Talent evaluation system and method based on multi-source data fusion and intelligent analysis

Abstract

The invention relates to a talent evaluation system and method based on multi-source data fusion and intelligent analysis, and belongs to the crossing field of information technology and human resource management. The technical problems to be solved are that in existing talent evaluation, multi-source heterogeneous data are difficult to fuse, an evaluation model is static and rigid, a decision is lack of data support, and closed-loop self-optimization cannot be realized. According to the technical scheme, the method is characterized in that multi-source heterogeneous data such as resumes, interview scores and background investigation in the whole recruitment process are cleaned and standardized through a data collection and structuring step, and the multi-source heterogeneous data are converted into unified structured data; based on the structured data, double quantization calculation of the flow stage quantization model and the capability dimension quantization model is executed in parallel, and flow stage linear scores and multi-dimensional capability radar map scores of candidates are generated respectively; fusing the double-model scores, and generating a comprehensive evaluation result and a visual report according to a preset decision rule; by collecting performance data of recorded employees and utilizing a machine learning algorithm to analyze relevance between historical evaluation and performance, weight parameters of the model are dynamically adjusted, and feedback and self-optimization of the system are achieved. The method is mainly used for enterprise recruitment, talent selection and performance prediction, and the objectivity, accuracy and decision-making efficiency of evaluation can be improved.

Description

Technical Field

[0001] This invention relates to the intersection of information technology and human resource management, specifically a talent assessment system and method based on multi-source data fusion and intelligent analysis, applicable to scenarios such as corporate recruitment, talent selection, and performance prediction. Background Technology

[0002] Current talent assessment primarily relies on two types of technologies: First, there are relational database-based recruitment management systems (ATS) whose core function is process management (such as resume screening and interview scheduling). However, their evaluation modules only support simple form scoring, resulting in isolated data and a lack of in-depth analysis capabilities. Second, there are online psychological assessment tools that, while capable of generating standardized reports, suffer from rigid models that are difficult to integrate with recruitment process data. Existing technologies have the following shortcomings: 1. Significant data heterogeneity: Data from multiple sources, such as resumes, interview scores, and background checks, are in different formats and cannot be shared between systems, creating information silos; 2. Static evaluation model: The scoring dimensions and online assessment algorithms of ATS are fixed, and the weights cannot be dynamically adjusted according to job requirements; 3. Weak decision support: Lack of cross-stage correlation analysis, making it impossible to mine the correlation between evaluation indicators and future performance through historical data; 4. Lack of closed-loop learning: The system does not incorporate performance data feedback after employees join the company, making it impossible to optimize the evaluation model. Summary of the Invention

[0003] I. Technical Problems to be Solved This invention breaks through the bottlenecks of existing technologies and solves the following problems: it realizes the unified integration and structured processing of multi-source heterogeneous data in the entire recruitment process; it provides a dynamically configurable quantitative evaluation model that supports adaptive adjustment according to job requirements; and it provides predictive support for decision-making through data mining and visualization analysis.

[0004] II. Technical Solution The core architecture of this system includes a data acquisition layer, a business logic layer, a storage layer, and an intelligent service layer. Key innovations are as follows: 1. Multi-source data fusion technology: Natural language processing (NLP) is used to parse unstructured data (such as interview comments and background investigation texts), extract keywords and transform them into structured indicators; a unified data model with candidate ID as the primary key is constructed to link heterogeneous data such as resumes, ratings, and assignments.

[0005] 2. Dual Quantization Dynamic Model: Process Stage Quantification Model: Define N consecutive recruitment stages (e.g., initial interview, final interview, background check), calculate the scores for each stage, and then sum them by weight to generate a linear process scoring chart. The formula is: ; Ability Dimension Quantification Model: Pre-sets "external training" (e.g., professional skills) and "internal cultivation" (e.g., intrinsic motivation) dimension sets, aggregates cross-stage scores for each dimension to generate a radar chart, and the scoring formula for each dimension is: .

[0006] 3. Closed-loop self-optimization mechanism: The system regularly collects the performance data (such as KPIs) of recruited employees, analyzes the correlation between historical scores and performance through machine learning algorithms (such as linear regression), and automatically adjusts the model weight parameters.

[0007] III. Beneficial Effects 1. Technical effects: By using multi-source data fusion technology, information silos are broken down to form a panoramic view of candidates; The dual-model dynamic configuration makes the evaluation results more closely match the actual needs of the job, thus improving accuracy; Data-driven predictive analytics reduces subjective bias and significantly enhances the scientific nature of decision-making.

[0008] 2. Commercial Value: Improved recruitment efficiency: Automated generation of visual reports shortens the decision-making cycle; Reduced risk of misrecording: Deeply explore candidates' potential traits through a radar chart of their capabilities; Accumulate corporate talent data assets to support talent pipeline development. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the overall system architecture, illustrating the interaction between the data acquisition, processing, storage, and service layers; Figure 2 The flowchart for calculating the dual quantification model includes the steps of process stage scoring and capability dimension aggregation; Figure 3 A visual comparison example of candidate competence radar charts and process linear scoring. Detailed Implementation

[0010] The present invention will be further described in detail below with reference to embodiments. These embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0011] Example 1: A talent assessment system and method based on multi-source data fusion and intelligent analysis This embodiment details the specific application process of the method of the present invention in a corporate recruitment scenario.

[0012] 1. Data Acquisition and Structure Steps The system collects multi-source heterogeneous data and performs cleaning and standardization in the following ways: Resume Data Collection and Structuring: The system receives electronic resumes (such as PDF and Word formats) submitted by candidates through the web front end. The built-in data parsing module automatically parses the resume file, extracts key fields, including but not limited to name, university, education, major, work experience (employer, service time, position, main responsibilities), etc., and converts them into unified JSON format structured data.

[0013] Interview scoring data collection: At each stage of the interview process, including phone invitations, initial interviews, second interviews, and final interviews, interviewers use a mobile app or web interface to score candidates based on a pre-set "Structured Question Collection Library" and "Scoring Details Reference Standards." For example, in the "Initial Interview," interviewers need to score candidates on a 5-point scale for dimensions such as "Job-Related Professional Abilities" and "Analytical and Decision-Making Abilities." All scoring data is submitted to the system in real time via an API.

[0014] Background Check Data Collection: The human resources department uses the "Applicant Background Check Form" generated by the system to conduct background checks. After the investigation is completed, the structured information in the form (such as employer, reason for leaving, background checker evaluation, etc.) is entered into the system. For the text comments (such as "outstanding abilities" and "areas for improvement"), the system will activate the natural language processing module to extract keywords (e.g., extract keywords such as "leadership" and "communication skills") and perform sentiment analysis (converting the comments into positive, negative, or neutral sentiment scores) as a supplement to the quantitative score.

[0015] Data standardization: The system maps scores from different sources and scales (such as 10-point scale and 5-point scale) to a preset scale of 0-100 and stores them in the database's data storage layer.

[0016] 2. Calculation steps of the dual quantization model The model calculation engine in the data processing and business logic layer executes the following two quantization model calculations in parallel: The system predefines N=7 consecutive recruitment stages: resume screening (10 points), telephone invitation (10 points), initial interview (10 points), written test (10 points), final interview (10 points), background check (10 points), and probation period (10 points). Each stage has its own weight (for example, the final interview has the highest weight, set at 0.25; the resume screening has a lower weight, set at 0.1).

[0017] According to the formula Calculate the candidate's overall process score S process。

[0018] The system generates a visual process scoring chart, which intuitively displays the candidate's scores at each stage.

[0019] The system predefines M capability dimensions, which are divided into an external training dimension set and an internal cultivation dimension set. The external training dimension set includes: job professional skills, analytical and decision-making skills, planning and execution skills, teamwork skills, and self-management skills. The internal cultivation dimension set includes: comprehension skills, intrinsic motivation, perseverance, honesty, and integrity.

[0020] For each capability dimension j, the system uses the formula... Calculate the weighted average score of the ability; for example, if "analytical decision-making ability" is scored in the preliminary test (weight 0.2), the second test (weight 0.2) and the final test (weight 0.4), then calculate its final score according to the weight.

[0021] The system generates a multi-dimensional capability radar chart, clearly showing the candidate's strengths and weaknesses in each capability dimension.

[0022] 3. Data Fusion and Comprehensive Evaluation Steps The data fusion module will use the total process score S calculated above. process and capability dimension radar chart score S ability(j) The system will then integrate the data. The system's preset decision rules are as follows: When the total process score is S process The score is greater than the threshold T1 (e.g., 70 points), and the score S of key competency dimensions (such as "job professional skills" and "integrity") is greater than the threshold T1. ability(j) When all scores are greater than the threshold T2 (e.g., 80 points), the system automatically marks the candidate as "recommended for recruitment".

[0023] The system generates a comprehensive evaluation report that includes quantitative scores, visual charts, and final recommendations.

[0024] 4. Model Feedback and Self-Optimization Steps The system's intelligent service layer continuously collects performance data of hired employees (such as KPI completion and supervisor evaluations). Using machine learning algorithms (e.g., linear regression or random forest), it analyzes the correlation between historical evaluation data (i.e., scores at each stage and scores for each competency dimension) and performance data. Based on the analysis results, it dynamically adjusts the weight parameters in the dual-quantification model (e.g., adjusting the weights of each recruitment stage and each competency dimension) to ensure the evaluation model continuously approximates actual successful hiring standards, achieving self-optimization.

[0025] Example 2: A talent assessment system implementing the method described in Example 1 This embodiment describes the system architecture for implementing the above method. The system adopts a layered architecture, including: Data collection and interaction layer: Provides web front-end and mobile APP interfaces for candidates to submit resumes, interviewers to enter scores, and HR to enter background check information.

[0026] Data processing and business logic layer (core layer): includes: Data parsing module: used to parse resume files and extract structured information; Natural Language Processing Module: Used to process text comments, extract keywords, and perform sentiment analysis; Model calculation engine: used to calculate the quantitative model and capability dimension quantitative model for the execution process stage; Data fusion module: used to perform data fusion and comprehensive evaluation steps.

[0027] Data storage layer: Relational databases and NoSQL databases are used to store processed structured data (such as candidate information and rating records) and unstructured data (such as original resumes and text comments), respectively.

[0028] Intelligent Service Layer: Provides reusable algorithm services in the form of APIs, including weight calculation algorithms, similarity matching algorithms, etc., for the business logic layer to call.

[0029] Example 3: An electronic device The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements any of the methods described in Embodiment 1.

[0030] Example 4: A computer-readable storage medium The computer-readable storage medium (such as an SSD, USB flash drive, or optical disc) stores a computer program thereon; when the program is executed by a processor, it implements any of the methods described in Embodiment 1.

[0031] The above embodiments are for illustrative purposes only and are not intended to limit the invention. In practical applications, the model parameters and data sources can be adjusted according to the needs of the enterprise.

Claims

1. A talent assessment method for the recruitment process based on multi-source data fusion, characterized in that, Includes the following steps: Data collection and structuring steps: Collect multi-source heterogeneous data from the recruitment process, including resume data, interview score data, written test data, and background investigation data; clean and standardize the multi-source heterogeneous data, transforming it into unified structured data and storing it; Dual quantification model calculation steps: Based on the structured data, perform parallel calculations of the process stage quantification model and the ability dimension quantification model; wherein, the process stage quantification model calculation is used to generate linear scores for candidates at each process stage, and the ability dimension quantification model calculation is used to generate multi-dimensional ability radar chart scores for candidates; Data fusion and comprehensive evaluation steps: Integrate the linear scores of the process stages with the multi-dimensional capability radar chart scores, and generate a comprehensive evaluation result according to preset decision rules; Model feedback and self-optimization steps: Collect the performance data of hired employees, analyze the correlation between historical evaluation data and performance data through machine learning algorithms, and dynamically adjust the weight parameters of the dual quantification model.

2. The method according to claim 1, characterized in that, The quantitative model calculation for the process stages includes: defining multiple consecutive recruitment stages, each with a full score and weight; and calculating according to the formula. ， Overall score of the calculation process S process， It also generates a visual process score chart.

3. The method according to claim 1, characterized in that, The capability dimension quantification model calculation includes: predefining multiple capability dimensions, which include an external training dimension set and an internal cultivation dimension set; according to the formula... Calculate the weighted average score S for each ability dimension. ability(j), And generate the multi-dimensional capability radar map.

4. The method according to claim 3, characterized in that, The external training dimension set includes job-related professional skills, analytical and decision-making abilities, planning and execution abilities, teamwork abilities, and self-management abilities; the internal training dimension set includes comprehension abilities, intrinsic motivation, perseverance, honesty, and integrity.

5. The method according to claim 1, characterized in that, In the data fusion and comprehensive evaluation step, the preset decision rule includes: when the total process score S process A score S greater than the threshold T1 and at least one predefined key capability dimension ability(j) When the value exceeds the threshold T2, the candidate is automatically marked as recommended for employment.

6. A full-process talent assessment system for implementing the method as described in any one of claims 1 to 5, characterized in that, The system adopts a layered architecture, including: a data acquisition and interaction layer, used to collect raw data input from multiple roles through a web front-end and mobile APP interface; a data processing and business logic layer, which is the core layer, including a data cleaning module, a standardization module, and a model calculation engine for performing the dual quantization model calculation; a data storage layer, used to store processed structured data and unstructured raw data; and an intelligent service layer, which provides reusable algorithm service interfaces, including weight calculation algorithms and model self-optimization algorithms.

7. The system according to claim 6, characterized in that, The data processing and business logic layer further includes: a data parsing module for parsing resume files and extracting structured information; a natural language processing module for processing text comments and background survey texts and performing keyword extraction and sentiment analysis; and a data fusion module for executing the data fusion and comprehensive evaluation steps and generating a visualization report.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 5.

Images