Enterprise-level employee digital duplication system based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence, and is used for solving the problems that an existing employee digital duplication system often neglects effective collection and analysis of task performance data of employees in a unit period and lacks a screening mechanism based on historical behaviors and task performance; the problems of high response delay, task omission and overtime completion of partial clones in task execution due to employees of a construction source are solved; the enterprise-level employee digital duplication system based on artificial intelligence comprises a server, and the server is further connected with a data acquisition module, a difficulty judgment module, a response analysis module, a reliability analysis module, a planning module and an effect verification module. Quantitative analysis is carried out through the process complexity, the employee response timeliness and the task reliability degree, and the intelligence and the accuracy of the digital duplication system are improved in combination with a judgment rule and an interval classification mechanism.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to an enterprise-level employee digital clone system based on artificial intelligence. BACKGROUND

[0002] The integration of artificial intelligence technology in organization management and enterprise operation has promoted the enterprise intelligence from basic information management to deep intelligent decision-making driven by behavior modeling and data analysis. Under this trend, many enterprises have begun to explore the construction of employee digital clones, that is, to replicate or simulate the work behavior and processing mode of real employees through artificial intelligence systems, so as to realize the proxy execution, process promotion and communication interaction of tasks. Such digital clones are usually embedded in intelligent office systems, collaborative work platforms or business process automation modules to undertake part of the daily transactional and rule-based tasks.

[0003] However, the existing employee digital clone system generally ignores the effective collection and analysis of the task performance data of employees in the unit cycle, lacks a screening mechanism based on historical behavior and task performance, and usually defaults that all employees are suitable for building clones, regardless of whether they have high task carrying capacity or reliability level in work. This non-screening construction strategy directly leads to the unevenness of the clone agent capability, and makes some clones continue the behavior defects such as high response delay, task omission or overtime completion in task execution due to the employees themselves, which ultimately leads to the continuation of these behavior defects in the actual application of the constructed digital clone, and further causes agent failure or work blockage. At the same time, the current system lacks quantitative modeling of task complexity, task adaptability and process response capability related dimensions, which makes it difficult to support effective candidate selection before building clones, which makes the deployment of digital clone system more like a surface form of automation, rather than a precise replication based on employee ability identification. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides an enterprise-level employee digital clone system based on artificial intelligence, which solves the problems in the background art.

[0005] To achieve the above purpose, the present application is realized by the following technical scheme: an enterprise-level employee digital clone system based on artificial intelligence, comprising a server, wherein the server is further connected with a data acquisition module, a difficulty determination module, a response analysis module, a reliability analysis module, a planning module and an effect verification module.

[0006] The data collection module is used to collect the task characteristic data information and the task completion state data information of the employees in the historical unit period through the plug-in embedded in the daily office tool and in combination with the local log collector of the enterprise end, and send the collected related data information to the difficulty determination module, the response analysis module and the reliability analysis module respectively;

[0007] The difficulty determination module is used to determine and analyze the difficulty level of the corresponding employees to complete the corresponding process task in the historical unit period according to the received task characteristic data information, obtain the task difficulty level of the corresponding employees to complete the corresponding process task, and match the difficulty value of the corresponding process task;

[0008] The response analysis module is used to analyze the comprehensive response pressure degree of the corresponding employees to complete each process task in the historical unit period according to the received task characteristic data information, and thereby build the bearing capacity label employee group;

[0009] The reliability analysis module is used to analyze the reliability level of the corresponding employees to complete each process task in the historical unit period according to the received task completion state data information, and thereby build the reliability label employee group;

[0010] The planning module is used to plan and analyze the digital twin determination situation of each employee in the historical unit period, and thereby obtain the digital twin selected employee list, the digital twin candidate employee list and the digital twin unselected employee list;

[0011] The effect verification module constructs the relevant category digital twin based on the digital twin selected employee list, compares and analyzes the performance score situation before and after the application of the digital twin, judges whether the application of the digital twin of the corresponding personnel is effective, and generates the feedback adjustment instruction of the corresponding level and executes it.

[0012] Preferably, the difficulty level of the corresponding employees to complete the corresponding process task in the historical unit period is determined and analyzed, and the specific analysis process includes:

[0013] The task characteristic data information collected in the historical unit period is subjected to feature recognition, the extracted prescribed approval times, prescribed pending time limit and prescribed uploaded file quantity of the corresponding employees to complete each process task are subjected to normalization analysis, the task complexity coefficient of the corresponding employees to complete the corresponding process task is determined, and specifically: ; In the formula, represents the task complexity coefficient of the jth employee to complete the ith process task, 、 and respectively represent the specified number of approvals, the specified time limit for pending, and the specified number of uploaded files for the jth employee to complete the ith process task, wherein a1, a2, and a3 represent the weight values of the specified number of approvals, the specified time limit for pending, and the specified number of uploaded files, respectively, and the specific values are set by a person skilled in the art;

[0014] The task complexity coefficient of the corresponding employee completing the corresponding process task is input into the task complexity interval matching table, and the task difficulty level of the corresponding employee completing the corresponding process task is output, and the difficulty value is matched, and the specific process is as follows:

[0015] If the task complexity coefficient of the corresponding process task is in the first task complexity interval in the task complexity interval matching table, the corresponding process task is matched as a first task difficulty, if the task complexity coefficient of the corresponding process task is in the second task complexity interval in the task complexity interval matching table, the corresponding process task is matched as a second task difficulty, and if the task complexity coefficient of the corresponding process task is in the third task complexity interval in the task complexity interval matching table, the corresponding process task is matched as a third task difficulty.

[0016] The difficulty value of the corresponding process task is matched according to the task difficulty level of the corresponding process task, wherein the first task difficulty is matched with a difficulty value of 3N, the second task difficulty is matched with a difficulty value of 2N, and the third task difficulty is matched with a difficulty value of 1N, and 3N>2N>1N, N represents a positive integer.

[0017] Preferably, the comprehensive response pressure degree of the corresponding employee completing each process task in the historical unit period is analyzed, and the specific analysis process includes:

[0018] The task feature data information collected in the historical unit period is subjected to feature recognition, and the task issuance time and the task first processing time of the corresponding employee completing each process task are subjected to difference processing, respectively, to obtain the task response time length of the corresponding employee completing each process task.

[0019] The task response time length of the corresponding employee completing each process task is associated with the difficulty value of the corresponding process task, and after dimensionless processing, the comprehensive response pressure degree of the corresponding employee completing each process task in the historical unit period is analyzed, and the comprehensive response pressure value of the corresponding employee in the historical unit period is determined, which is specifically: ; wherein, represents the comprehensive response pressure value of the jth employee in the historical unit period, and respectively represent the task response time length and the difficulty value of the jth employee completing the ith process task, wherein i=1, 2, 3,..., n, and n represents the total number of tasks.

[0020] According to the numerical size of the comprehensive response pressure value of each employee in the historical unit period, the corresponding employees are arranged in descending order from large to small, and a response pressure personnel sequence is obtained. The top 50% of employees in the response pressure personnel sequence are selected and marked as high task carrying capacity employee tags, and the last 50% of employees in the response pressure personnel sequence are selected and marked as low task carrying capacity employee tags.

[0021] According to the tag situation of the task carrying capacity level of the corresponding employee, a carrying capacity tag employee group is constructed.

[0022] Preferably, the reliability level of the corresponding employee in completing each process task in the historical unit period is analyzed, and the specific analysis process includes:

[0023] The feature recognition is performed on the task completion state data information collected in the historical unit period, and the number of on-time task completion and the number of overtime task completion of each employee in the historical unit period are extracted.

[0024] By comparing the number of on-time task completion and the number of overtime task completion of each employee in the historical unit period, a reliability tag employee group is constructed, and the specific construction process is:

[0025] If the number of on-time task completion of the corresponding employee exceeds the number of overtime task completion, it indicates that the corresponding employee is in a controllable state of task completion in the historical unit period, and the corresponding employee is marked as a high-reliability employee tag.

[0026] If the number of on-time task completion of the corresponding employee does not exceed the number of overtime task completion, it indicates that the corresponding employee is in an uncontrollable state of task completion in the historical unit period, and the corresponding employee is marked as a low-reliability employee tag.

[0027] According to the tag situation of the reliability level of the corresponding employee, a reliability tag employee group is constructed.

[0028] Preferably, the digital twin judgment of each employee in the historical unit period is planned and analyzed, and the specific analysis process includes:

[0029] The employee tags of the carrying capacity tag employee group and the reliability tag employee group are extracted, and the tag categories of each employee are counted.

[0030] If the corresponding employee is marked as a high task carrying capacity employee tag and a high reliability employee tag at the same time, a first level digital twin degree signal is generated, and the corresponding employee is listed in the digital twin selected employee list according to the generated first level digital twin degree signal.

[0031] If the corresponding employee is marked as both a high task load employee tag and a low reliability employee tag, or as both a low task load employee tag and a high reliability employee tag, a secondary digital clone degree signal is generated, and the corresponding employee is listed in the digital clone candidate employee list according to the generated secondary digital clone degree signal;

[0032] If the corresponding employee is marked as both a low task load employee tag and a low reliability employee tag, a tertiary digital clone degree signal is generated, and the corresponding employee is listed in the digital clone elimination employee list according to the generated tertiary digital clone degree signal.

[0033] Preferably, based on the digital clone selected employee list, relevant category digital clones are constructed, and the performance score situation before and after the application of the digital clone is analyzed, and the specific analysis process includes:

[0034] Based on the digital clone selected employee list, and in combination with the interaction of the employee in the platform office, relevant category digital clones of each person in the digital clone selected employee list are constructed, wherein the relevant category digital clones include request type digital clones, reply type digital clones, instruction type digital clones, summary type digital clones, and confirmation type digital clones.

[0035] The request type digital clone is used to represent the intention of the employee to assist colleagues, obtain information, and apply for resources, the reply type digital clone is used to respond to questions, feedback request results, and explain processing progress, the instruction type digital clone is used to assign tasks, issue operations, and promote process progress, the summary type digital clone is used to summarize stage achievements, sort out meeting conclusions, and report work status, and the confirmation type digital clone is used to verify information accuracy, verify task execution, and respond to transactional checks.

[0036] According to the performance score situation before and after the application of the relevant category digital clones, the performance score values before and after the application of the corresponding category digital clones of each person in the digital clone selected employee list are collected, and the performance score total values before and after the application of each person in the digital clone selected employee list are respectively calculated in combination with a statistical summation algorithm.

[0037] The performance score total values before and after the application of each person in the digital clone selected employee list are compared and analyzed to obtain the performance difference values of each person in the digital clone selected employee list before and after the application.

[0038] Preferably, the performance difference values of each person in the digital clone selected employee list before and after the application are respectively compared and analyzed with a preset difference threshold value to determine whether the digital clone application of the corresponding person is effective, to generate a feedback adjustment instruction of the corresponding level and execute it, and the specific process includes:

[0039] If the performance difference of the corresponding personnel in the digital twin selected employee list before and after the application exceeds the difference threshold, it indicates that the digital twin application of the corresponding personnel is invalid, and a first-level feedback adjustment instruction is generated, including: deleting the corresponding personnel from the digital twin selected employee list, and selecting personnel from the digital twin candidate employee list as a supplement;

[0040] If the performance difference of the corresponding personnel in the digital twin selected employee list before and after the application does not exceed the difference threshold, it indicates that the digital twin application of the corresponding personnel is effective, and a second-level feedback adjustment instruction is generated, including: continue the digital twin application operation of the corresponding personnel.

[0041] The present application provides an enterprise-level employee digital twin system based on artificial intelligence, which has the following beneficial effects:

[0042] (1) The present application introduces data collection, difficulty determination, response analysis, reliability modeling and effect verification related modules, breaks through the static split setting idea of the existing system in the overall structure design, realizes the precise split construction mechanism based on historical task performance and ability identification in a true sense; The system quantifies the dimensions of process complexity, employee response time, and task reliability by collecting and analyzing task behavior data in a unit cycle, and incorporates them into the digital twin adaptation evaluation index system, solving the generalization failure problem caused by the default of the existing system that all employees can build a split; With the help of judgment rules and interval classification mechanism, the system can dynamically identify high load and high reliability employees and preferentially build a split model to ensure stable, controllable and efficient proxy behavior in subsequent applications; At the same time, by introducing a performance difference comparison mechanism, the deployed split is continuously verified, and once the effect is not as expected, the system will automatically generate adjustment instructions and complete personnel replacement, forming a closed-loop management structure; This strategy significantly improves the intelligence, accuracy and continuous evolution ability of the digital twin system, effectively promoting the enterprise from surface automation to deep intelligent operation.

[0043] (2) By taking the task response stress value and the task completion reliability value of the staff as double indexes, a high bearing capacity label and a high reliability label double-dimensional label system is established, so that the precise selection and grading identification mechanism is realized. The system no longer depends on subjective evaluation or management experience, but analyzes the task data in the real historical period, including the task processing time and the task completion state, extracts the key behavior characteristics through the logical judgment model, and constructs the systematic staff ability portrait. According to the label combination, the staff is automatically divided into the selected, candidate or eliminated list, so as to ensure that only the staff with real behavior performance support is included in the construction range of the digital avatar. This mechanism effectively avoids the inclusion of execution procrastinators, frequent task failure and inefficient staff into the system, and guarantees the quality consistency and high execution force of the digital avatar from the source. Compared with the traditional system, the scheme significantly reduces the problems of agent errors, process blockage and transaction interruption caused by defects in the construction source of the digital avatar, and improves the running stability of the whole system.

[0044] (3) In most current digital avatar systems, there is a lack of systematic tracking and quantitative feedback mechanism for the actual agent effect of the avatar, which is often in an uncontrollable state for a long time after the avatar is online, and it is difficult to find and correct problems in time. The effect verification module proposed in the present application solves the core problem of avatar efficiency evaluation by designing a performance difference comparison model before and after application. This mechanism relies on the performance score collection mechanism in the unit period to quantify the execution performance of each type of digital avatar in the staff's original work and avatar agent state, calculates the performance total value difference, and compares it with the preset threshold. Once it is found that the performance does not increase but decreases or does not meet the standard after the avatar is applied, the system will automatically generate a first-level feedback adjustment instruction to start the replacement mechanism of the candidate staff, so as to ensure that the system is continuously maintained in the optimal operating state. This dynamic verification and self-adjustment strategy not only builds a closed-loop control framework for the avatar system, but also enables the avatar system to have self-optimization ability for the first time. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 The figure is a block diagram of an enterprise-level staff digital avatar system based on artificial intelligence. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0047] EMBODIMENT

[0048] Please refer to Figure 1The application provides an enterprise-level employee digital clone system based on artificial intelligence, comprising a server, wherein the server is further connected with a data acquisition module, a difficulty determination module, a response analysis module, a reliability analysis module, a planning module and an effect verification module;

[0049] The data acquisition module is used for acquiring task characteristic data information and task completion state data information of employees in a historical unit period through a plug-in embedded in daily office tools in combination with a local log collector of an enterprise end, and sending the acquired related data information to the difficulty determination module, the response analysis module and the reliability analysis module respectively.

[0050] Specifically, the difficulty determination module is used for determining and analyzing the difficulty level of corresponding employees in completing corresponding process tasks in a historical unit period according to the received task characteristic data information, obtaining the task difficulty level of corresponding employees in completing corresponding process tasks, and matching the difficulty value of corresponding process tasks, and the specific analysis process comprises:

[0051] The task characteristic data information collected in a historical unit period is subjected to feature recognition, and the extracted prescribed approval times, prescribed pending time limit and prescribed file upload quantity of corresponding employees in completing each process task are subjected to normalization analysis to determine the task complexity coefficient of corresponding employees in completing corresponding process tasks, and the specific process is as follows: ; wherein, represents the task complexity coefficient of the jth employee in completing the ith process task, 、 and respectively represent the prescribed approval times, prescribed pending time limit and prescribed file upload quantity of the jth employee in completing the ith process task, wherein a1, a2 and a3 respectively represent the weight values of the prescribed approval times, prescribed pending time limit and prescribed file upload quantity, and the specific numerical values are set by a person skilled in the art.

[0052] It should be noted that the prescribed approval times, prescribed pending time limit and prescribed file upload quantity refer to the standardized operation requirements set by an enterprise in a process task execution specification, and are used to measure the processing complexity of a process task at the system level; specifically, the prescribed approval times refer to the number of standard approval nodes required in the normal compliance completion process of a process task; the prescribed pending time limit refers to the longest period required by the system between the reception of a task and the processing thereof; and the prescribed file upload quantity refers to the number of standard documents and attachments required to be submitted in the execution process of the task; these data are obtained by analyzing and extracting the metadata structure of a process to which the task belongs from the process template and task configuration document defined in an enterprise process engine (an OA system, an ERP system or a BPM platform);

[0053] The task complexity coefficient of the corresponding employee completing the corresponding process task is input into the task complexity interval matching table, and the task difficulty level of the corresponding employee completing the corresponding process task is output, and the difficulty value is matched, and the specific process is as follows:

[0054] If the task complexity coefficient of the corresponding process task is in the first task complexity interval in the task complexity interval matching table, the corresponding process task is matched as a first task difficulty, if the task complexity coefficient of the corresponding process task is in the second task complexity interval in the task complexity interval matching table, the corresponding process task is matched as a second task difficulty, if the task complexity coefficient of the corresponding process task is in the third task complexity interval in the task complexity interval matching table, the corresponding process task is matched as a third task difficulty.

[0055] According to the task difficulty level of the corresponding process task, the difficulty value of the corresponding process task is matched, wherein the first task difficulty is matched with the difficulty value 3N, the second task difficulty is matched with the difficulty value 2N, and the third task difficulty is matched with the difficulty value 1N, and 3N>2N>1N, N represents a positive integer.

[0056] Specifically, the response analysis module is configured to analyze the comprehensive response pressure degree of the corresponding employee completing each process task in the historical unit period according to the received task characteristic data information, and to build a bearing capacity label employee group accordingly. The specific analysis process includes:

[0057] The task characteristic data information collected in the historical unit period is subjected to feature recognition, and the extracted task issuance time and task first processing time of the corresponding employee completing each process task are subjected to difference processing, respectively, to obtain the task response time length of the corresponding employee completing each process task.

[0058] The task response time length of the corresponding employee completing each process task is associated with the difficulty value of the corresponding process task, and after non-dimensional processing, the comprehensive response pressure degree of the corresponding employee completing each process task in the historical unit period is analyzed, and the comprehensive response pressure value of the corresponding employee in the historical unit period is determined, which is specifically: ; In the formula, represents the comprehensive response pressure value of the jth employee in the historical unit period, and respectively represent the task response time length and the difficulty value of the jth employee completing the ith process task, wherein i=1, 2, 3,..., n, and n represents the total number of tasks.

[0059] According to the numerical size of the comprehensive response pressure value of each employee in the historical unit period, the corresponding employee is arranged in descending order from large to small, and a response pressure personnel sequence is obtained. The top 50% of employees in the response pressure personnel sequence are selected and marked as high task carrying capacity employee tags. The last 50% of employees in the response pressure personnel sequence are selected and marked as low task carrying capacity employee tags.

[0060] According to the tag situation of the task carrying capacity level of the corresponding employee, a carrying capacity tag employee group is constructed.

[0061] Specifically, the reliability analysis module is configured to analyze the reliability level of the corresponding employee in completing each process task in the historical unit period according to the received task completion state data information, and to construct a reliability tag employee group accordingly. The specific analysis process includes:

[0062] The task completion state data information collected in the historical unit period is subjected to feature recognition, and the number of tasks completed on time and the number of tasks completed overtime of each employee in the historical unit period are extracted.

[0063] By comparing the number of tasks completed on time and the number of tasks completed overtime of each employee in the historical unit period, a reliability tag employee group is constructed. The specific construction process is as follows:

[0064] If the number of tasks completed on time of the corresponding employee exceeds the number of tasks completed overtime, it indicates that the corresponding employee is in a controllable state of task completion in the historical unit period, and the corresponding employee is marked as a high-reliability employee tag.

[0065] If the number of tasks completed on time of the corresponding employee does not exceed the number of tasks completed overtime, it indicates that the corresponding employee is in an uncontrollable state of task completion in the historical unit period, and the corresponding employee is marked as a low-reliability employee tag.

[0066] According to the tag situation of the reliability level of the corresponding employee, a reliability tag employee group is constructed.

[0067] Specifically, the planning module is configured to analyze the digital twin judgment of each employee in the historical unit period, and to obtain a digital twin selected employee list, a digital twin candidate employee list, and a digital twin eliminated employee list accordingly. The specific analysis process includes:

[0068] The employee tags of the carrying capacity tag employee group and the reliability tag employee group are extracted, and the tag categories of each employee are counted.

[0069] If the corresponding employee is marked as a high task carrying capacity employee tag and a high reliability employee tag at the same time, a first level digital twin degree signal is generated, and the corresponding employee is included in the digital twin selected employee list according to the generated first level digital twin degree signal.

[0070] If the corresponding employee is marked as a high task-bearing capacity employee tag and a low reliability employee tag at the same time, or as a low task-bearing capacity employee tag and a high reliability employee tag at the same time, a secondary digital shilling degree signal is generated, and the corresponding employee is listed in the digital shilling candidate employee list according to the generated secondary digital shilling degree signal;

[0071] If the corresponding employee is marked as a low task-bearing capacity employee tag and a low reliability employee tag at the same time, a tertiary digital shilling degree signal is generated, and the corresponding employee is listed in the digital shilling elimination employee list according to the generated tertiary digital shilling degree signal.

[0072] Specifically, the effect verification module constructs relevant category digital shillings based on the digital shilling selected employee list, compares and analyzes the performance scores before and after the application of the digital shilling, judges whether the application of the digital shilling of the corresponding personnel is effective, generates feedback adjustment instructions of the corresponding level and executes them. The specific process includes:

[0073] Based on the digital shilling selected employee list, and combined with the interaction of the employee in the platform office, the relevant category digital shillings of each personnel in the digital shilling selected employee list are constructed, wherein the relevant category digital shillings include request type digital shilling, reply type digital shilling, instruction type digital shilling, summary type digital shilling and confirmation type digital shilling.

[0074] The request type digital shilling is used to represent the intention of the employee to assist colleagues, obtain information and apply for resources, the reply type digital shilling is used to respond to questions, feedback request results and explain processing progress, the instruction type digital shilling is used to arrange tasks, issue operations and promote process progress, the summary type digital shilling is used to summarize stage achievements, sort out meeting conclusions and report work status, and the confirmation type digital shilling is used to verify information accuracy, verify task execution and respond to transactional checks.

[0075] It should be noted that in the process of constructing the relevant category digital shillings of each personnel in the digital shilling selected employee list, first, the daily office interaction data of the employee identified as a high bearing capacity and a high reliability tag within a unit period is extracted; including but not limited to email content, instant messaging records, collaborative document editing tracks, task instruction operation records and meeting minutes generation behavior data; the above data is subjected to semantic recognition and intention classification through natural language processing (NLP) technology in artificial intelligence, and the communication types exhibited by the employee in different scenarios are automatically classified into request type, reply type, instruction type, summary type and confirmation type; on this basis, a personalized digital shilling behavior model is further trained using a behavior modeling algorithm (a behavior sequence modeling network based on the Transformer structure);

[0076] According to the performance score situation of the relevant category digital clone application before and after the process, the performance score value before application and the performance score value after application of each person in the digital clone selected employee list corresponding to the category digital clone are collected respectively, and the statistical summation algorithm is combined to respectively count the performance score total value before application and the performance score total value after application of each person in the digital clone selected employee list;

[0077] The performance difference value before and after application of each person in the digital clone selected employee list is compared and analyzed, and the performance difference value of each person in the digital clone selected employee list is obtained.

[0078] The performance difference value before and after application of each person in the digital clone selected employee list is compared and analyzed, and the performance difference value of each person in the digital clone selected employee list is obtained.

[0079] If the performance difference value before and after application of the corresponding person in the digital clone selected employee list exceeds the difference threshold value, it means that the digital clone application of the corresponding person is invalid, and a first-level feedback adjustment instruction is generated, which includes: deleting the corresponding person from the digital clone selected employee list, and selecting a person from the digital clone candidate employee list as a supplement.

[0080] If the performance difference value before and after application of the corresponding person in the digital clone selected employee list does not exceed the difference threshold value, it means that the digital clone application of the corresponding person is effective, and a second-level feedback adjustment instruction is generated, which includes: continuing the digital clone application operation of the corresponding person.

[0081] In use, the plug-in embedded in the daily office tool is used in combination with the local log collector of the enterprise end to collect the task feature data information and the task completion state data information of the employees in the historical unit period, and the collected related data information is sent to the difficulty determination module, the response analysis module and the reliability analysis module, so that the multi-dimensional data accurate collection of the employee work behavior can be realized, and the real and complete data support for the subsequent analysis module is ensured; through the cooperative work of the plug-in and the log collector, the real-time and scene adaptability of data acquisition are significantly improved, which provides a basic guarantee for the intelligent judgment of the system.

[0082] According to the received task feature data information, the difficulty level of the corresponding employee completing the corresponding process task in the historical unit period is determined and analyzed, the task difficulty level of the corresponding employee completing the corresponding process task is obtained, and the difficulty value of the corresponding process task is matched, so that the subjective bias of human evaluation is effectively avoided, the accurate difficulty scale for subsequent task matching and clone ability docking is laid, and the scientific matching of task difficulty and employee ability is realized.

[0083] According to the received task characteristic data information, the comprehensive response pressure degree of corresponding employees completing each process task in the historical unit period is analyzed, and a bearing capacity label employee group is constructed accordingly. Through the division of the bearing capacity label, it is helpful to identify the key personnel who are efficient and quick in response, and to provide personnel guarantee for high-quality digital avatar construction;

[0084] According to the received task completion state data information, the reliability level of corresponding employees completing each process task in the historical unit period is analyzed, and a reliability label employee group is constructed accordingly. Through feature comparison of task completion state data, the controllability and stability of employees executing tasks are automatically identified, effectively excluding personnel with frequent task delays from entering the avatar system, and significantly improving the overall credibility of agent behavior;

[0085] The digital avatar determination of each employee in the historical unit period is analyzed, and accordingly the digital avatar selected employee list, the digital avatar candidate employee list and the digital avatar eliminated employee list are obtained, realizing the scientific classification and stratification of candidate employees, and significantly reducing the resource waste and misjudgment risk in the process of digital avatar construction;

[0086] Based on the digital avatar selected employee list, the relevant category digital avatar is constructed, and through comparative analysis of the performance score before and after the application of the digital avatar, it is determined whether the application of the digital avatar of the corresponding personnel is effective, to generate the feedback adjustment instruction of the corresponding level and execute it, realizing the closed-loop management of the avatar construction, and ensuring the quality stability and continuous optimization in the long-term operation of the system.

[0087] Although embodiments of the present application have been shown and described, it is to be understood that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An enterprise-level digital avatar system for employees based on artificial intelligence, characterized in that: It includes a server, which is also connected to a data acquisition module, a difficulty assessment module, a response analysis module, a reliability analysis module, a planning module, and an effect verification module; The data acquisition module is used to collect task characteristic data and task completion status data of employees within a historical unit period through a plugin embedded in daily office tools and in combination with the local log collector on the enterprise side, and sends the collected relevant data information to the difficulty judgment module, response analysis module and reliability analysis module respectively. The difficulty determination module is used to determine and analyze the difficulty level of the corresponding employee in completing the corresponding process task within a historical unit period based on the received task feature data information, obtain the task difficulty level of the corresponding employee in completing the corresponding process task, and match the difficulty value of the corresponding process task. The response analysis module is used to analyze the comprehensive response pressure of corresponding employees in completing each process task within a historical unit period based on the received task characteristic data information, and to construct a group of employees with carrying capacity labels accordingly. The reliability analysis module is used to analyze the reliability level of corresponding employees in completing each process task within a historical unit period based on the received task completion status data information, and to construct a group of employees with reliability labels accordingly. The planning module is used to perform planning analysis on the digital clone determination of each employee within a historical unit period, thereby obtaining the list of employees selected for digital clones, the list of candidates for digital clones, and the list of employees who were not selected for digital clones. The effect verification module constructs relevant categories of digital clones based on the list of selected employees for digital clones. By comparing and analyzing the performance scores before and after the application of digital clones, it determines whether the application of digital clones is effective for the corresponding personnel, and generates and executes corresponding feedback adjustment instructions.

2. The enterprise-level employee digital avatar system based on artificial intelligence according to claim 1, characterized in that: The difficulty level of corresponding employees completing corresponding process tasks within a historical timeframe is determined and analyzed. The specific analysis process includes: Feature identification is performed on the task characteristic data information collected within the historical unit period. The number of approvals, the time limit for completion, and the number of uploaded files for each process task extracted by the corresponding employee are normalized and analyzed to determine the task complexity coefficient for the corresponding employee to complete the corresponding process task. Input the task complexity coefficients of the corresponding employees for completing the corresponding process tasks into the task complexity interval matching table, output the task difficulty level of the corresponding employees for completing the corresponding process tasks, and match the difficulty values. The specific process is as follows: If the task complexity coefficient of the corresponding process task is in the first task complexity interval of the task complexity interval comparison table, then the corresponding process task will be matched as a level 1 task difficulty. If the task complexity coefficient of the corresponding process task is in the second task complexity interval of the task complexity interval comparison table, then the corresponding process task will be matched as a level 2 task difficulty. If the task complexity coefficient of the corresponding process task is in the third task complexity interval of the task complexity interval comparison table, then the corresponding process task will be matched as a level 3 task difficulty. The difficulty value of the corresponding task is matched according to the difficulty level of the task. The difficulty of the first-level task is matched with a difficulty value of 3N, the difficulty of the second-level task is matched with a difficulty value of 2N, and the difficulty of the third-level task is matched with a difficulty value of 1N, and 3N>2N>1N, where N represents a positive integer.

3. The enterprise-level employee digital avatar system based on artificial intelligence according to claim 1, characterized in that: Analyze the overall response pressure of employees in completing various process tasks within a historical timeframe. The specific analysis process includes: Feature identification is performed on the task feature data information collected within the historical unit period. The task issuance time and the first processing time of the corresponding employee for each process task are extracted and subtracted to obtain the task response time of the corresponding employee for each process task. The task response time of each employee in completing each process task is correlated with the difficulty value of the corresponding process task. After quantitative processing, the overall response pressure of each employee in completing each process task within a historical unit period is analyzed to determine the overall response pressure value of the employee within a historical unit period, specifically: In the formula, This represents the overall response stress value of the j-th employee within a historical unit period. and Let i and n represent the task response time and difficulty value of the j-th employee completing the i-th process task, respectively, where i = 1, 2, 3, ..., n, and n represents the total number of tasks. Based on the comprehensive response stress value of each employee within a historical unit period, the corresponding employees are sorted in descending order to obtain the response stress personnel sequence. The top 50% of employees in the response stress personnel sequence are labeled as employees with high task capacity, and the bottom 50% of employees in the response stress personnel sequence are labeled as employees with low task capacity. Based on the labels indicating the corresponding employees' task capacity levels, construct a group of employees with capacity labels.

4. The enterprise-level employee digital avatar system based on artificial intelligence according to claim 1, characterized in that: Analyze the reliability level of employees completing various process tasks within a historical timeframe. The specific analysis process includes: Feature identification is performed on the task completion status data collected within the historical unit period to extract the number of tasks completed on time and the number of tasks completed overtime for each employee within the historical unit period. By comparing the number of tasks completed on time and the number of tasks completed late for each employee within a historical period, a group of employees with reliability labels is constructed. The specific construction process is as follows: If the number of tasks completed on time by the corresponding employee exceeds the number of tasks completed overtime, it indicates that the corresponding employee is in a controllable state of task completion within the historical unit period, and the corresponding employee is marked as a high-reliability employee. If the number of tasks completed on time by the corresponding employee does not exceed the number of tasks completed overtime, it indicates that the corresponding employee was in an uncontrollable state of task completion within the historical unit period, and the corresponding employee will be marked as a low-reliability employee. Based on the reliability level labels of the corresponding employees, construct a group of employees with reliability labels.

5. The enterprise-level employee digital avatar system based on artificial intelligence according to claim 1, characterized in that: A planning analysis was conducted on the digital identity determination of each employee within a historical timeframe. The specific analysis process included: Extract employee tags from the employee groups labeled with carrying capacity and those labeled with reliability, and count the tag categories marked by each employee; If the corresponding employee is simultaneously tagged as a high task capacity employee and a high reliability employee, a first-level digital clone degree signal is generated, and the corresponding employee is included in the digital clone selection employee list based on the generated first-level digital clone degree signal. If the corresponding employee is simultaneously labeled as a high task capacity employee and a low reliability employee, or simultaneously labeled as a low task capacity employee and a high reliability employee, a secondary digital clone degree signal is generated, and the corresponding employee is included in the digital clone candidate employee list based on the generated secondary digital clone degree signal. If the corresponding employee is simultaneously tagged as a low task capacity employee and a low reliability employee, a three-level digital clone degree signal is generated, and the corresponding employee is included in the digital clone rejection list based on the generated three-level digital clone degree signal.

6. The enterprise-level employee digital avatar system based on artificial intelligence according to claim 1, characterized in that: Based on the list of employees selected for digital avatars, relevant categories of digital avatars were constructed, and the performance scores before and after the application of digital avatars were analyzed. The specific analysis process included: Based on the list of employees selected for digital avatars and combined with the employees' interactions on the platform, relevant categories of digital avatars for each person in the list of selected employees are constructed. Among them, relevant categories of digital avatars include request-type digital avatars, response-type digital avatars, instruction-type digital avatars, summary-type digital avatars, and confirmation-type digital avatars. Based on the performance scores before and after the application of digital clones in relevant categories, the performance scores before and after the application of digital clones for each person in the list of selected employees were collected. Then, combined with the statistical summation algorithm, the total performance scores before and after the application of digital clones for each person in the list of selected employees were calculated. By comparing and analyzing the total performance scores of each employee in the digital avatar selection list before and after application, the performance difference of each employee in the digital avatar selection list before and after application is obtained.

7. The enterprise-level employee digital avatar system based on artificial intelligence according to claim 1, characterized in that: The performance difference before and after the application of digital avatars for each employee on the selected employee list is compared and analyzed with a preset difference threshold to determine whether the application of digital avatars is effective for the corresponding personnel. Based on this analysis, corresponding feedback adjustment instructions are generated and executed. The specific process includes: If the performance difference before and after the application of the digital clone exceeds the difference threshold, it indicates that the application of the digital clone by the corresponding person is invalid, and a first-level feedback adjustment instruction is generated, which includes: removing the corresponding person from the list of selected employees for digital clone and selecting a person from the list of candidate employees for digital clone as a replacement. If the performance difference before and after the application of the digital clone by the selected employees does not exceed the difference threshold, it indicates that the application of the digital clone by the corresponding personnel is effective, and a secondary feedback adjustment instruction is generated, which includes: continue to perform the digital clone application operation for the corresponding personnel.

Images