model to be used for predicting task approval time of different approvers, integrating the trained task distribution model into an actiti flow engine, calling the model to calculate the predicted approval time of candidate approvers when the flow task reaches an approval node, dynamically distributing the task according to the predicted approval time, selecting the compliance approver with the shortest predicted approval time to execute the current approval task, periodically collecting new added approval data, and retraining the task distribution model to iteratively optimize the prediction effect.
Description
Technical Field
The invention relates to the technical field of machine learning, in particular to an intelligent flow task allocation method and system based on the combination of Activiti and machine learning.
Background
As the informatization degree increases gradually, more and more enterprises need to complete online electronization of various business approval processes. Activiti acts as a Java-based open source workflow and business process management engine, conforms to the apache2.0 open source protocol, and conforms to the BPMN2.0 standard. The system is favored because of light weight and easy integration with mainstream frames such as Spring, and is convenient for visual design, automatic execution and monitoring of business processes.
However, the native Activiti engine relies primarily on a fixed allocation when assigning approval tasks, i.e., assigning tasks to fixed approvers by respective administrators according to experience and judgment, or assigning according to simple rules. The allocation mode is firstly low in efficiency, difficult to quickly respond to a large number of or complex approval tasks, secondly unbalanced in allocation, insufficient in resource utilization caused by the fact that actual workload and processing capacity of an approver cannot be effectively estimated, and finally poor in predictability, analysis and prediction are difficult to carry out according to historical data, and approval process task allocation cannot be optimized, so that approval efficiency and quality are low.
Disclosure of Invention
The invention aims to provide an intelligent flow task distribution method and system based on the combination of actiti and machine learning, so as to solve the problems that the original actiti engine in the background art mainly depends on fixed distribution or distribution according to simple rules when performing approval task distribution, and is low in efficiency, unbalanced in distribution and lack of predictability.
The intelligent flow task allocation method based on the combination of actiti and machine learning comprises the following steps of collecting multidimensional road conference flow data and cleaning the data, processing and calculating the multidimensional road conference flow data, extracting road conference type characteristics, task urgency characteristics and task load characteristics of compliance approvers, constructing a task allocation model, training the task allocation model based on the extracted characteristics, wherein the model is used for predicting task approval time of different approvers, integrating the trained task allocation model into an actiti flow engine, calling the model to calculate predicted approval time of candidate approvers when the flow tasks reach approval nodes, dynamically allocating tasks according to the predicted approval time, selecting the compliance approvers with shortest predicted approval time to execute the current approval task, periodically collecting new added approval data, and retraining the task allocation model to iteratively optimize the predicted effect.
Optionally, the preprocessing step specifically includes recording data sources and processing logic in the process of integrating multidimensional route conference flow data, performing desensitization processing on sensitive information, retaining characteristics related to service, removing data missing values, converting category type data into unique binary vectors by adopting single thermal coding, and performing normalization processing on the numerical type data to enable the numerical type data to fall into a [0,1] interval.
Optionally, the multidimensional route conference flow data comprise a route conference type, task submission time, conference starting time, approver information, professional field matching degree and online state, wherein the route conference type is used for determining task workload according to different types, the task submission time and the conference starting time are used for calculating task emergency degree, the professional field matching degree is used for evaluating suitability of the approver and the task type, the online state is used for judging real-time availability of the approver, whether the approver is in compliance is judged by combining the professional field matching degree and the online state, and the approval passing time and the number of the approver to-be-processed tasks are used for calculating load of the approver.
Optionally, the step of judging whether the approver is qualified by combining the professional field matching degree and the online state specifically comprises the steps of judging whether the type of the professional field is matched with the conference type one by one, if yes, marking as 1, otherwise marking as 0, judging whether the approver is online, if yes, marking as 1, otherwise marking as 0, and if the type of the professional field is matched with the conference type and the approver is online, judging that the approver is qualified.
Optionally, the task emergency degree characteristic calculating step specifically comprises the steps of obtaining a time difference between task submitting time and meeting starting time, converting the time difference into an emergency degree value by adopting a time attenuation function, wherein the smaller the time difference is, the higher the emergency degree value is, introducing random noise conforming to normal distribution in the calculating process, and simulating uncertainty in an actual approval scene.
Optionally, the method for calculating the task load characteristics of the compliance approver comprises the steps of counting the number of the to-be-approved tasks of the current approver, calculating the emergency degree of each to-be-approved task, sequencing the to-be-approved tasks from high to low according to the emergency degree, and calculating the normalized task load value according to the position of the current task in sequencing and the historical maximum task number.
Optionally, the task allocation model building step specifically comprises the steps of building a linear regression model based on task type characteristics, task emergency degree characteristics and compliance approver task load characteristics, carrying out logarithmic transformation on the approval time, ensuring that the prediction time is positive, introducing cross item characteristics to optimize the prediction, and solving model parameters through a least square method to obtain optimal parameters.
Optionally, the step of integrating the trained task allocation model into the active flow engine specifically includes configuring a custom listener in the BPMN file, and triggering the listener when the flow task reaches the approval node.
The method comprises the steps of selecting a user-defined monitor, wherein the user-defined monitor specifically comprises the steps of inheriting TASKLISTENER interfaces of an action engine, rewriting a notify method, acquiring context information of a current task in the notify method, calling an API of a task distribution model, inputting task type characteristics, task emergency characteristics and task load characteristics of compliance approvers, acquiring predicted approval time of candidate approvers, determining optimal approvers, dynamically setting approver variables through a runtime service interface of the action engine, and updating task distribution through RuntimeService interfaces.
On the other hand, the invention also provides an intelligent flow task distribution system based on the combination of Activiti and machine learning, which comprises a data collection module, a feature extraction module, a model construction module, an integration module, a distribution module and a training module, wherein the data collection module is used for collecting multidimensional route meeting flow data and cleaning the data, the feature extraction module is used for processing and calculating the multidimensional route meeting flow data and extracting route meeting type features, task urgency features and compliance approval personnel task load features, the model construction module is used for constructing a task distribution model, training the task distribution model based on the extracted features, the model is used for predicting task approval time of different approval personnel, the integration module is used for integrating the trained task distribution model into an Activiti flow engine, when a flow task reaches an approval node, the model is called to calculate the predicted approval time of candidate approval personnel, the distribution module is used for dynamically distributing the task according to the predicted approval time, the compliance approval personnel with the shortest predicted approval time is selected to execute the current approval task, and the task distribution module is used for periodically collecting new increased approval data and retraining the task distribution model to iteratively optimize the predicted approval effect.
Compared with the prior art, the invention has the beneficial effects that:
The application combines the active workflow engine and the machine learning model to dynamically allocate tasks, reduces the approval waiting time and the approval time, and remarkably improves the efficiency of the approval process. According to the historical performance and the current working state of the approver, tasks are intelligently distributed, the balanced distribution of the workload is realized, and the resource waste is avoided. By utilizing the machine learning technology, the system can continuously learn and optimize the allocation strategy, adapt to the change of approval requirements and realize intelligent flow management. The intelligent task allocation method combining the active process engine and the machine learning not only improves the efficiency and quality of the approval process, but also improves the intelligent level of the system by dynamically optimizing the resource allocation, thereby providing a more efficient and accurate process task allocation scheme for enterprises.
Drawings
FIG. 1 is a flow chart of the steps of the method of the present invention.
FIG. 2 is a schematic diagram of the system structure of the present invention.
In the figure, a 10-data collection module, a 20-feature extraction module, a 30-model construction module, a 40-integration module, a 50-distribution module and a 60-optimization module are shown.
Detailed Description
The aspects of the present invention will become apparent from the following detailed description of embodiments of the invention, given in conjunction with the accompanying drawings, wherein it is evident that the embodiments described are merely some, but not all embodiments of the invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate in order to describe the embodiments of the application herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless expressly stated otherwise, as understood by those skilled in the art. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. The term "and/or" as used herein includes all or any element and all combination of one or more of the associated listed items.
It will be understood by those skilled in the art that all terms (including 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 unless defined otherwise. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
It should be understood that the sequence number and the size of each step in this embodiment do not mean the sequence of execution, and the execution sequence of each process is determined by the function and the internal logic of each process, and should not be construed as limiting the implementation process of the embodiment of the present application.
It should be noted that, without conflict, the embodiments of the present application and features of the embodiments may be combined with each other. The application will be described in detail below with reference to the drawings in connection with embodiments.
Referring to fig. 1, the intelligent flow task allocation method based on the combination of Activiti and machine learning includes the steps of collecting multidimensional route meeting flow data and cleaning the data, processing and calculating the multidimensional route meeting flow data, extracting route meeting type features, task urgency features and compliance approver task load features, constructing a task allocation model, training the task allocation model based on the extracted features, enabling the model to be used for predicting task approval time of different approvers, integrating the trained task allocation model into an Activiti flow engine, calling the model to calculate predicted approval time of candidate approvers when the flow tasks reach approval nodes, dynamically allocating tasks according to the predicted approval time, selecting compliance approvers with shortest predicted approval time to execute current approval tasks, periodically collecting new added approval data, and retraining the task allocation model to iteratively optimize the predicted effect.
The method comprises the steps of collecting multidimensional conference flow data of a road show, extracting key characteristics such as conference type characteristics, task emergency characteristics, task load characteristics of compliance approvers and the like, constructing a task distribution model to distribute tasks, integrating the model into an action flow engine, calling the model to calculate the predicted approval time of candidate approvers when the flow tasks reach approval nodes, dynamically distributing the tasks according to the predicted approval time, selecting the compliance approvers with the shortest predicted approval time to execute the current approval tasks, automatically selecting a mechanism of the optimal approvers by a system to obviously shorten approval flow time, realizing dynamic intelligent distribution of the approval tasks, overcoming the defects of low efficiency and uneven resource distribution of the traditional fixed distribution mode, periodically collecting the newly added approval data, and retraining the task distribution model to iteratively optimize the prediction effect. The application combines the active workflow engine and the machine learning model to dynamically allocate tasks, reduces the approval waiting time and the approval time, and remarkably improves the efficiency of the approval process. According to the historical performance and the current working state of the approver, tasks are intelligently distributed, the balanced distribution of the workload is realized, and the resource waste is avoided. By utilizing the machine learning technology, the system can continuously learn and optimize the allocation strategy, adapt to the change of approval requirements and realize intelligent flow management. By combining the active process engine and the intelligent task allocation of machine learning, the efficiency and quality of the approval process are improved, the intelligent level of the system is improved through dynamic optimization of resource allocation, and a more efficient and accurate process task allocation scheme is provided for enterprises.
In some embodiments, the preprocessing step specifically includes recording data sources and processing logic during the multidimensional route conference process data integration, performing desensitization processing on sensitive information, retaining characteristics related to service, eliminating data missing values, converting category type data into unique binary vectors by adopting single thermal coding, and performing normalization processing on the numerical type data to enable the numerical type data to fall into a [0,1] interval.
Specifically, the collected multidimensional road conference flow data is subjected to standardized data cleaning, the data quality is improved through desensitization processing and missing value elimination, and reliable input is provided for model training. The data tracing mechanism enhances the traceability of the feature engineering and reduces the influence of data abnormality on the distribution result. The processing treatment is carried out on the collected multidimensional road conference flow data, and the processing treatment concretely comprises the steps that professional field matching is recorded as 1, otherwise recorded as 0, on-line recording of approval personnel is recorded as 1, otherwise recorded as 0, the approval personnel records as 1 through approval, otherwise recorded as 0, and the road conference type comprises three types of industry viewpoint reporting, company viewpoint reporting and company performance exchanging, and is recorded as follows: In which, in the process, For the road conference type, different conference types need to be confirmed and checked by compliance approvers to have larger information difference, so that the time consumption is different, and the single-hot coding is adopted to convert the type data into unique binary vectors, and normalization processing is carried out on the numerical data to enable the numerical data to fall into the [0,1] interval.
In some embodiments, the multidimensional conference flow data comprises a conference type, task submission time, conference starting time, approver information, professional field matching degree, on-line state, approval passing time and number of to-be-approved tasks of the approvers, wherein the conference type is used for determining task workload according to different types, the task submission time and the conference starting time are used for calculating task emergency degree, the professional field matching degree is used for evaluating suitability of the approvers and the task type, the on-line state is used for judging real-time availability of the approvers, and judging whether the approvers are in compliance or not by combining the professional field matching degree and the on-line state, and the approval passing time and the number of to-be-processed tasks of the approvers are used for calculating load of the approvers.
Specifically, the comprehensive road conference type, task submission time, conference start time, approver information, professional field matching degree and online state are evaluated through multidimensional data such as duration, number of to-be-approved tasks of the approver and the like, and a comprehensive task allocation index system is constructed. The efficiency and quality of the approval process are improved while the compliance is ensured through a combined judging mechanism of the matching degree, the online state, the task emergency degree and the load of the approver in the professional field.
In some embodiments, the step of determining whether the approver is qualified by combining the professional field matching degree and the online state specifically includes the steps of determining whether the type of the professional field is matched with the conference type in a one-to-one correspondence, determining whether the type of the professional field is matched with the conference type, if yes, marking as 1, otherwise marking as 0, determining whether the approver is online, if yes, marking as 1, otherwise marking as 0, and if the type of the professional field is matched with the conference type and the approver is online, determining that the approver is qualified.
Specifically, a binary quantization standard is adopted to realize efficient compliance screening, and a candidate list is accurately generated through professional matching and on-line state double verification, so that invalid task allocation is effectively reduced.
In some embodiments, the calculation step of the task emergency degree characteristic specifically comprises the steps of obtaining a time difference between task submitting time and meeting starting time, converting the time difference into an emergency degree value by adopting a time attenuation function, wherein the smaller the time difference is, the higher the emergency degree value is, introducing random noise conforming to normal distribution in the calculation process, and simulating uncertainty in an actual approval scene.
Specifically, a self-defined time decay function is adopted to convert the time interval into the task emergency degree, and the calculation formula is as follows: In which, in the process, In order to achieve a degree of task urgency,For the start time of the road conference,For the time of the process task commit,Is distributed as a standard whole. The method is based on the emergency degree quantization model of the time attenuation function, combines with the random noise simulation actual scene, more accurately identifies the task priority difference, and ensures that the emergency task is processed in time.
In some embodiments, the method for calculating the task load characteristics of the compliance approver comprises the steps of counting the number of to-be-approved tasks of the current approver, calculating the emergency degree of each to-be-approved task, sequencing the to-be-approved tasks according to the emergency degree from high to low, and calculating a normalized task load value according to the position of the current task in sequencing and the historical maximum task number.
Specifically, after the approval task is submitted, the approval task of the compliance approver is as followsThe task number is n, the task emergency degree of each task is calculated respectively, the task emergency degree is sequenced from high to low, and the sequencing is carried outThe calculation formula of the compliance approver task load is as follows: In which, in the process, For compliance approver task loads,For the number of positions in which the task urgency is ordered from high to low,The maximum value of the historical task number n to be approved is used for all compliance approvers. According to the load evaluation method for dynamic sequencing and normalization processing, the work pressure of the approver is reflected in real time, and intelligent balanced allocation of the task quantity is realized.
In some embodiments, the step of constructing the task allocation model specifically comprises constructing a linear regression model based on task type characteristics, task emergency degree characteristics and compliance approver task load characteristics, carrying out logarithmic transformation on the approval time, ensuring that the prediction time is positive, introducing cross term characteristics to optimize the prediction, and solving model parameters through a least square method to obtain optimal parameters.
Specifically, an intelligent task allocation model is constructed, and the calculation formula is as follows: wherein y is approval time, logarithmic change is adopted to ensure that the prediction time is positive, Is a constant term which is used to determine the degree of freedom,For the type of conference on the wayCorresponding matrix coefficients in the form of,For the degree of task urgencyThe corresponding coefficient is used to determine the coefficient,Task load for compliance approverCorresponding coefficients. Because the task emergency degree has influence on the task load of the qualified personnel, cross items are introducedIts corresponding coefficient is,Is normally distributed and is standard。
First construct coefficient vectorsConstructing a feature matrix after data processingWherein n is the sample data amount, and the target vector isAnd solving an intelligent task allocation model by adopting a least square method, wherein the calculation formula of a residual square sum function of the least square method is as follows: In which, in the process, As a sum of squares function of the residuals of the least squares method,As a result of the fact that the target vector,Is a feature matrix after the data processing,For the coefficient vector, in order to find the result thatMinimum coefficient vectorCalculation ofThe derivative of (2) is calculated by the following formula: In which, in the process, For partial differential operator, let the derivative be 0, get the normal equation: From which it can be solved to obtain And taking the fact that the data volume is large into consideration, adopting mumpy libraries to complete the construction of the intelligent task allocation model. The application adopts the logarithmic linear model to match with the cross item characteristic, thereby improving the prediction precision while ensuring the calculation efficiency. The least square method parameter estimation enhances the stability of the model and meets the real-time response requirement of the flow engine.
In some embodiments, the step of integrating the trained task allocation model into the Activiti process engine specifically includes configuring a custom listener in the BPMN file, and triggering the listener when the process task reaches the approval node.
Specifically, when the route conference flow reaches the approval node of the compliance person, the description takes the node_a node of the flow_a as an example;
the bpmn file of node_a of flow_a is newly added with the following configuration:
<extensionElements>
<activiti:taskListener
event="start"class="listener.MyTaskCreateListener"/>
</extensionElements>
When flow_a runs to node_a node, custom listener MYTASKCREATELISTENER is triggered. The application realizes non-invasive system integration through standard BPMN expansion, and a listener mechanism seamlessly adds intelligent allocation capability on the basis of keeping an active native function.
In some embodiments, the self-defined monitor specifically includes the steps of inheriting TASKLISTENER interfaces of an action engine, rewriting a notify method, acquiring context information of a current task in the notify method, calling an API of a task allocation model, inputting task type characteristics, task urgency characteristics and compliance approver task load characteristics, acquiring predicted approval time of candidate approvers, determining optimal approvers, dynamically setting approver variables through a runtime service interface of the action engine, and updating task allocation through RuntimeService interfaces.
Specifically, the flow_a custom flow listener class MYTASKCREATELISTENER, which implements the TASKLISTENER interface of the acitiviti flow engine, rewrites the notify method, and triggers the notify method in the MYTASKCREATELISTENER listener when the approval task runs to the node_a node.
The logic in the notify method is that a trained machine learning model API is called and the current road conference type is inputThe emergency degree of the task isTask load of compliance approverA predicted approval passing time of the plurality of candidates is obtained. And selecting an approver which can pass through the approval task and has the shortest approval time as a current node approver, calling the setVariable method of the Activiti flow engine RuntimeService class, and distributing the optimal approver in real time as the approver of the current approval task. After the allocation is completed, the approver receives the task prompt to be done so as to carry out approval processing. The application realizes real-time seamless switching of the approver based on the dynamic variable setting technology of the runtime interface, and remarkably improves the task reassignment efficiency under abnormal conditions.
On the other hand, the invention also provides an intelligent flow task distribution system based on the combination of Activiti and machine learning, which comprises a data collection module 10, a feature extraction module 20, a model construction module 30, an integration module 40, a distribution module 50, a training module 60 and an optimization module, wherein the data collection module 10 is used for collecting multidimensional road meeting flow data and cleaning the data, the feature extraction module 20 is used for processing and calculating the multidimensional road meeting flow data and extracting road meeting type features, task urgency features and compliance approver task load features, the model construction module 30 is used for constructing a task distribution model and training the task distribution model based on the extracted features, the model is used for predicting task approval time of different approvers, the integration module 40 is used for integrating the trained task distribution model into an Activiti flow engine, and when the flow task reaches an approval node, the model is called to calculate the predicted approval time of candidate approvers, the distribution module 50 is used for dynamically distributing the task according to the predicted approval time, and the compliance approver with the shortest predicted approval time is selected to execute the current approval task, and the task distribution model is used for periodically collecting new added approval data so as to optimize the predicted effect.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of the embodiments of the present invention. The storage medium includes a U disk, a removable hard disk, a read-only memory (ROM), a random access memory (random access memory, RAM), a magnetic disk, an optical disk, or other various media capable of storing program codes.
Those of ordinary skill in the art will appreciate that all or a portion of the processes in the methods of the embodiments described above may be implemented by hardware associated with a computer program that may be stored in a non-volatile computer readable storage medium, which when executed may include the processes of the embodiments described above, wherein any references to memory, storage, database, or other medium used in the embodiments provided by the present invention may include non-volatile and/or volatile memory, which may include read-only memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM) or flash memory, which may include Random Access Memory (RAM) or external cache memory, which may be available in a variety of forms, such as static RAM (RAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link SYNCHLINK DRAM (SLDRAM), memory bus (Rambus), direct RAM (rdbus), direct RAM bus (RDRAM), dynamic RAM bus (RDRAM), and the like, as an illustrative but not limiting.
The foregoing is only illustrative of the present invention and is not to be construed as limiting the scope of the invention, and all equivalent changes made by the description of the invention and the accompanying drawings, or direct or indirect application in the relevant art, are intended to be included within the scope of the invention.
Claims (6)
1. An intelligent flow task allocation method based on the combination of Activiti and machine learning is characterized by comprising the following steps:
collecting multidimensional route conference flow data and preprocessing;
the multi-dimensional road conference flow data comprises:
The method comprises the steps of a road conference type, task submission time, conference starting time, approver information, professional field matching degree, on-line state, approval passing time and the number of to-be-approved tasks of an approver;
the road conference type is used for determining task workload according to different types;
the task submitting time and the meeting starting time are used for calculating the task emergency degree;
the professional field matching degree is used for evaluating the suitability of the approver and the task type, the online state is used for judging the real-time availability of the approver, and the professional field matching degree and the online state are combined for judging whether the approver is in compliance;
The approval passing time and the number of tasks to be processed by the approver are used for calculating the load of the approver;
processing and calculating the multidimensional road conference flow data, and extracting road conference type characteristics, task emergency degree characteristics and compliance approval personnel task load characteristics;
constructing a task allocation model, and training the task allocation model based on the extracted characteristics, wherein the model is used for predicting task approval time of different approvers;
The step of constructing the task allocation model specifically comprises the following steps:
Constructing a linear regression model based on task type characteristics, task emergency degree characteristics and compliance approver task load characteristics;
Carrying out logarithmic transformation on the batch trial time to ensure that the prediction time is a positive value, and introducing cross item characteristics to optimize prediction;
solving model parameters through a least square method to obtain optimal parameters;
Integrating the trained task allocation model into an Activiti flow engine, and calling the model to calculate the predicted approval time of candidate approvers when the flow task reaches an approval node;
the step of integrating the trained task allocation model into the Activiti flow engine specifically comprises the following steps:
Configuring a custom monitor in a BPMN file, and triggering the monitor when a flow task reaches an approval node;
The step of the self-defining monitor specifically comprises the following steps:
inheriting TASKLISTENER interfaces of the Activiti engine and rewriting a notify method;
Acquiring context information of a current task in a notify method;
calling an API of a task allocation model, inputting task type characteristics, task emergency degree characteristics and task load characteristics of compliance approvers, acquiring predicted approval time of candidate approvers, and determining optimal approvers;
dynamically setting an approver variable through a runtime service interface of an action engine, and updating task allocation through a RuntimeService interface;
According to the predicted approval time, dynamically distributing tasks, and selecting a compliance approver with the shortest predicted approval time to execute the current approval task;
And periodically collecting newly added approval data, and retraining the task allocation model to iteratively optimize the prediction effect.
2. The intelligent process task allocation method according to claim 1, wherein the preprocessing step specifically includes:
recording data sources and processing logic in the process of integrating multidimensional route conference flow data, performing desensitization processing on sensitive information, retaining characteristics related to business, and eliminating data missing values;
And converting the type data into unique binary vectors by adopting single-heat coding, and normalizing the numerical data to enable the numerical data to fall into the [0,1] interval.
3. The intelligent process task allocation method according to claim 1, wherein the step of determining whether the approver is compliant by combining the professional field matching degree and the online status specifically comprises:
judging whether the professional field of the approver is matched with the type of the road conference, if the professional field of the approver contains the type of the road conference, marking the professional field matching as1, otherwise marking as 0;
judging whether an approver is online, if so, marking the approver as 1, otherwise, marking the approver as 0;
and if the type of the professional field is matched with the conference type and the approver is online, judging that the approver is compliant.
4. The intelligent process task allocation method according to claim 1, wherein the step of calculating the task urgency feature specifically includes:
Acquiring a time difference between task submitting time and meeting starting time;
converting the time difference into an emergency degree value by adopting a time decay function, wherein the smaller the time difference is, the higher the emergency degree value is;
Random noise conforming to normal distribution is introduced in the calculation process, and uncertainty in an actual approval scene is simulated.
5. The intelligent process task allocation method according to claim 4, wherein the means for calculating the compliance approver task load feature comprises:
counting the number of tasks to be approved of current approvers;
Calculating the emergency degree of each task to be examined, and sequencing the tasks from high to low according to the emergency degree;
and calculating a normalized task load value according to the position of the current task in the sequencing and the historical maximum task number.
6. An intelligent process task distribution system based on the combination of Activiti and machine learning, which is characterized by comprising:
the data collection module is used for collecting multidimensional route meeting flow data and cleaning the data;
the multi-dimensional road conference flow data comprises:
The method comprises the steps of a road conference type, task submission time, conference starting time, approver information, professional field matching degree, on-line state, approval passing time and the number of to-be-approved tasks of an approver;
the road conference type is used for determining task workload according to different types;
the task submitting time and the meeting starting time are used for calculating the task emergency degree;
the professional field matching degree is used for evaluating the suitability of the approver and the task type, the online state is used for judging the real-time availability of the approver, and the professional field matching degree and the online state are combined for judging whether the approver is in compliance;
The approval passing time and the number of tasks to be processed by the approver are used for calculating the load of the approver;
The feature extraction module is used for processing and calculating the multidimensional route conference flow data and extracting the route conference type features, the task emergency degree features and the task load features of compliance approval personnel;
The model construction module is used for constructing a task allocation model and training the task allocation model based on the extracted characteristics, wherein the model is used for predicting task approval time of different approvers;
The step of constructing the task allocation model specifically comprises the following steps:
Constructing a linear regression model based on task type characteristics, task emergency degree characteristics and compliance approver task load characteristics;
Carrying out logarithmic transformation on the batch trial time to ensure that the prediction time is a positive value, and introducing cross item characteristics to optimize prediction;
solving model parameters through a least square method to obtain optimal parameters;
the integration module is used for integrating the trained task allocation model into an Activiti flow engine, and calling the model to calculate the predicted approval time of candidate approval personnel when the flow task reaches the approval node;
the step of integrating the trained task allocation model into the Activiti flow engine specifically comprises the following steps:
Configuring a custom monitor in a BPMN file, and triggering the monitor when a flow task reaches an approval node;
The step of the self-defining monitor specifically comprises the following steps:
inheriting TASKLISTENER interfaces of the Activiti engine and rewriting a notify method;
Acquiring context information of a current task in a notify method;
calling an API of a task allocation model, inputting task type characteristics, task emergency degree characteristics and task load characteristics of compliance approvers, acquiring predicted approval time of candidate approvers, and determining optimal approvers;
dynamically setting an approver variable through a runtime service interface of an action engine, and updating task allocation through a RuntimeService interface;
The distribution module is used for dynamically distributing tasks according to the predicted approval time, and selecting a compliance approver with the shortest predicted approval time to execute the current approval task;
And the optimizing module is used for periodically collecting the newly added approval data and retraining the task allocation model to iteratively optimize the prediction effect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510874800.4A CN120806449B (en) | 2025-06-27 | 2025-06-27 | A Smart Process Task Allocation Method and System Based on Activiti and Machine Learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510874800.4A CN120806449B (en) | 2025-06-27 | 2025-06-27 | A Smart Process Task Allocation Method and System Based on Activiti and Machine Learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN120806449A CN120806449A (en) | 2025-10-17 |
| CN120806449B true CN120806449B (en) | 2026-03-24 |
Family
ID=97323224
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202510874800.4A Active CN120806449B (en) | 2025-06-27 | 2025-06-27 | A Smart Process Task Allocation Method and System Based on Activiti and Machine Learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN120806449B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN121279959A (en) * | 2025-12-08 | 2026-01-06 | 上海上湖信息技术有限公司 | Approval process processing method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112598366A (en) * | 2020-11-27 | 2021-04-02 | 北京理工大学 | Automatic planning method and device for asset management approval process |
| CN117076096A (en) * | 2023-10-17 | 2023-11-17 | 腾讯科技（深圳）有限公司 | Task flow execution method and device, computer readable medium and electronic equipment |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117151608A (en) * | 2023-06-19 | 2023-12-01 | 杭州电力设备制造有限公司 | Construction method of mobile approval micro-front-end architecture based on business flow |
| CN118246880A (en) * | 2024-04-02 | 2024-06-25 | 中国水利水电科学研究院 | A design method for machine recognition approval process engine based on the combination of Activiti and LLaMA |
| CN118733859B (en) * | 2024-08-30 | 2024-11-22 | 深圳市实信达科技开发有限公司 | File information storage integrated management system and method based on artificial intelligence |
| CN119379350A (en) * | 2024-10-16 | 2025-01-28 | 河南中原消费金融股份有限公司 | Method, system and device for simulating customer approval results based on historical data learning |
| CN119398698A (en) * | 2024-10-23 | 2025-02-07 | 海通证券股份有限公司 | Task processing method, task processing device, electronic device, storage medium and program product |
| CN119415180A (en) * | 2024-11-01 | 2025-02-11 | 苏州元脑智能科技有限公司 | A method, device, equipment and service system for starting a workflow instance |
| CN120047255A (en) * | 2024-12-30 | 2025-05-27 | 北京行旅国际旅游科技有限公司 | Reimbursement data management system and method based on AI artificial intelligence |
| CN119784124A (en) * | 2024-12-30 | 2025-04-08 | 中科曙光国际信息产业有限公司 | Business process processing method, device, electronic device, storage medium and program product |
| DE202025102095U1 (en) * | 2025-04-16 | 2025-05-26 | Deepesh Vinodkumar Semlani | Voice-enabled approval workflow system integrated with an Oracle ERP |
-
2025
- 2025-06-27 CN CN202510874800.4A patent/CN120806449B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112598366A (en) * | 2020-11-27 | 2021-04-02 | 北京理工大学 | Automatic planning method and device for asset management approval process |
| CN117076096A (en) * | 2023-10-17 | 2023-11-17 | 腾讯科技（深圳）有限公司 | Task flow execution method and device, computer readable medium and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN120806449A (en) | 2025-10-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11650968B2 (en) | Systems and methods for predictive early stopping in neural network training | |
| Pfeiffer et al. | Manufacturing lead time estimation with the combination of simulation and statistical learning methods | |
| CN110799997B (en) | Industrial data service, data modeling, and data application platform | |
| CN112905340B (en) | System resource allocation method, device and equipment | |
| CN112508723B (en) | Financial risk prediction method, device and electronic equipment based on automatic optimal modeling | |
| US11816687B2 (en) | Personalized approach to modeling users of a system and/or service | |
| CN120806449B (en) | A Smart Process Task Allocation Method and System Based on Activiti and Machine Learning | |
| CN113191540B (en) | A method and device for constructing industrial chain manufacturing resources | |
| CN119006165B (en) | Data asset assessment method and system based on big data | |
| CN119988157B (en) | A data collection method and system based on big data of intelligent operation and maintenance platform | |
| CN114202179A (en) | Target enterprise identification method and device | |
| CN119338053B (en) | Platform management method and system based on low code development | |
| WO2017071369A1 (en) | Method and device for predicting user unsubscription | |
| CN115204648A (en) | Method and device for obtaining power engineering quality index based on mathematical model | |
| CN118115098A (en) | Big data analysis and processing system based on deep learning | |
| CN116883065A (en) | Merchant risk prediction method and device | |
| CN120196680B (en) | Industrial data asset distributed storage method, medium and system | |
| Hu et al. | Research on supply chain demand forecasting and dynamic adjustment model based on big data and artificial intelligence | |
| CN119963165A (en) | Multi-source sensing and joint charging infrastructure operation and maintenance management system | |
| CN118586782A (en) | A mathematical and statistical analysis method based on quality management decision-making system | |
| CN118967192A (en) | Stock copy trading customer churn analysis method, system and computer-readable storage medium | |
| CN118034905A (en) | Financial task execution method, device, equipment and storage medium | |
| CN113850609A (en) | A client management system, method, computer device and storage medium | |
| CN114548538A (en) | Task amount prediction method and device, electronic equipment and storage medium | |
| CN113850454A (en) | Method, device, equipment and medium for manufacturing parts based on industrial neural network |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |