Intelligent recommendation method for key post talents of enterprises based on knowledge graph

Abstract

The present application relates to a kind of enterprise key post talent intelligent recommendation method based on knowledge graph.The method includes: the personality characteristics of employee is evaluated based on Big Five Personality Theory.According to the personality characteristics and key post characteristics, a double-index portrait system is constructed.Relationship samples between personality characteristics and enterprise competition are obtained, clustering algorithm is used to cluster the relationship samples, and key post game mechanism is generated.A large language model is used to analyze resume information according to the double-index portrait system, and a potential key information corpus is obtained.The potential key information corpus is used as knowledge graph data, and the knowledge graph data is path-matched according to the key post game mechanism, and a key post talent recommendation scheme is output.This method can improve the accuracy and efficiency of key post talent recommendation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent human resource planning, and in particular to an enterprise key position talent intelligent recommendation method based on a knowledge graph. BACKGROUND

[0002] Under the wave of digital transformation, large enterprises are facing the dilemma of "information overload but inefficient talent decision-making": it is difficult to accurately match the personnel in charge with the needs of the post, and the traditional selection method relying on HR experience cannot meet the urgent needs of key positions in sudden projects, and at the same time, it rarely considers the confrontation game between peer enterprises. The technical method aims to realize scientific and efficient key position talent recommendation during the appointment of company executives or branch general managers, and provide quantitative decision support for human resource departments, which is of great significance for optimizing enterprise talent resource allocation, maximizing human capital efficiency, and improving enterprise core competitiveness.

[0003] Current personnel and post portraits have been applied to human resource management of local enterprises, but the key factor of management personnel personality traits has not been considered in the setting of portrait indicators. The method will focus on solving this problem, designing a management personnel personality trait model based on the Big Five personality, thereby improving the comprehensiveness and accuracy of personnel portraits and building a new indicator system for portrait models.

[0004] The theory of person-job matching originated from the concept of person-organization matching. Foreign researchers have analyzed the importance of person-job matching and continuously enriched the person-job matching model using various means, such as matching models based on career ability models and psychological measurement. However, there are few studies on person-job matching based on knowledge graph technology in published literature. The problem studied by this method is to apply intelligent recommendation algorithms based on knowledge graph to the field of human resource management planning in enterprise management, hoping to use the powerful processing capability of knowledge graph in handling unstructured data and clear visualization effect to contribute to improving the enterprise's ability in resource allocation, decision support, and other aspects, and help promote the development of enterprise informatization and intelligentization. SUMMARY

[0005] Therefore, it is necessary to provide an enterprise key position talent intelligent recommendation method based on a knowledge graph to improve the intelligence and efficiency of enterprise human resource allocation.

[0006] An enterprise key position talent intelligent recommendation method based on a knowledge graph, the method comprising:

[0007] Evaluating the personality characteristics of employees based on the Big Five personality theory.

[0008] Constructing a double-indicator portrait system according to the personality characteristics and key position characteristics.

[0009] Obtain a relationship sample between personality characteristics and enterprise competition, cluster the relationship sample by using a clustering algorithm, and generate a key post game mechanism.

[0010] Use a large language model to analyze resume information according to a double-index portrait system, and obtain a potential key information corpus.

[0011] Use the potential key information corpus as knowledge graph data, and perform path matching on the knowledge graph data according to the key post game mechanism, and output a key post talent recommendation scheme.

[0012] The above-mentioned enterprise key post talent intelligent recommendation method based on knowledge graph takes the Big Five personality theory as the basis, converts core dimensions such as openness and responsibility into quantifiable personality characteristic indicators, and fuses them with key features such as stress resistance and collaboration tendency required by the post, to build a "personality trait + post feature" double-index portrait system. This system breaks through the limitations of traditional subjective evaluation, realizes the accurate measurement of personality traits through standardized dimensions, and makes implicit personality factors into explicit data that can be calculated and compared, providing a basis for quantitative matching. Secondly, a closed-loop intelligent mechanism is built: first, the relationship sample between personality characteristics and enterprise competition is clustered by using a clustering algorithm to generate a key post game mechanism with suppression characteristics, which clearly defines the adaptation threshold and balance relationship of different personality traits in post competition; then, a large language model is used to deeply analyze resumes and extract potential key information to form a knowledge graph; finally, based on the game mechanism, the knowledge graph is path-matched to realize intelligent decision mapping from personality traits to post requirements. This deep integration of personality traits into the whole process of talent evaluation not only solves the quantification problem through the double-index system, but also endows the system with dynamic decision-making ability through the game mechanism and knowledge graph, promoting the shift from "experience judgment" to "data-driven" in the matching of people and posts, significantly improving the accuracy and efficiency of key post talent recommendation, and helping enterprises to upgrade the intelligence of human resource allocation and improve efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 FIG. 1 is a flowchart of an embodiment of the enterprise key post talent intelligent recommendation method based on knowledge graph;

[0014] Figure 2 FIG. 2 is an elbow rule diagram in an embodiment;

[0015] Figure 3 FIG. 3 is an innovative coordination type radar distribution diagram in an embodiment;

[0016] Figure 4 FIG. 4 is a high-pressure and decisive type radar distribution diagram in an embodiment;

[0017] Figure 5 FIG. 5 is a flexible and adaptable type radar distribution diagram in an embodiment;

[0018] Figure 6 For a robust and efficient type of each dimension radar profile in an embodiment. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0020] In one embodiment, as shown in Figure 1 A knowledge graph-based enterprise key position talent intelligent recommendation method is provided, including the following steps:

[0021] Step 102, evaluating the personality characteristics of employees based on the Big Five Personality Theory.

[0022] Specifically, the five dimensions of the Big Five Personality Theory are mapped to personality traits, and the characteristic dimensions of personality traits are established: neuroticism, extroversion, openness, agreeableness, and conscientiousness. A 5-level Likert scale (1=not at all, 2=not, 3=undecided, 4=agree, 5=fully agree) is used to design a questionnaire (as shown in Table 1), and the questions marked with "(R)" need to be scored in reverse. In the questionnaire design, the focus is mainly on the behavior performance between the enterprises, and how to lead the subordinates to complete the task is crucial.

[0023] Table 1 Self-evaluation form of personality traits

[0024]

[0025] Step 104, constructing a double-index portrait system according to the personality characteristics and the key position characteristics.

[0026] Specifically, the personnel profiling model fully considers multiple factors, and the constructed indicator system covers multiple dimensions including basic employee information, work experience, education and training, task experience, research achievements, and personality traits. The job profiling model fully integrates actual work needs, establishing an indicator system comprising seven dimensions: job requirements, professional skills, information literacy, task requirements, task experience, organizational management, and personality traits (competitors). These correspond to the personnel profiling, and each primary indicator is further subdivided into multiple secondary indicators for precise description. A knowledge graph is used to visually present the indicator system, achieving a multi-dimensional and objective description of personnel and positions, providing a solid data foundation for subsequent personnel-job matching decisions. Furthermore, personality traits are incorporated into the indicator system for the first time in the profiling model, fully considering the differences in management philosophy and corporate culture caused by the different personality traits of senior managers in competing companies. This represents the game-theoretic relationship in corporate competition, thereby improving the comprehensiveness and accuracy of the profiling model and providing support for the personnel-job matching algorithm.

[0027] Step 106: Obtain a sample of the relationship between personality traits and corporate competition, and use a clustering algorithm to cluster the relationship samples to generate a game mechanism for key positions.

[0028] Specifically, the questionnaire was distributed, and based on the sample results obtained from the feedback, the reliability of the questionnaire reached an acceptable level through Cronbach's α coefficient test (α=0.81).

[0029] Furthermore, the K-means algorithm is used to cluster personality traits. First, the sum of squared errors (SSE) within clusters corresponding to different k values ​​is calculated, and then an elbow diagram (e.g.) is plotted. Figure 2 As shown in the diagram, the elbow rule diagram clearly shows that the rate of descent decreases significantly when k is 4, indicating that the data should be clustered into 4 categories.

[0030] Furthermore, using the K-means algorithm, four points were randomly selected as centroids. Each data point was assigned to the cluster containing the nearest centroid, and the centroid of each cluster was recalculated. This process was repeated until the centroids no longer changed. Ultimately, four personality trait types were obtained: Innovative and Coordinating, High-Pressure and Autocratic, Agile and Adaptable, and Steady and Efficient (the scores for each dimension are shown in Table 2).

[0031] Table 2. Cluster Center Scores for Personality Trait Types

[0032]

[0033] Furthermore, the scores for all dimensions of innovative and coordinating senior talents are relatively high, with openness and rigor being the highest. The score distribution for each dimension is shown below. Figure 3Its core advantage lies in the ability to "break boundaries" and "integrate". The high openness feature drives its rapid application of cutting-edge technology, and it is good at using unexpected means; at the same time, high rigor ensures the executability of the innovation scheme, which means that it is relatively conservative in confrontation strategy and emphasizes systematic thinking. High sociability reflects excellent team collaboration ability, good at adjusting conflicts and contradictions, and can effectively coordinate various resources; high neuroticism has the risk of insufficient stress tolerance.

[0034] High-pressure and decisive senior talents have high neuroticism and low rigor. The distribution of their scores in each dimension is shown in Figure 4 In the early stage of crisis, it shows strong control ability. The sense of urgency driven by high neuroticism enables it to quickly issue orders in unexpected situations, and the advantage of extroversion supports its authoritative image, making it an extremely aggressive manager. However, low rigor may lead to execution vulnerabilities, causing unnecessary economic losses.

[0035] Agile and adaptable senior talents are good at achieving flexible management in dynamic environments. The distribution of their scores in each dimension is shown in Figure 5 Its low neuroticism allows it to maintain a stable mindset during prolonged work. Higher openness reflects the ability of such senior talents to accept method innovation or their willingness to believe in the advantages of new methods, and they tend to be aggressive in management. However, low rigor may lead to loss of control or inaccurate degree.

[0036] Stable and efficient senior talents have low neuroticism, low openness, and high rigor. The distribution of their scores in each dimension is shown in Figure 6 This type of senior talent shows strong stability and execution in their work; their low neuroticism allows them to remain calm and make rational decisions even when faced with temporary strategy changes. High rigor reflects their extreme reliance on standardized processes. Low openness indicates their limited acceptance of innovative methods and their preference for proven mature solutions.

[0037] Further, based on the personality trait model based on the Big Five personality, by analyzing the score differences of different types of senior talents in each dimension, the differences in decision-making, team interaction, and execution of solutions are obtained, and the suppression of different types of enterprises is constructed. The game mechanism aims to reveal the different personality traits in the game mechanism of enterprise confrontation, and achieve efficient allocation of enterprise human resources.

[0038] Further, according to the score differences of different types in the five dimensions, the personality trait suppression chain is constructed: stable and efficient type → innovative coordination type → high-pressure and decisive type → agile and adaptable type → stable and efficient type.

[0039] Step 108, use a large language model to analyze resume information based on a double-index portrait system to obtain a potential key information corpus.

[0040] Step 110, taking the potential key information corpus as knowledge graph data, performing path matching on the knowledge graph data according to the key post game mechanism, and outputting a key post talent recommendation scheme.

[0041] In the above-mentioned enterprise key post talent intelligent recommendation method based on knowledge graph, based on the Big Five personality theory, the core dimensions such as openness and responsibility are converted into quantifiable personality characteristic indicators, which are fused with the key features such as stress resistance and cooperation tendency required by the post to construct a "personality trait + post feature" double-index portrait system. This system breaks through the limitations of traditional subjective evaluation, realizes the accurate measurement of personality traits through standardized dimensions, and converts implicit personality factors into explicit data that can be calculated and compared, providing a basis for quantitative matching. Secondly, a closed-loop intelligent mechanism is constructed: first, the relationship samples between personality characteristics and enterprise competition are clustered through clustering algorithm to generate a key post game mechanism with anti-suppression characteristics, and the adaptive threshold and balance relationship of different personality traits in post competition are clearly defined; secondly, the resume is analyzed in depth using a large language model to extract potential key information and form a knowledge graph; finally, based on the game mechanism, the knowledge graph is path matched to realize intelligent decision mapping from personality traits to post requirements. This deep integration of personality traits into the whole process of talent evaluation not only solves the quantization problem through the double-index system, but also endows the system with dynamic decision-making ability through the game mechanism and knowledge graph, promoting the shift from "experience judgment" to "data-driven" in the matching of people and positions, significantly improving the accuracy and efficiency of key post talent recommendation, and helping enterprises to upgrade the intelligence of human resource allocation and improve efficiency.

[0042] In one of the embodiments, based on the Big Five personality theory, the personality characteristics of the five dimensions of the employees are constructed, a Likert scale is used to design a questionnaire, and the personality characteristics are evaluated by using the questionnaire.

[0043] In one of the embodiments, the Delphi method is used to determine the first-level indicators and second-level indicators from the objective data of the employees according to the personality characteristics, the second-level indicators are fine-tuned according to the key post characteristics of different enterprises to obtain specific second-level indicators, the weights between the first-level indicators and the specific second-level indicators are determined by using the analytic hierarchy process, and the double-index portrait system is constructed according to the weights.

[0044] In one of the embodiments, the relationship samples between the personality characteristics and the enterprise competition are obtained according to the evaluation results of the personality characteristics evaluated by the questionnaire, the intra-cluster error sum of squares corresponding to different k values of the relationship samples is calculated to determine the personality trait type according to the elbow diagram, and the key post game mechanism is generated by clustering the relationship samples according to the personality trait type.

[0045] In one of the embodiments, according to the differences in the scores of the relationship samples in the five dimensions of the personality trait type, the personality trait suppression chain is generated as: robust and efficient type suppresses innovative and coordinated type, innovative and coordinated type suppresses high-pressure and decisive type, high-pressure and decisive type suppresses agile and adaptable type, agile and adaptable type suppresses robust and efficient type, and the competition game mechanism of different personality traits in key positions is obtained.

[0046] It is worth noting that the robust and efficient type suppresses the innovative and coordinated type: the robust and efficient type can effectively coordinate team resources, pay attention to execution details and specifications, and ensure stable task progress by virtue of the advantages of agreeableness and conscientiousness. When facing the innovative and coordinated type, although the latter has strong innovation ability, it may ignore the standardization and stability of execution. The robust and efficient type can evaluate the feasibility of the innovation scheme, convert innovation into actual results through meticulous management and stable execution, and avoid chaos caused by excessive innovation, thereby suppressing the innovative and coordinated type in terms of execution stability and risk control.

[0047] The innovative and coordinated type suppresses the high-pressure and decisive type: the high openness of the innovative and coordinated type gives it the ability to break through the rigid mode caused by the low openness of the high-pressure and decisive type; the team cohesion driven by the high agreeableness of the former can dissolve the individual authority dependence derived from the low agreeableness of the latter. When facing the high-pressure and decisive type, the latter has a higher neuroticism score, with large emotional fluctuations and impulsivity, which may lead to decision-making errors and team instability. The innovative and coordinated type can flexibly respond with innovative thinking and suppress the high-pressure and decisive type through coordination and calm analysis.

[0048] The high-pressure and decisive type suppresses the agile and adaptable type: the high-pressure and decisive type has higher scores in extraversion and neuroticism, with strong senior talent temperament and interpersonal communication ability, and can make rapid decisions in a high-pressure environment. The agile and adaptable type has lower scores in neuroticism and extraversion, and may appear conservative and hesitant in high-pressure situations. The high-pressure and decisive type suppresses the flexibility of the agile and adaptable type by making decisive decisions and having strong senior talent temperament to quickly adjust the direction of the team.

[0049] The agile and adaptable type suppresses the robust and efficient type: the agile and adaptable type has a higher score in openness and can quickly adapt to new methods, flexibly respond to changes, and be good at using situations to create favorable situations for themselves. The robust and efficient type has a lower score in openness and lacks innovative thinking, preferring traditional methods. The agile and adaptable type breaks the conservative mode of the robust and efficient type by flexibly adjusting strategies in a changing environment, thereby suppressing its flexibility.

[0050] In one of the embodiments, the large language model is used to iteratively train the characteristics of the associated statements in the relationship samples according to the double-index profiling system, generate the matching relationship corpus of key positions and employees, and use the word segmentation tool to process the matching relationship corpus and store it in the new file of the database. The Skip-Gram model is used to retrieve the matching relationship corpus of the new file to generate word vectors, and calculate the cosine similarity between each word vector:

[0051] ;

[0052] wherein, is the i-th word vector, is the j-th word vector, is the cosine similarity between the i-th word vector and the j-th word vector. According to the cosine similarity, the resume information is analyzed to obtain the potential key information corpus.

[0053] In one of the embodiments, the potential key information corpus and the preset weight information are input as knowledge graph data, and the path matching of the knowledge graph data is performed according to the key position game mechanism through the input query statement:

[0054] ;

[0055] wherein, M is the matching degree, R is the path set, P is a type of path, p is a path, w 1 is the weight of the path p from the position to the first index of the position, w 2 is the weight of the path p from the first index of the position to the specific second index of the position, is the similarity correlation between indexes, w 3 is the weight of the path p from the employee to the first index of the employee profile, w 4 is the weight of the path p from the first index of the employee to the specific second index of the employee. Complete path information is obtained. The complete path information is exported in JSON format using the export tool, the knowledge graph data exported in JSON format is parsed, the complete path information between the position and the personnel is obtained, and the key position talent recommendation scheme is output.

[0056] It's worth noting that, firstly, through expert interviews, the potential relationships between the secondary indicators of each individual profile and the secondary indicators of which job profiles existed were identified and recorded. Then, based on these identified potential relationships, a large-scale artificial intelligence model was used to generate a large number of statements describing the connections between the indicators. These, combined with work experience and personnel file data, formed a corpus of approximately 50,000 words. Finally, the corpus was preprocessed using the jieba word segmentation tool, including word segmentation, retention of indicator words, and filtering of stop words and punctuation. The processed corpus was then saved to a new file.

[0057] Furthermore, word vectors generated using the Skip-Gram model are used to measure the similarity between words by calculating the cosine similarity between the vectors. For two word vectors... v i and v j The formula for calculating cosine similarity is:

[0058] ;

[0059] The similarity correlation score ranges from -1 to 1; a value of 1 indicates that the two vectors are identical, while a value of -1 indicates that the two vectors are completely opposite. Therefore, the closer the correlation score is to 1, the greater the correlation between the two words. The finalized index system and weight information are input into the Neo4j database to build a knowledge graph, providing data support for subsequent matching analysis.

[0060] Furthermore, by inputting a query, all complete path information is retrieved, and the provided export tool is used to export the data into JSON format. The exported JSON file contains the nodes, relationships, and attribute information of the knowledge graph, providing a data foundation for subsequent path search and matching degree calculation. By parsing the JSON-formatted knowledge graph data, all complete paths between personnel and positions can be obtained.

[0061] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0062] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0063] Any combination of the technical features of the above embodiments can be made, and in order to make the description simple, not all possible combinations of the technical features in the above embodiments are described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0064] The above-mentioned embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the application. It should be noted that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A knowledge graph-based intelligent recommendation method for key post talents in an enterprise, characterized in that, The method comprises: Evaluate the personality characteristics of employees based on the Big Five Personality Theory; According to the personality characteristics and the key post characteristics, a double-index portrait system is constructed, and the specific steps are as follows: adopting the Delphi method to determine the first-level index and the second-level index from the objective data of employees according to the personality characteristics, and the second-level index is fine-tuned according to the key post characteristics of different enterprises to obtain specific second-level indexes; The weights between the first-level indexes and the specific second-level indexes are determined by using the analytic hierarchy process, and the double-index portrait system is constructed according to the weights; Obtain the relationship samples between the personality characteristics and the enterprise competition, and use clustering algorithm to cluster the relationship samples to generate the key post game mechanism, and the specific steps are as follows: according to the personality trait type, the score difference of the relationship samples in five dimensions is clustered to generate the personality trait suppression chain: stable and efficient type suppresses innovative and coordinated type, the innovative and coordinated type suppresses high-pressure and decisive type, the high-pressure and decisive type suppresses agile and adaptable type, and the agile and adaptable type suppresses stable and efficient type, to obtain the competition game mechanism of different personality traits in key posts; Use a large language model to analyze resume information according to the double-index portrait system to obtain a potential key information corpus; Take the potential key information corpus as knowledge graph data, and perform path matching on the knowledge graph data according to the key post game mechanism to output a key post talent recommendation scheme, and the specific steps are as follows: input the potential key information corpus and the preset weight information as knowledge graph data, and perform path matching on the knowledge graph data according to the key post game mechanism through the input query statement: wherein, M is the matching degree, R is the path set, P is a class of paths, p is a path, w 1 is the path p the weight of the post-level indicator to the post, w 2 is the path p the weight of the post-level indicator to the specific two-level indicator, is the similarity correlation between indicators, w 3 is the path p the weight of the employee portrait-level indicator to the employee, w 4 is the path p the weight of the employee-level indicator to the specific two-level indicator; Obtain complete path information; After exporting the complete path information in JSON format, parse the JSON format exported knowledge graph data to obtain all complete path information between posts and personnel, and output a key post talent recommendation scheme.

2. The method of claim 1, wherein, Evaluate the personality characteristics of employees based on the Big Five Personality Theory, including: Based on the Big Five Personality Theory, the personality characteristics of employees in five dimensions are constructed, a Likert scale is designed to design a questionnaire, and the personality characteristics are evaluated by using the questionnaire.

3. The method of claim 2, wherein, Obtain the relationship samples between the personality characteristics and the enterprise competition, and use clustering algorithm to cluster the relationship samples to generate the key post game mechanism, including: According to the evaluation results of the personality characteristics evaluated by the questionnaire, the relationship samples between the personality characteristics and the enterprise competition are obtained, the intra-cluster error sum of squares corresponding to the different k values of the relationship samples is calculated to determine the personality trait type according to the elbow diagram, the relationship samples are clustered according to the personality trait type, and the key post game mechanism is generated.

4. The method according to any one of claims 1 to 3, characterized in that, Use a large language model to analyze resume information according to the double-index portrait system to obtain a potential key information corpus, including: Iteratively train the relationship samples according to the double-index portrait system by using a large language model to generate a matching relationship corpus of key posts and employees, and store the processed matching relationship corpus in a new file in the database after processing the matching relationship corpus by using a word segmentation tool; The Skip-Gram model is used to call the matching relationship corpus of the new file to generate word vectors, and the cosine similarity between each word vector is calculated. wherein, v i is the i-th word vector, v j is the j-th word vector, is the cosine similarity of the i-th word vector and the j-th word vector; According to the cosine similarity, the resume information is analyzed to obtain a potential key information corpus.

Images