AI Chat human resource management method based on context awareness

Abstract

The invention provides an AI Chat human resource management method based on context awareness, and relates to the technical field of artificial intelligence and human resource management system integration, and the method comprises the steps: extracting the context information of a human resource system, storing the context information to an AI Chat robot, the context information comprises function description, an associated agent list and event driver information; generating a recommendation list on an AI chat interface according to the context information, combining the recommendation list with the text input by the user, and identifying through an intention matching algorithm to obtain a user intention; and converting the user intention into a standardized event definition based on the human resource event definition library.

Description

Technical Field

[0001] This application belongs to the field of artificial intelligence and human resource management system integration technology, specifically involving a context-aware AI Chat human resource management method. Background Technology

[0002] With the rapid development of artificial intelligence (AI) technology, AI chat has shown great potential in enterprise applications, especially in the field of human resource management. However, the integration of AI chat with traditional HR system software is currently ineffective and faces numerous challenges. First, the complex architecture and diverse business logic of HR systems make it difficult to deeply integrate AI chat in a universal way. Second, in terms of page function perception, existing technologies struggle to enable AI chat to accurately understand the functions of each page, preventing it from precisely driving corresponding page actions based on user needs. Furthermore, the lack of standardized methods for integrating AI output to drive page actions further limits the effectiveness of AI chat in HR management systems. Summary of the Invention

[0003] This application provides a context-aware AI Chat human resource management method to address one of the aforementioned technical problems.

[0004] The technical solution adopted in this application is as follows:

[0005] This application provides a context-aware AI Chat-based human resource management method, including:

[0006] Extract the context information from the human resources system and store it in the AI ​​chatbot. The context information includes function descriptions, a list of associated intelligent agents, and event-driven information.

[0007] The AI ​​chat interface generates a recommendation list based on contextual information, combines the recommendation list with the user's input text, and identifies the user's intent through an intent matching algorithm.

[0008] Based on the human resources event definition library, user intents are converted into standardized event definitions.

[0009] According to one embodiment of this application, the extraction of context information from the human resources page and the storage of the context information in the AI ​​chat robot, the context information includes a function description, a list of associated intelligent agents, and event driver information, specifically:

[0010] Context information is stored in the AI ​​chatbot using a hierarchical key-value pair data structure, where the functional description serves as the first-level key, and the list of intelligent agents and event driver information serve as the second-level key.

[0011] According to one embodiment of this application, the step of generating a recommendation list in the AI ​​chat interface based on contextual information, combining the recommendation list with user input text, and identifying the user intent through an intent matching algorithm specifically includes:

[0012] Based on the function description and intelligent agent list in the context information, a set of intent candidates related to the current page's business scenario is preset;

[0013] The recommendation list is generated using a priority sorting algorithm, prioritizing the display of intent items that frequently match the user's historical operations, and labeling the confidence score of each intent item;

[0014] The intent matching algorithm is implemented through a context-aware natural language processing model. The input includes the user's input text and the functional description in the context information. The output is the matched user intent, and the intent irrelevant to the current page is excluded through a context filtering mechanism.

[0015] According to one embodiment of this application, the step of converting user intent into standardized event definitions based on the human resources event definition library specifically includes:

[0016] Retrieve event templates that match the user's intent from the human resources event definition library. The event templates contain operation instructions and parameter rules.

[0017] Standardized event definitions include permission verification rules, which are used to verify whether a user has the permission to execute an event;

[0018] The event definition library supports dynamic expansion and automatically updates event templates based on user feedback or system logs.

[0019] According to one embodiment of this application, it also includes:

[0020] When a user intent conflicts with event-driven information in the context, the user intent is automatically corrected using priority rules.

[0021] After the standardized event definition is generated, it is sent back to the human resources system through the event driver interface, and the functional description status in the context information is updated at the same time.

[0022] According to one embodiment of this application, it also includes:

[0023] Continuously monitor the interaction between AI chat and users, and record user operation paths and intent recognition accuracy;

[0024] Optimize the intent matching algorithm through log analysis and adjust the weight of the recommendation list;

[0025] The extracted context information and event definition library are dynamically updated based on user feedback.

[0026] According to one embodiment of this application, the information display of the AI ​​chat interface adopts a diversified component mixed layout technology, specifically:

[0027] The diverse components include text flow components, card components, list components, and chart components;

[0028] Diverse components are linked to system data sources in real time through a metadata binding mechanism to ensure that the displayed content is consistent with the latest status of the human resources system;

[0029] When a user clicks on an interactive element in a variety of components, a pop-up event chain is triggered, dynamically loading content.

[0030] According to one embodiment of this application, it also includes a data compliance check:

[0031] Check whether the data involved in the incident complies with privacy protection rules;

[0032] Check whether the incident handling meets business compliance requirements;

[0033] If the check fails, the event blocking mechanism is triggered, and the user is prompted to correct the operation or submit supplementary materials.

[0034] A second aspect of this application provides a computer-readable storage medium having a program stored thereon that, when executed by a processor, implements the steps described in the method.

[0035] A third aspect of this application provides an electronic device including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method as described.

[0036] Due to the adoption of the above technical solution, the beneficial effects achieved by this application are as follows:

[0037] This application extracts contextual information (including function descriptions, a list of associated agents, and event-driven information) from human resources pages and stores it in an AI chatbot. This enables the AI ​​chatbot to deeply understand the functions and operational logic of each page in the human resources system, achieving seamless integration with the system. For example, in the employee recruitment process, the AI ​​chatbot can accurately identify user intent based on context and automatically perform tasks such as resume screening and interview scheduling, significantly improving recruitment efficiency and accuracy.

[0038] This application utilizes contextual information to generate a recommendation list and combines this list with the user's input text. An intent matching algorithm is then used to identify the user's intent, significantly improving the accuracy of AI chat's understanding of user questions. For example, on a performance evaluation page, when a user asks "How to improve performance scores," AI chat can provide targeted suggestions and operational guidance, enhancing the accuracy and fluency of human-computer interaction.

[0039] This application, based on a human resources event definition library, transforms user intent into standardized event definitions, clarifying the conversion rules between AIchat output and human resources system operations, and ensuring accurate and efficient communication between the two. For example, in the payroll calculation process, the event definitions output by AIchat can precisely guide the system to complete a series of operations such as data collection and calculation rule application, ensuring the standardization and accuracy of the operations.

[0040] This application significantly improves the user experience by embedding AI event-driven pop-ups next to the AI ​​chat dialog box. For example, when filling out an employee information form, the relevant pop-up is displayed adjacent to the AI ​​chat dialog box, allowing users to interact with the AI ​​chat for guidance while quickly entering information into the pop-up form, effectively improving work efficiency.

[0041] This application supports the mixed display of various components, such as text flow components, card components, list components, and chart components, providing rich and powerful information display and analysis capabilities. For example, when analyzing employee performance data, users can see specific numerical values ​​and intuitively grasp the overall trend. At the same time, each component supports metadata event definition, which can call pop-up windows to display more details, meeting the diverse and in-depth analysis and display needs in human resource management. Attached Figure Description

[0042] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0043] Figure 1 A flowchart illustrating a context-aware AI Chat human resource management method provided in this application embodiment;

[0044] Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0045] Figure label:

[0046] 810, Processor; 820, Communication interface; 830, Memory; 840, Communication bus. Detailed Implementation

[0047] To more clearly illustrate the overall concept of this application, a detailed explanation is provided below with reference to the accompanying drawings.

[0048] Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application may also be implemented in other ways different from those described herein. Therefore, the scope of protection of this application is not limited to the specific embodiments disclosed below. It should be noted that, unless otherwise specified, the embodiments of this application and the features thereof can be combined with each other.

[0049] In this application, unless otherwise expressly specified and limited, the "above" or "below" of the second feature can mean that the first and second features are in direct contact, or that the first and second features are in indirect contact through an intermediate medium. In the description of this specification, references to terms such as "an embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples.

[0050] Example 1

[0051] like Figure 1 As shown, a context-aware AI Chat-based human resource management method includes:

[0052] S100. Extract the context information from the human resources system and store the context information in the AI ​​chat robot. The context information includes function descriptions, a list of associated intelligent agents, and event-driven information.

[0053] As mentioned above, extracting contextual information refers to retrieving key information from the current page from the Human Resources Management System (HRMS) and passing it to the AI ​​chatbot. This information includes, but is not limited to:

[0054] Function Description: This section details the functions of each page, such as employee file query, attendance data entry, or payroll calculation.

[0055] List of associated agents: Lists all AI models available on the current page, such as agents for data analysis or agents for document processing.

[0056] Event driver information: Defines the various interactive events that may occur on the page, as well as their triggering conditions and execution logic, such as calling a specific API after a button is clicked.

[0057] This information is stored in a structured manner in the AI ​​chatbot's memory to facilitate subsequent intent recognition and operation execution based on user input.

[0058] For example, suppose you're using a human resources system to manage a company's hiring process. When a hiring manager opens the "Candidate Screening" page, its function description might be "View and screen applicants' resumes." In this scenario, the associated agent list might include one agent specifically for parsing resume text, and another agent for evaluating candidate fit. Event-driven information could include "Load the candidate's resume file when the user clicks the 'View Resume' button."

[0059] Once this information is successfully extracted and stored in the AI ​​chatbot, if a user asks, "Which candidate's resume should I prioritize?", the AI ​​chatbot can understand the user's intent based on the stored context information and invoke the appropriate agent to analyze the candidate data, thereby providing suggestions.

[0060] It should be noted that, in specific implementation scenarios, in order to adapt to constantly changing business needs, the context information can not only be extracted and stored during initialization, but also updated in real time with user operations. For example, if a new functional module (such as performance evaluation) is added, the system can automatically recognize this change and update the context information to support the new business process.

[0061] In specific implementation scenarios, based on the above solution, although initially designed for human resource management systems, this method is also applicable to other types of enterprise management systems, such as customer relationship management (CRM) or financial management systems. By adjusting the way contextual information is extracted, the AI ​​chatbot can be seamlessly integrated with different types of systems.

[0062] In specific implementation scenarios, based on the above solutions and considering data sensitivity, encryption technology can be introduced to ensure information security before storing context information. Furthermore, access control policies can be implemented to ensure that only authorized users can access certain types of context information.

[0063] In specific implementation scenarios, building upon the above solutions and leveraging the collected contextual information, AIchat can not only provide general services but also offer personalized suggestions and services based on user preferences and historical behavior. For example, for managers who frequently focus on team-building activities, AIchat can proactively push information about relevant training courses.

[0064] In specific implementation scenarios, to enable businesses worldwide to benefit from this technology, the context information extraction mechanism can be expanded to support multiple languages, building upon the above solutions. This means that regardless of the user's native language, AI chat can accurately understand and respond to their requests.

[0065] S200. Generate a recommendation list in the AI ​​chat interface based on context information, combine the recommendation list with the user's input text, and identify the user's intent through an intent matching algorithm.

[0066] As described above, based on contextual information extracted and stored from the human resources system (such as page function descriptions and agent lists), AI chat automatically generates a recommendation list containing pre-set operation suggestions or frequently asked questions for the current business scenario. Then, when a user enters text into the AI ​​chat interface, the system not only analyzes the user's direct input but also combines it with the current contextual information to more accurately understand the user's actual needs or intentions. Finally, an intent matching algorithm comprehensively processes this information to determine the operation instructions or query requests that best meet the user's needs.

[0067] For example, suppose an HR professional is using the "Employee Performance Appraisal" page in a human resources system and opens AI chat. Based on the page's functional description (i.e., performance appraisal), AI chat will display a recommendation list, possibly including "View the most recent performance appraisal results," "Start a new performance appraisal cycle," or "How to set performance goals." If the user enters, "I want to know how Xiao Zhang performed last year?", AI chat will combine the context (in this case, performance appraisal) and the user's input, using an intent matching algorithm to identify the user's true intent: to query the historical performance records of a specific employee. Subsequently, AI chat can invoke the appropriate agent to execute the query and provide the user with the necessary information.

[0068] It's worth noting that in specific implementation scenarios, in addition to generating a recommendation list based on the current page context, the above solutions can be further customized based on the user's historical behavior data and personal preferences. For example, for HR professionals who frequently participate in training management, AI chat can prioritize displaying recommendations related to training arrangements and effectiveness evaluations on relevant pages, thereby improving work efficiency.

[0069] In specific implementation scenarios, in addition to page-level contextual information, the above solutions can be expanded to include more types of contextual information, such as time and user role dimensions, to enrich the recommendation list. For example, during annual performance evaluations, targeted recommendation lists can be provided to managers at different levels to help them better perform their performance management tasks.

[0070] In specific implementation scenarios, based on the above approach, as users interact with the AI ​​chat more, the system can continuously optimize the weight of each suggestion in the recommendation list based on feedback. For example, if a recommendation is found to be frequently ignored, its frequency of appearance will be reduced; conversely, popular recommendations will be given higher priority.

[0071] In specific implementation scenarios, building upon the above solutions, in large enterprises, human resource management systems are often closely integrated with other business systems (such as financial systems and project management systems). Therefore, AI chat can not only generate recommendation lists based on the context of a single system, but also provide comprehensive suggestions across data sources integrated from multiple systems. For example, when handling performance evaluations involving salary adjustments, AI chat can obtain budget constraint information from the financial system to provide HR with more comprehensive operational guidance.

[0072] In specific implementation scenarios, based on the above solutions, to improve the accuracy of the intent matching algorithm, natural language processing technology can be continuously improved, enabling it not only to understand explicit question statements but also to accurately capture implicit needs or emotional tendencies. For example, when a user expresses "I'm not satisfied with the recruitment situation this month," AI chat should be able to identify the negative emotions behind this and proactively offer improvement suggestions or ask if they need to view a detailed recruitment data analysis report.

[0073] S300, based on the human resources event definition library, converts user intent into standardized event definitions.

[0074] As mentioned above, a standardized bridge is built to enable AI chat to understand and execute user intents. First, the "HR Event Definition Library" is a predefined set of rules that details the event types, parameter requirements, and execution logic corresponding to various operations within the HR system. Once AI chat identifies the user's intent using contextual information and an intent matching algorithm, it queries this event definition library to find standard event definitions that match the user's intent. Then, based on these standard definitions, AI chat translates the user's intent into specific system operation instructions, ensuring that the operation can be accurately understood and executed by the HR system.

[0075] For example, suppose an HR professional wants to schedule onboarding training for a new employee on the "Employee Onboarding Management" page. The user might type in AI chat: "Please schedule onboarding training for newly hired Xiao Li." AI chat first recognizes this as a request for "onboarding training scheduling," and considers the current page context (i.e., employee onboarding management). Next, AIchat will search its event definition library for standard event definitions related to "scheduling onboarding training." These definitions might include:

[0076] Event Type: Initiating Onboarding Training Process

[0077] Parameter requirements: Employee ID, training module selection, and other information are required.

[0078] Execution logic: Call a specific API or service to create a training plan and send notifications to relevant personnel.

[0079] Once the correct event definition is determined, AI Chat will generate corresponding operation instructions and automatically complete the onboarding training arrangement process for new employee Xiao Li.

[0080] It should be noted that, in specific implementation scenarios, in addition to the above solutions, a mechanism can be designed to adapt to constantly changing enterprise needs, enabling the event definition library to automatically expand and adjust based on user feedback or system updates. For example, if an enterprise introduces a new performance evaluation method, the event definition library can be updated through simple configuration without redeveloping the entire system.

[0081] In specific implementation scenarios, building upon the above solutions and considering the diversity of global enterprises, support for multiple languages ​​can be added to the event definition library. This ensures that regardless of the language used by the user, AI chat can correctly translate their intent into standardized event definitions. This not only enhances the user experience but also helps break down language barriers and promotes collaboration between multinational teams.

[0082] In specific implementation scenarios, to ensure data security and compliance, an access control system can be integrated into the process of converting user intent into event definitions, building upon the above solutions. Only users with the appropriate permissions can trigger certain sensitive operations (such as salary adjustments, approval of important documents, etc.), thereby enhancing system security.

[0083] In specific implementation scenarios, based on the above solution, although initially designed for human resources systems, this approach is equally applicable to other business management systems. By adapting to different event definition formats and API interfaces, AI chat can seamlessly integrate with other systems (such as financial systems and project management systems), enabling a wider range of applications.

[0084] In specific implementation scenarios, a real-time monitoring system can be established based on the above solutions to track the actual execution effect of each event definition and its impact on business processes. Based on the collected data, the rules in the event definition library can be optimized periodically to improve the overall system efficiency and accuracy. For example, if a certain type of event definition is found to cause frequent operation failures or delays, relevant parameters or logic can be adjusted promptly to resolve the issue.

[0085] According to one embodiment of this application, the extraction of context information from the human resources page and the storage of the context information in the AI ​​chat robot, the context information includes a function description, a list of associated intelligent agents, and event driver information, specifically:

[0086] Context information is stored in the AI ​​chatbot using a hierarchical key-value pair data structure, where the functional description serves as the first-level key, and the list of intelligent agents and event driver information serve as the second-level key.

[0087] As mentioned above, to achieve deep integration between AI chat and the Human Resources Management System (HRMS), it is first necessary to extract key contextual information from various pages of the HRMS and store it in the AI ​​chat robot's memory. This contextual information mainly includes the following three parts:

[0088] Function Description: Provide a detailed explanation of the function of each page. For example, on the "Employee Profile Management" page, the function description might be "Used to view and edit basic employee information." This helps AI chat understand the user's current business scenario.

[0089] A list of associated agents: This lists all AI models or agents that can be invoked on the current page. For example, the "Performance Appraisal" page might include one agent for analyzing employee performance data and another for generating performance reports. This allows AI Chat to select the appropriate agent to handle the request based on the user's needs.

[0090] Event-driven information: Defines the various interactive events that may occur on the page, along with their triggering conditions and execution logic. For example, "When the user clicks the 'Submit' button, call a specific API to save the data."

[0091] To effectively manage and utilize this contextual information, a hierarchical key-value data structure is used for storage. Specifically:

[0092] First-level keys: Use function descriptions as first-level keys. This means that the main function of each page will have a unique identifier. For example, "Employee File Management" or "Performance Evaluation".

[0093] The second-level key, under the first-level key, is further subdivided into two subkeys: the agent list and the event driver information. These two subkeys store information about all agents related to this page and information about all triggerable events, respectively.

[0094] Through this hierarchical data structure design, AI chat can quickly retrieve the necessary contextual information, thereby more accurately understanding and responding to user requests. For example, when a user asks a question about a certain operation on a specific page, AI chat can quickly locate the correct agent or event driver based on the stored contextual information and take appropriate action. Furthermore, this structure supports dynamic updates, allowing the contextual information to be adjusted in real time as user actions change, ensuring that AI chat always has the latest contextual data.

[0095] According to one embodiment of this application, the step of generating a recommendation list in the AI ​​chat interface based on contextual information, combining the recommendation list with user input text, and identifying the user intent through an intent matching algorithm specifically includes:

[0096] Based on the function description and intelligent agent list in the context information, a set of intent candidates related to the current page's business scenario is preset;

[0097] The recommendation list is generated using a priority sorting algorithm, prioritizing the display of intent items that frequently match the user's historical operations, and labeling the confidence score of each intent item;

[0098] The intent matching algorithm is implemented through a context-aware natural language processing model. The input includes the user's input text and the functional description in the context information. The output is the matched user intent, and the intent irrelevant to the current page is excluded through a context filtering mechanism.

[0099] As mentioned above, firstly, based on the contextual information extracted from the Human Resources page, particularly the function descriptions and associated agent list, the system pre-defines a candidate set of intents related to the current page's business scenario. For example, when a user is on the "Performance Appraisal" page, the function description is "Perform employee performance rating and feedback," and the agent list includes "Performance Data Calculation Agent" and "Performance Report Generation Agent." Based on this, the system pre-defines possible user intents, such as "Start Performance Appraisal," "View an employee's historical ratings," or "Generate a department performance summary." These pre-defined intents constitute the candidate set, serving as the foundation for the recommendation list.

[0100] Next, the system generates the final recommendation list using a priority ranking algorithm. This algorithm combines users' historical behavior data to rank the intents in the candidate set. Intents corresponding to actions frequently performed by the user in the past have higher priority in the recommendation list. For example, if a user frequently uses the "Export Performance Data" function, this intent will be prioritized when visiting the performance page subsequently. Simultaneously, the system assigns a confidence score to each intent item in the list, reflecting the degree to which the intent matches the current context and user behavior patterns. A higher score indicates that the system believes the intent is more likely to be the user's next action.

[0101] After the user actually inputs text, the system activates the intent matching algorithm. This algorithm employs a context-aware natural language processing model, whose input includes not only the text content entered by the user but also the contextual information currently stored in the AI ​​chatbot, especially the function description. When analyzing user statements, the model combines the business context of the current page to improve its understanding of professional terminology and scenario-based expressions. For example, if a user enters "View recent resumes" on the "Recruitment Management" page, the model can accurately identify their intent as "View the latest received candidate resumes," rather than referring to all documents in general, based on the function description of "Recruitment."

[0102] In addition, the system incorporates a contextual filtering mechanism to exclude intents that are clearly irrelevant to the current page. For example, when a user is on the "Attendance Management" page, even if the input text contains the word "salary," the system will determine, based on the current function description, that "adjusting salary structure" is not within the scope of operations supported by this page, thus filtering out this intent from the matching results and avoiding misidentification. The final output user intent is a precise result after context enhancement, priority ranking, and contextual filtering, ensuring the accuracy of subsequent operations and the relevance of the system's response.

[0103] According to one embodiment of this application, the step of converting user intent into standardized event definitions based on the human resources event definition library specifically includes:

[0104] Retrieve event templates that match the user's intent from the human resources event definition library. The event templates contain operation instructions and parameter rules.

[0105] Standardized event definitions include permission verification rules, which are used to verify whether a user has the permission to execute an event;

[0106] The event definition library supports dynamic expansion and automatically updates event templates based on user feedback or system logs.

[0107] As described above, after the user's intent is identified, the system accesses the human resources event definition library to find an event template that matches that intent. This event definition library is a pre-established set of rules containing standardized definitions for various executable operations in the human resources management process. Each event template corresponds to a specific business operation type, such as "creating an employee file," "initiating a performance evaluation process," or "approving a leave application." Each template contains explicit operation instructions, indicating the specific actions the system needs to perform, as well as parameter rules, specifying the data fields, input formats, and data sources required to perform the operation. For example, the "creating an employee file" event template might require parameters such as employee name, employee ID, department, and date of employment, specifying the format and mandatory nature of these parameters.

[0108] After matching a suitable event template, the system converts it into a standardized event definition. This definition not only includes operation instructions and parameter rules but also integrates permission verification rules. Permission verification rules are used to verify whether the current user has the corresponding operation permissions before executing the event. For example, only users with the "HR Supervisor" role can perform the "Adjust Salary" operation, while ordinary employees can only view their own salary information. When generating standardized event definitions, the system automatically attaches corresponding permission verification conditions to ensure that subsequent execution processes comply with enterprise security policies and data access control requirements, preventing unauthorized operations.

[0109] Furthermore, the HR event definition library has dynamic expansion capabilities. The system can automatically identify new operational needs or optimize existing templates based on user feedback or backend operation logs. For example, when multiple users frequently request a type of operation that is not yet supported (such as "batch import training records"), the system can automatically generate a new event template and add it to the event definition library by analyzing high-frequency keywords and operation patterns in the logs. Similarly, if an event template frequently encounters parameter errors or user cancellations during execution, the system can adjust its parameter rules or optimize its matching logic accordingly. This dynamic update mechanism enables the event definition library to continuously adapt to changes in enterprise business processes, improving AI chat's adaptability to new scenarios and its long-term availability.

[0110] According to one embodiment of this application, it also includes:

[0111] When a user intent conflicts with event-driven information in the context, the user intent is automatically corrected using priority rules.

[0112] After the standardized event definition is generated, it is sent back to the human resources system through the event driver interface, and the functional description status in the context information is updated at the same time.

[0113] As mentioned above, in practice, there may be situations where the user's intent conflicts with the contextual information of the current page (especially event-driven information). For example, a user might enter a request for "performance evaluation" on a page specifically for "employee file management." In this case, the user's intent does not perfectly match the page's functional description.

[0114] To address this situation, the system has a built-in set of priority rules to automatically adjust user intent. These rules are based on the following factors:

[0115] Page Function Priority: The core functions of certain pages have higher priority. For example, on the "Employee File Management" page, all operations should first consider needs related to file management.

[0116] Historical behavior patterns: If a user frequently performs a specific type of action on the same page, the system will adjust the results of intent recognition based on this pattern.

[0117] Business logic constraints: Some operations may require specific business logic conditions to be met before they can be executed. For example, salary adjustments can only be made after the initial review is completed.

[0118] When a conflict is detected, the system will re-evaluate and correct the user's intent according to the aforementioned priority rules. For example, if a user asks how to conduct performance evaluations on the "Employee File Management" page, the system may automatically correct their intent to view the employee's historical performance records, as this better aligns with the current page's functional description and business logic.

[0119] Once the user intent is accurately identified and transformed into standardized event definitions, the next step is to send these definitions back to the human resources system via the event-driven interface to perform the corresponding operations.

[0120] The specific process is as follows:

[0121] Event definition generation: Based on user intent and context information, the system retrieves suitable templates from the event definition library and generates specific standardized event definitions by combining them with permission verification rules. For example, if the user intent is "update an employee's basic information", an event definition containing all necessary fields (such as name, department, position, etc.) and corresponding verification rules will be generated.

[0122] Event-driven interface calls: The generated standardized event definitions are sent to the human resources system through a pre-defined event-driven interface. This interface ensures that events are executed according to predetermined processes and rules. For example, the event definition for "update employee basic information" will be passed to the service module responsible for data updates, triggering the corresponding database operation.

[0123] Context information updates: As the event definition is executed, the system also synchronously updates the context information stored in the AI ​​chatbot. For example, if a user has just completed an information update operation for an employee, the function description status in the context information will be updated to reflect this change. This helps maintain the real-time and accuracy of the context information, providing more precise support for subsequent operations or queries.

[0124] Feedback Mechanism: The system can also be designed with a feedback mechanism to monitor the success rate and effectiveness of event definition execution. If any anomalies or unexpected results are detected, the system can further adjust the context information or optimize the event definition to improve the accuracy and efficiency of future operations.

[0125] According to one embodiment of this application, it also includes:

[0126] Continuously monitor the interaction between AI chat and users, and record user operation paths and intent recognition accuracy;

[0127] Optimize the intent matching algorithm through log analysis and adjust the weight of the recommendation list;

[0128] The extracted context information and event definition library are dynamically updated based on user feedback.

[0129] As mentioned above, to ensure the efficient operation of the AI ​​chat system and the continuous improvement of user experience, the system needs to monitor every interaction between the AI ​​chat and the user in real time. Specifically, this includes:

[0130] Record user action paths: The system meticulously records every step a user takes in AI chat, such as the questions they enter, the recommendations they choose, and the commands they trigger. This data helps us understand user behavior patterns and preferences.

[0131] Tracking intent recognition accuracy: The system also records the result and accuracy of each intent recognition. If the user's actual needs do not match the intent recognized by the system, the recognition will be marked as an error and recorded for subsequent analysis.

[0132] Through this continuous monitoring, the system can collect a large amount of user interaction data, providing a basis for subsequent optimization.

[0133] Based on the collected log data, the system can perform in-depth log analysis to continuously optimize the intent matching algorithm. The specific steps are as follows:

[0134] Data Analysis: The system analyzes the recorded user action paths and intent recognition results to identify common error types and failure patterns. For example, certain types of queries may be frequently misidentified as other irrelevant intents.

[0135] Adjusting Recommendation List Weights: Based on the analysis results, the system can dynamically adjust the weights of each suggestion in the recommendation list. Options that are frequently selected by users will have their priority increased; while options that are rarely selected may have their display frequency reduced or their relevance reassessed.

[0136] Optimize the intent matching algorithm: By learning from a large number of real-world cases, the system can improve the intent matching algorithm itself. For example, adding new feature variables to improve recognition accuracy, or introducing more complex model structures to enhance the ability to understand complex contexts.

[0137] This data-driven approach enables AI chat to become more intelligent and accurate over time.

[0138] Besides automated log analysis, direct user feedback is also a significant source of system improvement. Specific measures include:

[0139] Dynamically updating contextual information: When receiving user feedback about a specific page or function, the system can dynamically adjust the extracted contextual information based on this feedback. For example, if multiple users report that the function description on a certain page is not clear enough, the system can update the function description on that page to more accurately reflect its actual use.

[0140] Expanding the event definition library: User feedback can also help identify shortcomings in the existing event definition library. For example, if users frequently request certain operations that are not yet supported (such as "batch import employee training records"), the system can automatically generate new event templates and add them to the event definition library by analyzing this feedback. This not only enriches the system's feature set but also improves the responsiveness to new business needs.

[0141] Personalized services: In addition, the system can further customize service content based on user feedback. For example, for users who frequently ask similar questions, the system can pre-include relevant solutions in the recommendation list, thereby reducing repeated inquiries and improving efficiency.

[0142] According to one embodiment of this application, the information display of the AI ​​chat interface adopts a diversified component mixed layout technology, specifically:

[0143] The diverse components include text flow components, card components, list components, and chart components;

[0144] Diverse components are linked to system data sources in real time through a metadata binding mechanism to ensure that the displayed content is consistent with the latest status of the human resources system;

[0145] When a user clicks on an interactive element in a variety of components, a pop-up event chain is triggered, dynamically loading content.

[0146] As mentioned above, in order to meet the display needs of different types of information and improve the clarity of information expression and ease of operation during AI chat interaction with users, the system adopts a diversified component mixing technology. This technology allows for the flexible combination and arrangement of various display components, including text flow components, card components, list components, and chart components, based on the nature of the response content within the same AI chat interface.

[0147] Text flow components are used to display continuous natural language content, such as explanations of user questions, operating instructions, or process guidance. Card components are used to highlight key information units, such as basic information about an employee, performance ratings, or summaries of to-do items, featuring clear structure and visual focus. List components are suitable for presenting multiple pieces of structured data, such as employee attendance records, salary details, or lists of job candidates, facilitating quick browsing and comparison for users. Chart components are used to visualize data trends and distributions, such as departmental performance score distribution charts or monthly recruitment trend charts, helping users intuitively understand complex data.

[0148] All diverse components are linked in real-time to the data sources of the human resources system through a metadata binding mechanism. This mechanism means that each component defines its dependent data fields and their source system interfaces during configuration. When the system status changes (such as updates to employee salary information or additions to attendance data), the binding mechanism automatically triggers a refresh of the component content, ensuring that the information displayed in the AI ​​chat interface is always consistent with the latest data in the human resources system, avoiding information lag or inconsistencies.

[0149] Furthermore, all interactive elements in each component (such as links in text, operation buttons in cards, items in lists, and data points in charts) are configured with event listener mechanisms. When a user clicks on these interactive elements, the system triggers a pre-defined pop-up event chain. This event chain first parses the context information and operation intent corresponding to the currently clicked element, and then dynamically loads and displays the relevant content. For example, if a user clicks the "View Details" button in the "Performance Rating" card, the system will call the relevant interface to obtain the detailed composition of the rating and display it in a structured table or chart in a pop-up window; clicking on a data point in the chart will pop up the detailed business record corresponding to that point in time. The pop-up content is dynamically generated based on actual data, supports multi-level information expansion, and achieves a seamless browsing experience from overview to detail.

[0150] According to one embodiment of this application, it also includes a data compliance check:

[0151] Check whether the data involved in the incident complies with privacy protection rules;

[0152] Check whether the incident handling meets business compliance requirements;

[0153] If the check fails, the event blocking mechanism is triggered, and the user is prompted to correct the operation or submit supplementary materials.

[0154] As mentioned above, before converting user intent into standardized event definitions and preparing to execute related operations, the system performs a series of data compliance checks to ensure that all involved operations and data processing comply with established privacy protection rules and business compliance requirements. This step is crucial because it not only ensures the security and legality of data but also prevents unauthorized or non-compliant operations from being executed.

[0155] First, the system checks the data to be processed against privacy protection rules. These rules typically include, but are not limited to:

[0156] The principle of data minimization: Only collect and process the data necessary to complete a specific task. For example, when making salary adjustments, the system will only access fields related to that employee's salary and will not touch other irrelevant information.

[0157] Anonymization and desensitization: For sensitive personal information (such as ID card number, home address, etc.), the system will anonymize or desensitize it before displaying or transmitting it to ensure that even if the data is leaked, the individual's identity will not be directly exposed.

[0158] Access control: Verifies whether the current user has permission to view or modify specific types of data. For example, a regular employee may only be able to view their own attendance records, while an HR manager may have access to the attendance data for the entire department.

[0159] If any violation of privacy rules is detected, the system will immediately block further operations and prompt the user to correct the problem or provide necessary supplementary materials.

[0160] In addition to privacy protection, the system must also ensure that event operations comply with the company's business compliance requirements. These requirements may come from internal policies, industry standards, or laws and regulations, and include, but are not limited to:

[0161] Approval Process: Some operations (such as salary adjustments and approval of important documents) require multiple levels of approval before they take effect. The system will automatically check whether the current operation has obtained all the necessary approvals.

[0162] Time constraints: Some business operations have strict time windows. For example, performance evaluations must be completed within a specified period; new evaluation results are not allowed to be submitted after the deadline.

[0163] Format and completeness requirements: The system will check whether the input data conforms to the preset format specifications and whether the content is complete and error-free. For example, when filling in employee files, required fields cannot be empty, and the date format must be correct, etc.

[0164] If an operation fails to pass the above business compliance checks, the system will also trigger a blocking mechanism to prevent non-compliant operations from being executed.

[0165] Once a data or business compliance issue is detected, the system will activate the event blocking mechanism, suspending the current operation. Simultaneously, the system will issue a clear notification to the user, explaining the specific problem and the corrective actions required. For example:

[0166] If the issue stems from a lack of necessary approval signatures, the system will prompt the user to contact the appropriate approver to sign the document as soon as possible.

[0167] If the error is due to an incorrect input data format, a specific error message will be given, along with suggestions on how to correct it.

[0168] In addition, depending on the specific circumstances, the system may require users to provide additional supporting documents or supplementary information. For example, when making interdepartmental transfers, it may be necessary to upload relevant meeting minutes or other supporting documents as evidence.

[0169] A second aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in any of the embodiments of the first aspect above.

[0170] Figure 2 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 2 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 may call logical instructions in the memory 830 to execute the method in any of the embodiments of the first aspect described above, the method including:

[0171] Extract the context information from the human resources system and store it in the AI ​​chatbot. The context information includes function descriptions, a list of associated intelligent agents, and event-driven information.

[0172] The AI ​​chat interface generates a recommendation list based on contextual information, combines the recommendation list with the user's input text, and identifies the user's intent through an intent matching algorithm.

[0173] Based on the human resources event definition library, user intents are converted into standardized event definitions.

[0174] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.

[0175] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program, the computer program being able to be stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer being able to perform the methods provided by the above methods, the method comprising:

[0176] Extract the context information from the human resources system and store it in the AI ​​chatbot. The context information includes function descriptions, a list of associated intelligent agents, and event-driven information.

[0177] The AI ​​chat interface generates a recommendation list based on contextual information, combines the recommendation list with the user's input text, and identifies the user's intent through an intent matching algorithm.

[0178] Based on the human resources event definition library, user intents are converted into standardized event definitions.

[0179] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the cigarette box image recognition method provided by the methods described above, the method comprising:

[0180] Extract the context information from the human resources system and store it in the AI ​​chatbot. The context information includes function descriptions, a list of associated intelligent agents, and event-driven information.

[0181] The AI ​​chat interface generates a recommendation list based on contextual information, combines the recommendation list with the user's input text, and identifies the user's intent through an intent matching algorithm.

[0182] Based on the human resources event definition library, user intents are converted into standardized event definitions.

[0183] For any parts not mentioned in this application, existing technologies may be used or referenced.

[0184] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0185] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A context-aware AI Chat-based human resource management method, characterized in that, include: Extract the context information from the human resources system and store it in the AI ​​chatbot. The context information includes function descriptions, a list of associated intelligent agents, and event-driven information. The AI ​​chat interface generates a recommendation list based on contextual information, combines the recommendation list with the user's input text, and identifies the user's intent through an intent matching algorithm. Based on the human resources event definition library, user intents are converted into standardized event definitions.

2. The method according to claim 1, characterized in that, The extraction of context information from the human resources page and storage of this context information in the AI ​​chatbot includes a function description, a list of associated intelligent agents, and event-driven information, specifically: Context information is stored in the AI ​​chatbot using a hierarchical key-value pair data structure, where the functional description serves as the first-level key, and the list of intelligent agents and event driver information serve as the second-level key.

3. The method according to claim 1, characterized in that, The process involves generating a recommendation list on the AI ​​chat interface based on contextual information, combining the recommendation list with the user's input text, and identifying the user's intent using an intent matching algorithm. Specifically: Based on the function description and intelligent agent list in the context information, a set of intent candidates related to the current page's business scenario is preset; The recommendation list is generated using a priority sorting algorithm, prioritizing the display of intent items that frequently match the user's historical operations, and labeling the confidence score of each intent item; The intent matching algorithm is implemented through a context-aware natural language processing model. The input includes the user's input text and the functional description in the context information. The output is the matched user intent, and the intent irrelevant to the current page is excluded through a context filtering mechanism.

4. The method according to claim 1, characterized in that, The process of converting user intents into standardized event definitions based on the human resources event definition library specifically involves: Retrieve event templates that match the user's intent from the human resources event definition library. The event templates contain operation instructions and parameter rules. Standardized event definitions include permission verification rules, which are used to verify whether a user has the permission to execute an event; The event definition library supports dynamic expansion and automatically updates event templates based on user feedback or system logs.

5. The method according to claim 1, characterized in that, Also includes: When a user intent conflicts with event-driven information in the context, the user intent is automatically corrected using priority rules. After the standardized event definition is generated, it is sent back to the human resources system through the event driver interface, and the functional description status in the context information is updated at the same time.

6. The method according to claim 1, characterized in that, Also includes: Continuously monitor the interaction between AI chat and users, and record user operation paths and intent recognition accuracy; Optimize the intent matching algorithm through log analysis and adjust the weight of the recommendation list; The extracted context information and event definition library are dynamically updated based on user feedback.

7. The method according to claim 1, characterized in that, The information display in the AI ​​chat interface employs a diverse component mix-and-match technique, specifically: The diverse components include text flow components, card components, list components, and chart components; Diverse components are linked to system data sources in real time through a metadata binding mechanism to ensure that the displayed content is consistent with the latest status of the human resources system; When a user clicks on an interactive element in a variety of components, a pop-up event chain is triggered, dynamically loading content.

8. The method according to claim 1, characterized in that, This also includes data compliance checks: Check whether the data involved in the incident complies with privacy protection rules; Check whether the incident handling meets business compliance requirements; If the check fails, the event blocking mechanism is triggered, and the user is prompted to correct the operation or submit supplementary materials.

9. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-8.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1-8.

Images