Human resource data warehouse construction method and system based on big data

Abstract

The invention provides a human resource data warehouse construction method and system based on big data. According to the method, related information of candidates is collected, feature extraction is carried out on the related information of the candidates, a digital twin model is constructed, an AI agent is deployed according to the features of each candidate, and behavior data of the candidates in a virtual working environment are obtained; and according to behavior data of the AI agent in the virtual working environment, a behavior mode graph of the candidates is established, the candidates are scored, and finally related information of the candidates, the behavior mode graph and scores are stored in a human resource data warehouse. Through the method provided by the invention, the enterprise can more accurately evaluate the true ability and the fitness of the candidates, so that a more intelligent recruitment decision is made, comprehensive data support is provided for the enterprise, and long-term tracking, trend analysis and strategic planning are facilitated.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and in particular to a method and system for constructing a human resources data warehouse based on big data. Background Art

[0002] As the informatization and digitalization of human resource management in enterprises continues to deepen, information systems in various professional fields are becoming increasingly complex, and the amount of business data generated is also growing explosively. Today, millions of human resource data are scattered in various independent information systems.

[0003] Existing human resource data warehouses are usually stored in a decentralized form, such as rosters, address books, and performance tables, which are distributed in multiple independent tables. When responding to specific business needs, relevant information must be dispatched and integrated from multiple tables, which not only leads to low work efficiency, but also makes it difficult to grasp the overall human resource situation. It is difficult for managers to clearly understand the organization's human resource structure, and it is not conducive to flexible expansion and optimization according to changing needs.

[0004] In order to improve efficiency and better support decision-making, enterprises need to build an integrated and intelligent human resources data platform. Such a platform can centrally manage various types of human resources data, provide real-time data insights, simplify information retrieval processes, and support more complex data analysis and application requirements. In this way, managers can understand human resources information more intuitively, make more informed decisions, and lay a solid foundation for future human resources application needs. Summary of the invention

[0005] The present invention describes a method and system for constructing a human resources data warehouse based on big data, which can solve the above technical problems.

[0006] According to a first aspect, a method for constructing a human resources data warehouse based on big data is provided. The method comprises:

[0007] Collecting relevant information of candidates and preprocessing the relevant information of the candidates;

[0008] Extracting features from the relevant information of the candidate to obtain the candidate's skills, experience, and personality characteristics;

[0009] Create a virtual work environment and integrate candidates’ skills, experience, and personality traits to build a digital twin model;

[0010] Deploy an AI agent based on the characteristics of each candidate based on a reinforcement learning algorithm, place the AI ​​agent in the virtual work environment, track the performance of the candidate in the virtual work environment over a long period of time, and record the behavior data of the AI ​​agent in the virtual work environment;

[0011] Build a behavioral pattern map of candidates based on the behavioral data of AI agents in a virtual work environment and score the candidates;

[0012] Store the candidate's relevant information, behavior pattern maps and scores in the human resources data warehouse.

[0013] In some embodiments, the method for preprocessing the relevant information of the candidate specifically includes:

[0014] Remove duplicates to ensure each candidate's information is unique;

[0015] For missing data, mean filling, median filling or KNN interpolation methods are used to fill them;

[0016] Convert categorical related information into numerical representation and standardize or normalize the numerical related information.

[0017] In some embodiments, the candidate's relevant information includes one or more of a resume, an interview record, and social media public information.

[0018] In some embodiments, the method for creating a virtual work environment specifically includes:

[0019] Define the virtual work environment based on the characteristics of the target industry and the requirements of specific positions;

[0020] Determine the types of tasks the candidate may need to complete;

[0021] Design specific scenarios for each task type, introducing randomness and uncertainty.

[0022] In some embodiments, a method for establishing a candidate's behavior pattern map based on the behavior data of an AI agent in a virtual work environment specifically includes:

[0023] Apply clustering algorithms to classify similar behavior patterns and identify different behavior groups;

[0024] Use association rule learning algorithms to discover associations between behaviors;

[0025] Define the nodes and edges of the behavior pattern graph based on the behavior data, and set attributes for each node and edge.

[0026] In some embodiments, a method for discovering association relationships between behaviors using an association rule learning algorithm specifically includes:

[0027] Convert candidate behavior sequences into transactional format;

[0028] Use the Apriori algorithm or FP-Growth algorithm to mine frequent item sets and find frequently occurring behavior combinations;

[0029] Filter meaningful association rules based on business needs;

[0030] Explain the discovered association rules and apply the association rules to real-world scenarios.

[0031] In some embodiments, a visualization tool is used to create an interactive chart that displays key information of the behavior pattern map.

[0032] According to a second aspect, a human resources data warehouse construction system based on big data is provided, the system comprising:

[0033] Data collection module, used to collect relevant information of candidates;

[0034] Data processing module, used to pre-process relevant information of candidates;

[0035] The feature extraction module is used to extract features from the candidate's relevant information to obtain the candidate's skills, experience, and personality characteristics;

[0036] A digital twin model building module is used to create a virtual work environment and integrate the skills, experience, and personality traits of candidates to build a digital twin model;

[0037] The work simulation module is used to deploy an AI agent according to the characteristics of each candidate based on a reinforcement learning algorithm, place the AI ​​agent in the virtual work environment, track the performance of the candidate in different situations over a long period of time, and record the behavioral data of the AI ​​agent in the virtual work environment.

[0038] The behavioral data processing module is used to build a behavioral pattern map of candidates based on the behavioral data of the AI ​​agent in the virtual work environment and score the candidates;

[0039] The data storage module is used to store relevant information, behavior pattern maps and scores of candidates.

[0040] The present invention provides a method and system for constructing a human resources data warehouse based on big data. By collecting relevant information of candidates, extracting features of the relevant information of the candidates, and constructing a digital twin model, an AI agent is deployed according to the characteristics of each candidate, and the behavioral data of the candidates in a virtual work environment is obtained. The behavioral pattern map of the candidates is established based on the behavioral data of the AI ​​agent in the virtual work environment, and the candidates are scored, and finally the relevant information of the candidates, the behavioral pattern map and the score are stored in the human resources data warehouse. With the method provided by the present invention, enterprises can more accurately evaluate the real ability and suitability of candidates, so as to make more intelligent recruitment decisions, and also provide comprehensive data support for enterprises, which is convenient for long-term tracking, trend analysis and strategic planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 A schematic diagram of a method for constructing a human resources data warehouse based on big data provided in an embodiment of this specification is shown;

[0043] Figure 2 A schematic diagram of a system process framework for building a human resources data warehouse based on big data provided in an embodiment of this specification is shown. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0045] In the description of the embodiments of the present application, words such as "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a concrete way.

[0046] In the description of the embodiments of the present application, the term "and / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, B exists alone, and A and B exist at the same time. In addition, unless otherwise specified, the term "plurality" means two or more.

[0047] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprises", "has" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0048] Figure 1 A flowchart of a method for constructing a human resources data warehouse based on big data provided by an embodiment of this specification is shown as follows: Figure 1 As shown, the method comprises the following steps:

[0049] Step S101, collecting relevant information of the candidate and preprocessing the relevant information of the candidate.

[0050] In one embodiment, the candidate's relevant information includes but is not limited to one or more of a resume, an interview record, and public information on social media.

[0051] Specifically, with regard to the acquisition of resume-related information, optical character recognition (OCR) technology can be used to automatically extract text from resumes in PDF or image format. Use pre-trained language models (such as BERT) to understand and structure the content of the resume, including educational background, work experience, skill certificates, etc. With regard to the acquisition of interview records, a structured interview record form can be designed, covering key evaluation indicators (such as communication skills, professional skills, teamwork, etc.), and recorded through a spreadsheet tool. Speech recognition technology can be used to convert interview recordings or videos into text for easy retrieval and analysis, or sentiment analysis tools can be combined to evaluate the candidate's emotional state and expression during the interview process, providing a more comprehensive evaluation perspective. With regard to the acquisition of public information on social media, the official API interface provided by the social platform can be used to legally and compliantly obtain the candidate's public information, such as work experience, project contributions, personal blogs, etc. Under the premise of complying with the terms of use of each platform, you can consider using Web Scraping tools (such as BeautifulSoup, Scrapy) to capture some public information. It can also include relevant information of other candidates, and this application does not limit this.

[0052] In one embodiment, the method for preprocessing the relevant information of the candidate specifically includes:

[0053] Remove duplicates to ensure the uniqueness of each candidate's information. For missing data, use mean filling, median filling or KNN interpolation methods to fill in. Convert categorical related information into numerical representation and standardize or normalize the numerical related information.

[0054] Specifically, ensure that each candidate has a unique identifier (such as ID number, email address) to prevent duplicate records. For cases without unique identifiers, use string similarity algorithms (such as Levenshtein distance, Jaccard similarity) to detect possible duplicate records and merge them manually or automatically. Use specialized data cleaning tools (such as OpenRefine) to identify and delete duplicate entries. Further confirm and clean potential duplicate records through logical checks (such as the same name, contact information appears multiple times in different time periods). Data visualization tools (such as Matplotlib, Seaborn) can be used to draw heat maps or histograms to intuitively display the distribution of missing values ​​in the data, and fill them with mean filling, median filling or KNN interpolation methods. Convert categorical related information into numerical representations, such as: converting categorical features into integer values, which is suitable for ordered categories (such as education level: high school, undergraduate, master's, doctoral), converting categorical features into binary vectors, which is suitable for unordered categories (such as gender, position type), encoding categorical features based on the statistical information of the target variable (such as mean, frequency), which is suitable for high cardinality categories (such as city names). All numerical information in the data (including information converted from categorical to numerical) can be converted to a standard normal distribution with a mean of 0 and a standard deviation of 1, or scaled to the interval [0,1].

[0055] Step S102: extracting features from the candidate's relevant information to obtain the candidate's skills, experience, and personality characteristics.

[0056] Specifically, the relevant information of the candidate can be analyzed, and its characteristics can be further extracted to obtain the candidate's skills, experience, and personality characteristics.

[0057] About skill feature extraction. For example, you can analyze resumes: use natural language processing (NLP) technology to identify skill keywords in resumes, such as programming languages, development tools, frameworks, etc., through pre-trained language models (such as BERT) or custom dictionaries. Communication skills, teamwork, leadership, etc. Classify the identified skills into different fields (such as technical skills, management skills), and calculate the weight or frequency of each field. Analyze interview records: For structured questions designed in the interview, you can directly extract the candidate's answers, especially those related to skills. Obtain examples of candidates' skill application in specific situations through behavioral event interviews, and combine sentiment analysis tools to evaluate candidates' attitudes and confidence when discussing skills. Analyze public information on social media: For technical positions, you can obtain candidates' code contribution records from platforms such as GitHub, analyze the open source projects and technical fields they are involved in, and check certifications and recommendations on LinkedIn to verify the skills claimed by candidates.

[0058] About experience feature extraction. By analyzing their resume, you can extract the candidate's past positions, company names, working time range, and calculate the total years of work experience. You can extract key tasks and achievements from the job description to evaluate the depth and breadth of the candidate's actual work experience. Analyze the interview records and evaluate the candidate's problem-solving ability and experience level in complex situations through case analysis questions in the interview. Ask candidates to share their past work experiences, paying special attention to how they cope with challenges and achieve success. Analyze public information on social media and analyze the candidate's career development path on LinkedIn, including job promotion, industry transition, etc. Check the projects displayed by the candidate on personal blogs and portfolio websites to understand their actual operational experience.

[0059] Regarding the extraction of personality traits. Analyze the resume. Although the self-evaluation in the resume may be subjective, it can still be used as a preliminary reference, especially the candidate's description of his or her own personality. Analyze the interview records. You can use the behavioral interview method to ask situational questions, observe the candidate's response, evaluate his or her personality traits, and score the candidate based on his or her answers against the Big Five personality model (openness, conscientiousness, extroversion, agreeableness, and neuroticism). Pay attention to the candidate's body language, voice changes and other non-verbal signals to assist in judging personality traits. Analyze public information on social media, analyze the candidate's interaction patterns on social media, such as comment style, topic preferences, etc., and infer his or her personality traits. Analyze the candidate's interaction patterns on social media, such as comment style, topic preferences, etc., and infer his or her personality traits. Examine the candidate's activity and influence in professional communities, such as the frequency and quality of speeches in technical forums.

[0060] Furthermore, after the initial feature extraction, the features can be quantified and integrated. For example, a scoring system can be established for each feature dimension (skills, experience, personality) to ensure that information from different sources can be compared and comprehensively evaluated. According to the requirements of different positions, the weights of each feature dimension can be adjusted to make the final score more in line with specific needs.

[0061] Step S103, creating a virtual work environment and integrating the skills, experience, and personality traits of the candidates to build a digital twin model.

[0062] Specifically, you can choose to use a dedicated simulation platform to create a virtual work environment, such as Unity or UnrealEngine. You can integrate the skills, experience, and personality characteristics of the candidates extracted in the previous step to build a relationship graph containing their skills, experience, and personality characteristics, and use a graph database to store the candidate's feature relationship graph to build a digital twin model.

[0063] In one embodiment, the method for creating a virtual work environment specifically includes: defining a virtual work environment according to the characteristics of a target industry and the requirements of a specific position; determining the types of tasks that a candidate may need to complete; and designing specific scenarios for each task type, introducing randomness and uncertainty.

[0064] Specifically, gain an in-depth understanding of the characteristics of the target industry and the requirements of specific positions, including industry standards, technology trends, core skills, soft skills, and key performance indicators. Design a series of tasks based on the job requirements, covering daily tasks, emergencies, and long-term projects, to ensure comprehensive coverage of various scenarios that candidates may face. Simulate the tasks that candidates need to complete on a daily basis, such as writing code, writing reports, handling customer requests, etc. Create different work scenarios, including normal workflows, emergency response, teamwork, and customer interaction. Provide tools and resources that candidates may use, such as development environments, CRM systems, data analysis software, etc., to ensure that they can operate in a realistic environment.

[0065] Step S104, deploying an AI agent according to the characteristics of each candidate based on a reinforcement learning algorithm, placing the AI ​​agent in the virtual work environment, tracking the performance of the candidate in the virtual work environment for a long time, and recording the behavior data of the AI ​​agent in the virtual work environment.

[0066] Specifically, first, determine the specific goals for each position, such as task completion time, quality standards, customer satisfaction, etc. Set a reasonable reward function based on the goals. The reward function should be able to reflect the performance of candidates in different situations, such as giving positive rewards for successfully completing tasks and giving negative feedback for errors or inefficient behaviors. Second, select a suitable reinforcement learning algorithm (such as DQN, PPO, TRPO) to create an initial state AI agent for each candidate. This initial state can be personalized according to the candidate's skills, experience, and personality traits. Convert the candidate's skill characteristics (such as programming language, tool proficiency) into the ability parameters of the AI ​​agent. These parameters can affect the success rate of the AI ​​agent in performing specific tasks in a virtual environment. Adjust the learning rate and decision-making ability of the AI ​​agent based on the candidate's years of work experience and past project experience. Experienced candidates may have faster learning speeds and better problem-solving skills. Use the candidate's personality traits (such as responsibility, extroversion) to influence the behavior pattern of the AI ​​agent. For example, candidates with high responsibility may pay more attention to details and task accuracy.

[0067] After that, simulate common tasks in the candidate's daily work, such as writing code, writing reports, handling customer requests, etc. Introduce some unexpected events, such as system failures, changes in customer requirements, etc., to test the candidate's adaptability and problem-solving ability. You can create a multi-agent system (MAS) to simulate the interaction and collaboration between team members to evaluate the candidate's communication and teamwork skills. You can gradually increase the difficulty of the task according to the requirements of different stages to observe whether the performance of the AI ​​agent improves with the accumulation of experience. You can also introduce changes in external factors, such as market trends, policies and regulations, to examine the adaptability and flexibility of candidates under different environmental conditions.

[0068] Finally, record every action of the AI ​​agent in the virtual work environment in detail, including the type of task selected, the length of time to execute, the problems encountered and the solutions, etc. Record the process of the AI ​​agent making decisions, including the factors considered, the criteria for weighing, and the final choice. This helps to analyze its thinking style and decision-making logic. A monitoring system can be established to track the performance of the AI ​​agent in real time to ensure that its behavior meets the expected goals. If abnormal behavior is found, adjust the reward function or retrain the model in time. Conduct a comprehensive evaluation of the AI ​​agent at regular intervals, compare the changes in its performance before and after, and analyze which aspects have been improved and which areas need to be strengthened. The behavior patterns of AI agents can be identified through reinforcement learning algorithms, and common paths and abnormal behaviors can be found, which can help understand the typical working methods of candidates and their potential strengths and weaknesses.

[0069] For example, suppose you create a digital twin model for a software engineer: first, extract the candidate's skills, experience, and personality traits based on his or her resume, interview records, and social media profile to build a personalized AI agent. Select a reinforcement learning algorithm suitable for software development scenarios, such as PPO, to train the AI ​​agent. Then, simulate daily work such as writing code and debugging programs. Introduce emergencies such as system crashes and emergency repairs to test the candidate's ability to respond. Create a multi-agent system to simulate cooperation with other developers and evaluate communication and coordination capabilities. Record every operation of the AI ​​agent in the virtual environment in detail, including coding efficiency, error rate, etc. Record the AI ​​agent's thinking process and solutions when facing complex problems, and analyze its decision-making logic. Through data analysis, identify the candidate's behavior patterns when handling specific types of tasks. Compare the candidate's performance before and after the stage, evaluate his or her progress, and optimize the model based on feedback.

[0070] By simulating real workplace situations, it is possible to more comprehensively and accurately assess candidates' actual work capabilities and suitability, going beyond the limitations of traditional resumes and interviews. By observing candidates' performance in different situations, we can better predict their development potential and adaptability in future positions and reduce recruitment risks. Automated behavioral data analysis reduces the time and subjectivity of manual evaluation, and improves the efficiency and objectivity of recruitment and other human resource decisions. By tracking and analyzing candidates' behavioral data over a long period of time, companies can also continuously optimize evaluation models and scoring criteria to ensure that they always meet the latest business needs and technology development trends.

[0071] Step S105, establishing a behavioral pattern map of the candidate based on the behavioral data of the AI ​​agent in the virtual work environment, and scoring the candidate.

[0072] Mathematical tools can be used to group similar behavior patterns together, discover the typical behavior paths of candidates in different situations, find the correlation between different behaviors, reveal the possible action combinations that candidates may take in specific situations, and build the nodes, edges, and attributes of the behavior pattern map based on the basic information of the candidates. Finally, the candidates are scored based on their performance and recorded.

[0073] In one embodiment, a method for establishing a candidate's behavior pattern map based on the behavior data of an AI agent in a virtual work environment specifically includes: applying a clustering algorithm to classify similar behavior patterns and identify different behavior groups. Using an association rule learning algorithm to discover the association relationship between behaviors. Defining nodes and edges of the behavior pattern map based on the behavior data, and setting attributes for each node and edge.

[0074] Specifically, you can choose algorithms such as K-means or hierarchical clustering to group similar behavior patterns of candidates into one category and identify different behavior groups. Through association rule mining (such as the Apriori algorithm), find the correlation between different behaviors and reveal the possible combination of actions that candidates may take in specific situations. Extract representative key behaviors from a large amount of behavioral data, which can reflect the core capabilities and characteristics of candidates. Analyze the changing trends of behavioral data over time to evaluate the learning curve and growth potential of candidates. Take each behavior of the candidate as a node and the relationship between behaviors (such as cause and effect, sequence) as an edge to construct a behavioral network map. Add attribute labels to each node and edge, such as behavior type, occurrence time, influencing factors, etc., to describe the behavioral characteristics in more detail.

[0075] In one embodiment, a method for discovering associations between behaviors using an association rule learning algorithm specifically includes: converting a candidate's behavior sequence into a transaction format. Using an Apriori algorithm or an FP-Growth algorithm to mine frequent item sets to find frequently occurring behavior combinations. Screening meaningful association rules based on business requirements. Explaining the discovered association rules and applying the association rules to actual scenarios.

[0076] Specifically, first, extract specific behavioral actions from the behavior log of the AI ​​agent in the virtual work environment. These behaviors can be task completion, error handling, communication, etc. Add a timestamp to each behavior to record the specific time and order of the behavior.

[0077] Divide the candidate's behavior sequence in a specific time period into multiple "transactions". For example, you can take one day's work time as the unit and treat all behaviors in one day as a transaction. Encode each behavior and assign a unique identifier. Represent each transaction as a set that contains all behaviors that occurred in that time period. For example, transaction T1 = {encoding 1, encoding 2, encoding 3}. Use the Apriori algorithm or FP-Growth algorithm to mine frequent item sets.

[0078] Set a minimum support threshold based on business requirements. A behavior combination is considered a frequent item set only when it occurs more than this threshold. For the Apriori algorithm, gradually generate larger candidate item sets and check their frequencies by scanning the database; for the FP-Growth algorithm, extract frequent item sets directly from the FP-tree. Finally, a set of frequent item sets is obtained, each of which represents a frequently occurring behavior combination. For each frequent item set, calculate the confidence of its association rule. Confidence refers to the probability that when an item A appears, another item B also appears. Set a minimum confidence threshold based on business requirements and only retain association rules with confidence higher than this threshold. Evaluate the relevance of each association rule to business requirements, sort the association rules according to their importance or impact, and ensure that the most relevant rules are ranked first.

[0079] Convert the discovered association rules into easy-to-understand language descriptions. For example, "If candidates often write unit tests, they have a lower error rate when debugging code." Try to analyze the reasons behind the association rules and explore why certain combinations of behaviors appear frequently. Find out the association rules of those candidates who perform well and summarize their success stories. For example, "Efficient candidates usually conduct detailed requirements analysis at the beginning of the project, which reduces the possibility of rework later." Identify the association rules of those candidates who perform poorly and analyze the problems they encounter and their root causes. For example, "Candidates who frequently modify code but lack documentation are prone to difficulties in team collaboration." Finally, apply the discovered association rules to actual scenarios.

[0080] Step S106, storing the candidate's relevant information, behavior pattern map and score in the human resources data warehouse.

[0081] You can use relational databases (such as MySQL, PostgreSQL) to store structured data, such as candidate information, scores, etc. Use graph databases (such as Neo4j, Amazon Neptune) to store behavioral pattern graphs and manage complex relationship networks. In addition, you can establish a regular update mechanism to ensure that the information in the data warehouse is always up to date. For example, synchronize data once every night to update the latest behavioral records and scores. Implement strict access rights management to allow only authorized personnel to view and process sensitive information. For example, the HR department can access the information of all candidates, while other departments can only access part of the information related to them.

[0082] In one embodiment, a visualization tool is used to create an interactive chart that displays key information of the behavior pattern map.

[0083] Specifically, you can use visualization tools such as Tableau, Power BI, Looker, D3.js, etc. to create interactive charts. For example, first determine the key information and the core indicators that need to be displayed, such as behavior frequency, task completion rate, error rate, communication efficiency, etc. Analyze the changing trend of behavior over time and evaluate the development potential of candidates. Show the association rules between the discovered behaviors and reveal hidden behavioral patterns. Secondly, choose the appropriate chart type, including but not limited to maps, heat maps, scatter plots, and line charts. Different charts and information modules can be laid out reasonably. For example, a filter is placed on the left, the main chart is displayed on the right, and a detailed data table is displayed at the bottom. Dynamic links can be established between multiple charts. When an element is selected in a chart, other related charts will be automatically updated to display the corresponding data.

[0084] Corresponding to the above method provided by the present invention, the present invention also provides a system. Figure 2 The following is a schematic diagram showing a process framework of a human resources data warehouse construction system based on big data provided by an embodiment of this specification. Figure 2 As shown, the system includes:

[0085] The data collection module 201 is used to collect relevant information of candidates.

[0086] The data processing module 202 is used to pre-process the relevant information of the candidates.

[0087] The feature extraction module 203 is used to extract features from the candidate's relevant information to obtain the candidate's skills, experience, and personality features.

[0088] The digital twin model construction module 204 is used to create a virtual work environment and integrate the skills, experience, and personality traits of the candidates to build a digital twin model.

[0089] The work simulation module 205 is used to deploy an AI agent according to the characteristics of each candidate based on a reinforcement learning algorithm, place the AI ​​agent in the virtual work environment, track the performance of the candidate in different situations over a long period of time, and record the behavior data of the AI ​​agent in the virtual work environment.

[0090] The behavior data processing module 206 is used to establish a behavior pattern map of the candidate based on the behavior data of the AI ​​agent in the virtual work environment and score the candidate.

[0091] The data storage module 207 is used to store the candidate's relevant information, behavior pattern graphs and scores.

[0092] It is worth mentioning that Figure 2 The system shown is Figure 1Corresponding to the method embodiments shown, those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0093] According to another embodiment of the present invention, a computing device is provided, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the Figure 1 The method described.

[0094] According to another embodiment, there is also provided a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute a combination of Figure 1 The method described.

[0095] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present invention can be implemented by hardware, software, firmware or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0096] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solution of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for constructing a human resources data warehouse based on big data, characterized in that: The method comprises: Collecting relevant information of candidates and preprocessing the relevant information of the candidates; Extracting features from the relevant information of the candidate to obtain the candidate's skills, experience, and personality characteristics; Create a virtual work environment and integrate candidates’ skills, experience, and personality traits to build a digital twin model; Deploy an AI agent based on the characteristics of each candidate based on a reinforcement learning algorithm, place the AI ​​agent in the virtual work environment, track the performance of the candidate in the virtual work environment over a long period of time, and record the behavior data of the AI ​​agent in the virtual work environment; Build a behavioral pattern map of candidates based on the behavioral data of AI agents in a virtual work environment and score the candidates; Store the candidate's relevant information, behavior pattern maps and scores in the human resources data warehouse.

2. The method according to claim 1, characterized in that The method for preprocessing the relevant information of the candidate specifically includes: Remove duplicates to ensure each candidate's information is unique; For missing data, mean filling, median filling or KNN interpolation methods are used to fill them; Convert categorical related information into numerical representation and standardize or normalize the numerical related information.

3. The method according to claim 1, characterized in that The candidate's relevant information includes one or more of a resume, interview records, and public information on social media.

4. The method according to claim 1, characterized in that: Methods for creating a virtual work environment include: Define the virtual work environment based on the characteristics of the target industry and the requirements of specific positions; Determine the types of tasks the candidate may need to complete; Design specific scenarios for each task type, introducing randomness and uncertainty.

5. The method according to claim 1, characterized in that: The method of establishing a behavioral pattern map of a candidate based on the behavioral data of an AI agent in a virtual work environment specifically includes: Apply clustering algorithms to classify similar behavior patterns and identify different behavior groups; Use association rule learning algorithms to discover associations between behaviors; Define the nodes and edges of the behavior pattern graph based on the behavior data, and set attributes for each node and edge.

6. The method according to claim 5, characterized in that The method of using association rule learning algorithm to discover the association relationship between behaviors includes: Convert candidate behavior sequences into transactional format; Use the Apriori algorithm or FP-Growth algorithm to mine frequent item sets and find frequently occurring behavior combinations; Filter meaningful association rules based on business needs; Explain the discovered association rules and apply the association rules to real-world scenarios.

7. The method according to claim 1, characterized in that Use visualization tools to create interactive charts that display key information about behavioral pattern maps.

8. A human resources data warehouse construction system based on big data, characterized in that: The data warehouse is constructed by the method according to claim 1, wherein the system comprises: Data collection module, used to collect relevant information of candidates; Data processing module, used to pre-process relevant information of candidates; The model extraction module is used to extract features from the candidate's relevant information to obtain the candidate's skills, experience, and personality characteristics; A digital twin model building module is used to create a virtual work environment and integrate the skills, experience, and personality traits of candidates to build a digital twin model; The work simulation module is used to deploy an AI agent according to the characteristics of each candidate based on a reinforcement learning algorithm, place the AI ​​agent in the virtual work environment, track the performance of the candidate in different situations over a long period of time, and record the behavioral data of the AI ​​agent in the virtual work environment. The behavioral data processing module is used to build a behavioral pattern map of candidates based on the behavioral data of the AI ​​agent in the virtual work environment and score the candidates; The data storage module is used to store relevant information, behavior pattern maps and scores of candidates.

Images