Talent analysis and selection system based on AI auxiliary interview

Abstract

The invention discloses a talent analysis and selection system based on AI auxiliary interview, relates to the technical field of talent analysis and selection, and solves the problems that in the prior art, evaluation on the honesty degree of an applicant is not considered, many applicants excessively describe the ability and personality of the applicant on a resume in order to embody the advantages of the applicant, even counterfeiting is caused, and the evaluation efficiency is poor. As a result, enterprises may find that candidates do not meet post requirements in the later period and have to perform recruitment procedures again, and loss of time and economic cost is caused. The method comprises the following steps: acquiring ability data on a resume of an applicant and interview data when the applicant interviews; processing the interview data to obtain interview text data; calculating cosine similarity between the interview text data and the capability data to obtain similarity data; calculating to obtain an integrity evaluation coefficient based on the similarity data; and judging whether the honesty degree of the applicant is qualified based on the honesty evaluation coefficient and a preset honesty threshold. The technical problem is solved.

Description

Technical Field

[0001] The present invention belongs to the field of talent analysis and selection, and specifically is a talent analysis and selection system based on AI-assisted interviews. Background Art

[0002] The traditional talent selection process relies on manual interviews, which are time-consuming and highly subjective. They are easily influenced by the interviewer's personal preferences and are difficult to achieve large-scale, efficient, fair and objective evaluation.

[0003] The prior art (invention patent with application number 2021108786489) discloses an intelligent interview method based on expression analysis, including: the interview begins; real-time voice data collection; the server performs voice recognition on the applicant's answer to obtain text information; the server compares the text information with the vocabulary information in the corpus to complete lexical analysis; the server completes pragmatic reasoning based on the lexical analysis results; the server generates evaluation results based on the lexical analysis and pragmatic reasoning results; the server pushes the evaluation results to the employer in real time. This invention patent realizes a practical and targeted full-process product for talent assessment and interview, so that the employer can screen out talents that match the recruitment position, so that each job seeker can get a more fair and objective personality evaluation. In addition, not only does the interview complete the assessment of job seekers in the traditional interview, but it also connects to the data processing process, making the personality assessment results more fair, objective and accurate, and improving the efficiency of corporate talent selection. However, existing technologies do not take into account the evaluation of applicants' integrity. In order to highlight their own strengths, many applicants over-describe their abilities and personalities on their resumes, or even falsify them. As a result, companies may find at a later stage that the candidates do not meet the job requirements and have to go through the recruitment process again, resulting in loss of time and economic costs.

[0004] Therefore, the present invention proposes a talent analysis and selection system based on AI-assisted interview to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a talent analysis and selection system based on AI-assisted interviews, which is used to solve the technical problem that the prior art does not take into account the evaluation of the integrity of applicants, and many applicants over-describe their abilities and personalities on their resumes in order to highlight their advantages, or even falsify them, which may result in companies discovering that the candidates do not meet the job requirements at a later stage and have to repeat the recruitment process, resulting in loss of time and economic costs.

[0006] To achieve the above-mentioned object, a first aspect of the present invention provides a talent analysis and selection system based on AI-assisted interview, comprising: a data collection module, a data analysis module and an integrity assessment module;

[0007] Data collection module: used to obtain the ability data on the applicant's resume and the interview data of the applicant during the interview;

[0008] Data analysis module: obtains interview text data by processing the interview data; wherein the interview text data includes: emotional data and skill data; and,

[0009] By calculating the cosine similarity between the interview text data and the ability data, similarity data is obtained; wherein the similarity data includes: emotional similarity and skill similarity;

[0010] Integrity assessment module: The integrity assessment coefficient is calculated based on the similarity data; based on the integrity assessment coefficient and the preset integrity threshold, the applicant's integrity is judged to be qualified; if yes, the applicant is considered as a candidate; if no, the applicant is not considered as a candidate.

[0011] Preferably, the data analysis module is communicated and / or electrically connected with the data acquisition module and the integrity assessment module respectively.

[0012] Preferably, the obtaining of the ability data on the applicant's resume includes:

[0013] Keywords related to personal abilities are extracted from applicants' resumes through natural language processing technology to obtain ability data; natural language processing technologies include: TF-IDF, TextRank or LDA; personal abilities include: personality, mastered technology or areas of expertise; ability data include: personality keywords and skill keywords.

[0014] It should be noted that the mastered technology refers to a series of skills, software, platforms, methods or tools that job seekers are proficient in using in a specific field or industry; the personality keywords refer to the keywords related to personality in the resume; the skill keywords refer to the keywords related to mastered technology or areas of expertise in the resume.

[0015] Preferably, the interview data of the applicant during the interview includes:

[0016] The interview data of the applicants during the interview is collected in real time through data collection equipment; the interview data includes: video, audio and text information during the interview process.

[0017] It should be noted that the video of the interview process records the changes in the candidate's facial expressions and behavioral behaviors during the interview; the audio records the specific content of the candidate's answers to questions as well as the speed and expression ability of the answers; the text information records the specific content of the candidate's answers to questions on paper or on the computer.

[0018] Preferably, the processing of the interview data to obtain the interview text data includes:

[0019] Identify and obtain the candidate's emotion data and the candidate's skill data; wherein the emotion data includes: emotion information and emotion category;

[0020] S1: Train an artificial intelligence model based on the FER2013 dataset to obtain an expression recognition model; wherein the artificial intelligence model includes: a convolutional neural network or a deep belief network;

[0021] S2: Based on the video of the interview process, human feature recognition technology and expression recognition model are used to extract the emotional changes of the applicant during the interview to obtain emotional information; human feature recognition technology includes: PoseNet or VIBE; emotional information includes: confidence, nervousness, happiness, fear and surprise, etc.;

[0022] S3: Convert the audio of the interview process into speech text through speech recognition technology; the speech recognition technology includes: Google Cloud Speech-to-Text or IBM Watson Speech to Text;

[0023] S4: Extract emotional features from speech text through a deep learning NLP model; classify the speech text based on the emotional features through a sentiment analysis model to obtain emotional categories; wherein the sentiment analysis model includes: VADER model or TextBlob model; the emotional categories include: positive, negative or neutral.

[0024] It should be noted that the FER2013 dataset is a dataset for facial expression recognition, which is provided by researchers from the University of California, Los Angeles and the International Face Recognition Competition (FERET), and contains 35,887 images, each of which is 48x48 pixels in size; the emotional features refer to keywords that express emotions.

[0025] Preferably, the applicant's skill data includes:

[0026] Keywords related to personal abilities are extracted from voice text and text information through natural language processing technology, and all extracted keywords are named skill data.

[0027] The present invention acquires and analyzes applicants' non-verbal signals, such as facial expressions and body language, through expression recognition models and human feature recognition technology to quantify their emotional information, such as their confidence level and nervousness. Through speech recognition and NLP models, the applicants' verbal expressions are converted into analyzable text data, which reduces the influence of subjective judgment and can comprehensively consider applicants' performance from multiple dimensions, helping companies to identify talents more accurately and optimize recruitment processes.

[0028] Preferably, the calculating of the cosine similarity between the interview text data and the capability data comprises:

[0029] The keywords in the interview text data and the competence data are represented as vectors of fixed dimensions by a word vector model, and the word vectors of the sentiment data and the skill data in the interview text data, as well as the word vectors of the personality keywords and the skill keywords in the competence data are obtained; wherein the word vector model includes Word2Vec or GloVe;

[0030] Compute the cosine similarity between each word vector.

[0031] It should be noted that the keywords in the interview text data refer to emotional information, emotional categories, and skill data.

[0032] Preferably, the calculating the cosine similarity between each word vector includes:

[0033] The cosine similarity algorithm is used to calculate the cosine similarity between each word vector in the sentiment data and each word vector in the personality keyword, and the average value is taken to obtain the sentiment similarity;

[0034] The cosine similarity algorithm is used to calculate the cosine similarity between each word vector in the skill data and each word vector in the skill keyword, and the average value is taken to obtain the skill similarity.

[0035] Preferably, the calculation of the integrity assessment coefficient based on the similarity data includes:

[0036] Label the sentiment similarity as Q and the skill similarity as J;

[0037] The integrity assessment coefficient is calculated through the formula: P = A×e^Q / (e^Q+1)+B×e^J / (e^J+1); where P is the integrity assessment coefficient, and A and B are proportional coefficients.

[0038] Preferably, judging whether the applicant's integrity is qualified based on the integrity evaluation coefficient and the preset integrity threshold includes:

[0039] Determine whether the integrity assessment coefficient is greater than or equal to the preset integrity threshold; if yes, the integrity is determined to be qualified; if not, the integrity is determined to be unqualified; among which, the preset integrity threshold is set according to the actual enterprise recruitment requirements.

[0040] The present invention obtains similarity data by calculating the cosine similarity between interview text data and ability data; calculates a credibility evaluation coefficient based on the similarity data; and judges whether the credibility of the applicant is qualified based on the credibility evaluation coefficient and a preset credibility threshold, thereby solving the technical problem that the prior art does not consider evaluating the credibility of the applicant, resulting in time and economic cost losses in the later stage.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. The prior art does not consider evaluating the integrity of applicants; in order to highlight their own advantages, many applicants over-describe their abilities and personalities on their resumes, or even falsify them, which may result in the company discovering that the candidates do not meet the job requirements at a later stage and having to redo the recruitment process, resulting in a technical problem of loss of time and economic costs; the present invention obtains the ability data on the applicant's resume and the interview data of the applicant during the interview; obtains interview text data by processing the interview data; and obtains similarity data by calculating the cosine similarity between the interview text data and the ability data; calculates an integrity evaluation coefficient based on the similarity data; and determines whether the applicant's integrity is qualified based on the integrity evaluation coefficient and a preset integrity threshold, thereby solving the problem that the prior art does not consider evaluating the applicant's integrity.

[0043] 2. The present invention can capture and analyze the non-verbal signals of candidates, such as facial expressions and body language, through expression recognition models and human feature recognition technology, so as to quantify their emotional information such as confidence level and nervousness; through speech recognition and NLP models, it can convert the verbal expressions of candidates into analyzable text data, and further explore clues to their emotional tendencies and professional skills; it not only improves the efficiency of interview evaluation, but also enhances the accuracy and fairness of the results. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 paying creative work.

[0045] Figure 1 A schematic diagram of a system module according to an embodiment of the present invention;

[0046] Figure 2 A schematic diagram of a specific process of an embodiment of the present invention;

[0047] Figure 3 Schematic diagram of applicant integrity assessment according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] See also Figure 1-Figure 2 , the first aspect of the present invention provides a talent analysis and selection system based on AI-assisted interview, including: a data collection module, a data analysis module and an integrity assessment module;

[0050] Data collection module: used to obtain the ability data on the applicant's resume and the interview data of the applicant during the interview;

[0051] Data analysis module: obtains interview text data by processing the interview data; the interview text data includes: emotional data and skill data;

[0052] Obtain competency data from applicants’ resumes, including:

[0053] Keywords related to personal abilities are extracted from applicants' resumes through natural language processing technology to obtain ability data; natural language processing technologies include: TF-IDF, TextRank or LDA; personal abilities include: personality, mastered technology or areas of expertise; ability data include: personality keywords and skill keywords.

[0054] Interview data of candidates during interviews, including:

[0055] The interview data of the applicants during the interview is collected in real time through data collection equipment; wherein the data collection equipment includes: a recorder and a camera; the interview data includes: video, audio and text information during the interview process.

[0056] By processing the interview data, we get the interview text data, including:

[0057] Identify and obtain the candidate's emotion data and the candidate's skill data; wherein the emotion data includes: emotion information and emotion category;

[0058] S1: Train an artificial intelligence model based on the FER2013 dataset to obtain an expression recognition model; wherein the artificial intelligence model includes: a convolutional neural network or a deep belief network;

[0059] S2: Based on the video of the interview process, human feature recognition technology and expression recognition model are used to extract the emotional changes of the applicant during the interview to obtain emotional information; human feature recognition technology includes: PoseNet or VIBE; emotional information includes: confidence, nervousness, happiness, fear and surprise, etc.;

[0060] S3: Convert the audio of the interview process into speech text through speech recognition technology; the speech recognition technology includes: Google Cloud Speech-to-Text or IBM Watson Speech to Text;

[0061] S4: Extract emotional features from speech text through a deep learning NLP model; classify the speech text based on the emotional features through a sentiment analysis model to obtain emotional categories; wherein the sentiment analysis model includes: VADER model or TextBlob model; the emotional categories include: positive, negative or neutral.

[0062] The candidate’s skills data, including:

[0063] Keywords related to personal abilities are extracted from voice text and text information through natural language processing technology, and all extracted keywords are named skill data.

[0064] For example, suppose a technology company is hiring a software development engineer. Combine resume analysis with interview data analysis to more comprehensively assess the candidate's abilities and emotional state. The specific steps are as follows:

[0065] Step 1: Data acquisition module;

[0066] Resume capability data acquisition: By using natural language processing technology (such as TF-IDF), keywords such as "Java", "Python", "Agile development", "Teamwork" etc. are extracted from the applicant's resume. These keywords represent the applicant's skills and personality traits.

[0067] Step 2: Data analysis module;

[0068] Interview data collection: During the interview, a high-definition camera is used to record the candidate’s facial expressions and body movements. Professional recording equipment is used to capture the candidate’s voice, including speaking speed, tone, etc., as well as the candidate’s written or electronic answers to questions.

[0069] Step 3: Interview text data processing;

[0070] Expression recognition and emotion analysis: The expression recognition model (such as convolutional neural network) trained with the FER2013 dataset is used to analyze the interview video and identify the emotional changes of the candidate during the interview, such as confidence, nervousness, happiness, etc. At the same time, the PoseNet model is used to analyze body language to further understand the emotional state of the candidate.

[0071] Speech-to-text conversion and analysis: Use speech recognition technologies such as Google Cloud Speech-to-Text to convert interview audio into text, and then use deep learning NLP models (such as LSTM) to extract emotional features. Then use sentiment analysis models such as VADER or TextBlob to classify the text and determine whether it is positive, negative, or neutral.

[0072] Skill data extraction: From the interview text, by using natural language processing techniques such as word frequency statistics or topic modeling (LDA), we can extract job-related skill keywords such as "microservice architecture" and "cloud computing" to evaluate the technical strength of the applicant.

[0073] Comprehensive analysis and decision making;

[0074] The recruitment team combines the resume analysis results with the interview data analysis results to form an assessment report on the candidate's comprehensive capabilities. For example, if a candidate shows rich programming skills in his resume, and the interview video analysis shows that he is highly confident and proactive in the interview, and the voice text analysis also shows that he is clear and emotionally stable, then this candidate will be considered a very suitable candidate.

[0075] Through a systematic analysis process, companies can more objectively and comprehensively evaluate the capabilities and potential value of applicants, reduce subjective bias, improve the accuracy and efficiency of recruitment decisions, and ultimately find the most suitable talent for the team.

[0076] By calculating the cosine similarity between the interview text data and the ability data, similarity data is obtained; wherein the similarity data includes: emotional similarity and skill similarity;

[0077] Integrity assessment module: The integrity assessment coefficient is calculated based on the similarity data; based on the integrity assessment coefficient and the preset integrity threshold, the applicant's integrity is judged to be qualified; if yes, the applicant is considered as a candidate; if no, the applicant is not considered as a candidate.

[0078] See also Figure 3 , by calculating the cosine similarity between the interview text data and the ability data, including:

[0079] The keywords in the interview text data and the competence data are represented as vectors of fixed dimensions by a word vector model, and the word vectors of the sentiment data and the skill data in the interview text data, as well as the word vectors of the personality keywords and the skill keywords in the competence data are obtained; wherein the word vector model includes Word2Vec or GloVe;

[0080] Compute the cosine similarity between each word vector.

[0081] Calculate the cosine similarity between each word vector, including:

[0082] The cosine similarity algorithm is used to calculate the cosine similarity between each word vector in the sentiment data and each word vector in the personality keyword, and the average value is taken to obtain the sentiment similarity;

[0083] The cosine similarity algorithm is used to calculate the cosine similarity between each word vector in the skill data and each word vector in the skill keyword, and the average value is taken to obtain the skill similarity.

[0084] The integrity evaluation coefficient is calculated based on the similarity data, including:

[0085] Label the sentiment similarity as Q and the skill similarity as J;

[0086] The integrity assessment coefficient is calculated through the formula: P = A×e^Q / (e^Q+1)+B×e^J / (e^J+1); where P is the integrity assessment coefficient, and A and B are proportional coefficients.

[0087] The integrity of the applicant is determined based on the integrity evaluation coefficient and the preset integrity threshold, including:

[0088] Determine whether the integrity assessment coefficient is greater than or equal to the preset integrity threshold; if yes, the integrity is determined to be qualified; if not, the integrity is determined to be unqualified; among which, the preset integrity threshold is set according to the actual enterprise recruitment requirements.

[0089] For example, suppose a large software development company is hiring a project manager. Combine the ability data on the candidate's resume with the text data during the interview and calculate the cosine similarity to evaluate the candidate's integrity. The specific steps are as follows:

[0090] Step 1: Data preparation and vectorization;

[0091] Resume competency data: Keywords such as “leadership,” “agile management,” and “communication skills” are extracted from resumes using natural language processing technology and expressed as personality keywords and skill keywords.

[0092] Interview text data: Based on the audio-to-text and video analysis results during the interview process, keywords that reflect the candidate’s emotional state and skill display, such as “confidence,” “nervousness,” and “agile development,” are extracted and represented as sentiment data and skill data.

[0093] Step 2: word vector representation;

[0094] Use a word vector model (such as Word2Vec or GloVe) to convert all keywords into vectors of fixed dimensions. For example, keywords such as "leadership" and "confidence" are represented as corresponding word vectors.

[0095] Step 3: Calculate cosine similarity;

[0096] Sentiment similarity: Calculate the cosine similarity between the personality keyword vector in the resume and the sentiment data vector in the interview text, and then take the average to get the sentiment similarity.

[0097] Skill similarity: Calculate the cosine similarity between the skill keyword vector in the resume and the skill data vector in the interview text, and then take the average to get the skill similarity.

[0098] Step 4: Calculate the integrity assessment coefficient;

[0099] The proportional coefficients A=0.4 and B=0.6 are set to represent the weights of sentiment similarity and skill similarity in integrity evaluation.

[0100] Apply the formula P = A × e^Q / (e^Q+1) + B × e^J / (e^J+1), where Q is the sentiment similarity and J is the skill similarity, to calculate the integrity assessment coefficient P.

[0101] Step 5: Integrity judgment;

[0102] Assume that the preset integrity threshold is set to 0.7. If the calculated integrity evaluation coefficient P is greater than or equal to 0.7, the applicant's integrity is considered qualified; otherwise, it is considered unqualified.

[0103] Assume that the keywords in a candidate's resume include "leadership", "communication skills", and "agile management", and the keywords in the interview text data include "confidence", "fluent communication", and "agile development". Through the word vector model and cosine similarity calculation, the sentiment similarity Q is 0.8 and the skill similarity J is 0.9. According to the formula, the integrity evaluation coefficient P is 0.82 (assuming the proportional coefficient A = 0.4, B = 0.6). Since P is greater than the preset integrity threshold of 0.7, the applicant is judged to have qualified integrity and can enter the subsequent recruitment process.

[0104] By implementing the above-mentioned integrity assessment system, the company can more objectively and scientifically evaluate the integrity performance of applicants, reduce the loss of time and economic costs caused by the later discovery of integrity issues, thereby optimizing the recruitment process and improving recruitment efficiency and quality.

[0105] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0106] Working principle of the present invention:

[0107] The present invention obtains the ability data on the applicant's resume and the interview data of the applicant during the interview; obtains the interview text data by processing the interview data; and obtains the similarity data by calculating the cosine similarity between the interview text data and the ability data; calculates the integrity evaluation coefficient based on the similarity data; judges whether the applicant's integrity is qualified based on the integrity evaluation coefficient and a preset integrity threshold; if yes, the applicant is hired or considered as a candidate; if not, the applicant is eliminated, thereby solving the technical problem that the prior art does not take into account the evaluation of the applicant's integrity; many applicants over-describe their abilities and personalities on their resumes, or even falsify them, in order to reflect their own advantages, which results in the company possibly finding that the candidate does not meet the job requirements at a later stage and having to go through the recruitment process again, resulting in the loss of time and economic costs.

[0108] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A talent analysis and selection system based on AI-assisted interview, characterized in that: Data collection module, data analysis module and integrity assessment module; Data collection module: used to obtain the ability data on the applicant's resume and the interview data of the applicant during the interview; Data analysis module: obtains interview text data by processing the interview data; wherein the interview text data includes: emotional data and skill data; and, By calculating the cosine similarity between the interview text data and the ability data, similarity data is obtained; wherein the similarity data includes: emotional similarity and skill similarity; Integrity assessment module: The integrity assessment coefficient is calculated based on the similarity data; based on the integrity assessment coefficient and the preset integrity threshold, it is judged whether the applicant's integrity is qualified; if yes, the applicant will be considered as a candidate; if not, the applicant will not be considered as a candidate.

2. According to claim 1, a talent analysis and selection system based on AI-assisted interviewing is characterized in that: The data analysis module is respectively communicated and / or electrically connected with the data acquisition module and the integrity assessment module.

3. According to claim 1, a talent analysis and selection system based on AI-assisted interviewing is characterized in that: The obtaining of the ability data on the applicant's resume includes: Keywords related to personal abilities are extracted from applicants' resumes through natural language processing technology to obtain ability data; natural language processing technologies include: TF-IDF, TextRank or LDA; personal abilities include: personality, mastered technology or areas of expertise; ability data include: personality keywords and skill keywords.

4. According to claim 1, a talent analysis and selection system based on AI-assisted interviewing is characterized in that: The interview data of the applicant during the interview includes: The interview data of the applicants during the interview is collected in real time through data collection equipment; the interview data includes: video, audio and text information during the interview process.

5. According to claim 1, a talent analysis and selection system based on AI-assisted interviewing is characterized in that: The interview text data is obtained by processing the interview data, including: Identify and obtain the candidate's emotion data and the candidate's skill data; wherein the emotion data includes: emotion information and emotion category; S1: Train an artificial intelligence model based on the FER2013 dataset to obtain an expression recognition model; wherein the artificial intelligence model includes: a convolutional neural network or a deep belief network; S2: Based on the video of the interview process, human feature recognition technology and expression recognition model are used to extract the emotional changes of the applicant during the interview and obtain emotional information; S3: Convert the audio of the interview process into speech text through speech recognition technology; the speech recognition technology includes: Google Cloud Speech-to-Text or IBM Watson Speech to Text; S4: Extract emotional features from speech text through a deep learning NLP model; classify the speech text based on the emotional features through a sentiment analysis model to obtain emotional categories; wherein the sentiment analysis model includes: VADER model or TextBlob model; the emotional categories include: positive, negative or neutral.

6. The talent analysis and selection system based on AI-assisted interview according to claim 5, characterized in that: The applicant's skill data includes: Keywords related to personal abilities are extracted from voice text and text information through natural language processing technology, and all extracted keywords are named skill data.

7. The talent analysis and selection system based on AI-assisted interview according to claim 1 is characterized in that: The method of calculating the cosine similarity between the interview text data and the capability data includes: The keywords in the interview text data and the competence data are represented as vectors of fixed dimensions by a word vector model, and the word vectors of the sentiment data and the skill data in the interview text data, as well as the word vectors of the personality keywords and the skill keywords in the competence data are obtained; wherein the word vector model includes Word2Vec or GloVe; Compute the cosine similarity between each word vector.

8. The talent analysis and selection system based on AI-assisted interview according to claim 7, characterized in that: The calculation of the cosine similarity between each word vector includes: The cosine similarity algorithm is used to calculate the cosine similarity between each word vector in the sentiment data and each word vector in the personality keyword, and the average value is taken to obtain the sentiment similarity; The cosine similarity algorithm is used to calculate the cosine similarity between each word vector in the skill data and each word vector in the skill keyword, and the average value is taken to obtain the skill similarity.

9. The talent analysis and selection system based on AI-assisted interview according to claim 1, characterized in that: The integrity evaluation coefficient calculated based on the similarity data includes: Label the sentiment similarity as Q and the skill similarity as J; The integrity assessment coefficient is calculated through the formula: P = A×e^Q / (e^Q+1)+B×e^J / (e^J+1); where P is the integrity assessment coefficient, and A and B are proportional coefficients.

10. The talent analysis and selection system based on AI-assisted interview according to claim 1, characterized in that: The judging of whether the applicant's integrity is qualified based on the integrity evaluation coefficient and the preset integrity threshold includes: Determine whether the integrity assessment coefficient is greater than or equal to the preset integrity threshold; if yes, the integrity is determined to be qualified; if not, the integrity is determined to be unqualified; among which, the preset integrity threshold is set according to the actual enterprise recruitment requirements.

Images