Interview question generation method and device based on knowledge graph, equipment and medium

Abstract

The invention provides an interview question generation method and device based on a knowledge graph, equipment and a medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining post knowledge information of a workplace post and character job application information of candidates; constructing a post knowledge graph according to the post knowledge information, and constructing a figure portrait knowledge graph according to the figure job application information; generating a post interview question according to the post knowledge graph, the figure portrait knowledge graph and the large language model; and under the condition that the post interview question answering result does not meet the post requirements of the workplace posts, according to the post interview question answering result and the large language model, generating a target post interview questioning question. According to the technical scheme, the post knowledge graph and the figure portrait knowledge graph are constructed, then the post interview questions are generated, dynamic analysis is performed based on the post interview questions, and finally the post interview question answering results of the candidates are determined based on the post interview questions to dynamically generate the target post interview questions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a knowledge graph-based interview question generation method, device, equipment and medium. BACKGROUND

[0002] Intelligent generation of interview questions is an important technical means in the interview process.

[0003] Currently, in the prior art, traditional interview question generation can generally use a question bank or a training class to provide only static interview questions, lacking dynamic follow-up capabilities. Or it can use a specific format to ask questions in one direction, lacking effective understanding of the candidate's answer content and targeted follow-up questions, which cannot help the interviewer to dig into the candidate's knowledge depth and breadth and ultimately determine the matching degree of the candidate and the position, resulting in a disconnection between the questions and the job requirements, and an inability to deeply analyze the candidate's own situation.

[0004] Therefore, there is an urgent need for a knowledge graph-based interview question generation method to generate dynamic target post interview follow-up questions. SUMMARY

[0005] The present application provides a knowledge graph-based interview question generation method, device, equipment and medium to solve the problem that in the prior art, traditional interview question generation can generally use a question bank or a training class to provide only static interview questions, lacking dynamic follow-up capabilities; or it can use a specific format to ask questions in one direction, lacking effective understanding of the candidate's answer content and targeted follow-up questions, which cannot help the interviewer to dig into the candidate's knowledge depth and breadth and ultimately determine the matching degree of the candidate and the position, resulting in a disconnection between the questions and the job requirements, and an inability to deeply analyze the candidate's own situation. By constructing a post knowledge graph and a character portrait knowledge graph, then generating a post interview question, and dynamically analyzing based on the post interview question, the final post interview question answer result of the candidate is determined based on the post interview question to dynamically generate a target post interview follow-up question, greatly improving the authenticity and immersion of the target post interview follow-up question.

[0006] The present application provides a knowledge graph-based interview question generation method, comprising the following steps.

[0007] Obtain post knowledge information of a job position and character job-seeking information of a candidate; wherein the post knowledge information represents the recruitment situation, industry standard situation and post ability situation of the job position, and the character job-seeking information represents the candidate's own situation when job-seeking; According to the post knowledge information, a post knowledge graph is constructed, and according to the job-seeking information of a person, a person portrait knowledge graph is constructed; wherein the post knowledge graph represents a graph constructed after analyzing the recruitment situation, the industry standard situation and the post ability situation of a job position; the person portrait knowledge graph represents a graph constructed after analyzing the self situation of a candidate; According to the post knowledge graph, the person portrait knowledge graph and the large language model, a post interview question is generated; wherein the large language model represents a language model with analysis ability, creation ability, reasoning ability and execution ability; According to the post interview question, a post interview question answering result of the candidate is determined; In the case that the post interview question answering result does not meet the post demand of the job position, according to the post interview question answering result and the large language model, a target post interview follow-up question is generated.

[0008] According to the post knowledge information, a post knowledge graph is constructed, including: according to the post knowledge information and a deep learning model, a post knowledge entity recognition result and a post knowledge relationship extraction result are determined; wherein the deep learning model is a model used for entity recognition and relationship extraction; according to the post knowledge entity recognition result and the post knowledge relationship extraction result, the post knowledge graph is constructed.

[0009] According to the job-seeking information of a person, a person portrait knowledge graph is constructed, including: according to the job-seeking information of a person and a deep learning model, a person job-seeking entity recognition result and a person job-seeking relationship extraction result are determined; according to the person job-seeking entity recognition result and the person job-seeking relationship extraction result, the person portrait knowledge graph is constructed.

[0010] According to the post knowledge graph, the person portrait knowledge graph and the large language model, a post interview question is generated; wherein the large language model represents a language model with analysis ability, creation ability, reasoning ability and execution ability;

[0011] According to the method, the post matching situation of the candidate for the job post is determined according to the target question answer result, the post knowledge graph and the large language model, including: inputting the interview question answer result and the post knowledge graph into a semantic analysis layer in the large language model to obtain a similarity result output by the semantic analysis layer; inputting the interview question answer result into a logic analysis layer in the large language model to obtain an element detection result output by the logic analysis layer; inputting the interview question answer result into a psychological analysis layer in the large language model to obtain a psychological analysis result output by the psychological analysis layer; and inputting the similarity result, the element detection result and the psychological analysis result into a post matching analysis layer in the large language model to obtain the post matching situation of the candidate for the job post output by the post matching analysis layer.

[0012] According to the method, the post matching situation of the candidate for the job post is determined according to the target question answer result, the post knowledge graph and the large language model, including: inputting the interview question answer result and the post knowledge graph into a semantic analysis layer in the large language model to obtain a similarity result output by the semantic analysis layer; inputting the interview question answer result into a logic analysis layer in the large language model to obtain an element detection result output by the logic analysis layer; inputting the interview question answer result into a psychological analysis layer in the large language model to obtain a psychological analysis result output by the psychological analysis layer; and inputting the similarity result, the element detection result and the psychological analysis result into a post matching analysis layer in the large language model to obtain the post matching situation of the candidate for the job post output by the post matching analysis layer.

[0013] According to the method, the post matching situation of the candidate for the job post is determined according to the target question answer result, the post knowledge graph and the large language model, including: inputting the interview question answer result and the post knowledge graph into a semantic analysis layer in the large language model to obtain a similarity result output by the semantic analysis layer; inputting the interview question answer result into a logic analysis layer in the large language model to obtain an element detection result output by the logic analysis layer; inputting the interview question answer result into a psychological analysis layer in the large language model to obtain a psychological analysis result output by the psychological analysis layer; and inputting the similarity result, the element detection result and the psychological analysis result into a post matching analysis layer in the large language model to obtain the post matching situation of the candidate for the job post output by the post matching analysis layer.

[0014] The application further provides an interview question generation device based on a knowledge graph, including the following modules.

[0015] The information acquisition module is configured to acquire post knowledge information of the job post and character job-seeking information of the candidate, wherein the post knowledge information represents the recruitment situation, the industry standard situation and the post ability situation of the job post, and the character job-seeking information represents the self situation of the candidate when the candidate is job-seeking. The knowledge graph construction module is configured to construct a post knowledge graph according to the post knowledge information and construct a character portrait knowledge graph according to the character job-seeking information, wherein the post knowledge graph represents a graph constructed after analyzing the recruitment situation, the industry standard situation and the post ability situation of the job post, and the character portrait knowledge graph represents a graph constructed after analyzing the self situation of the candidate. The interview question generation module is configured to generate a post interview question according to the post knowledge graph, the character portrait knowledge graph and a large language model, wherein the large language model represents a language model with analysis ability, creation ability, reasoning ability and execution ability. The answer result determination module is configured to determine a post interview question answer result of the candidate according to the post interview question. The follow-up question generation module is configured to generate a follow-up interview question for the target position according to the interview question answer result and the large language model when the interview question answer result does not meet the position requirements of the position.

[0016] The application further provides an electronic device, including a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the knowledge graph-based interview question generation method according to any one of the above when executing the computer program.

[0017] The application further provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the knowledge graph-based interview question generation method according to any one of the above.

[0018] The application further provides a computer program product, which includes a computer program, and the computer program is executed by a processor to implement the knowledge graph-based interview question generation method according to any one of the above.

[0019] The application provides a knowledge graph-based interview question generation method, device and equipment and a medium. The method comprises the following steps: obtaining post knowledge information of a job position and character job-seeking information of a candidate; the post knowledge information represents the recruitment situation, industry standard situation and post ability situation of the job position, and the character job-seeking information represents the self situation of the candidate when the candidate is job-seeking; constructing a post knowledge graph according to the post knowledge information, and constructing a character portrait knowledge graph according to the character job-seeking information; the post knowledge graph represents a graph constructed after analyzing the recruitment situation, industry standard situation and post ability situation of the job position; the character portrait knowledge graph represents a graph constructed after analyzing the self situation of the candidate; generating a post interview question according to the post knowledge graph, the character portrait knowledge graph and a large language model; the large language model represents a language model with analysis ability, creation ability, reasoning ability and execution ability; determining a post interview question answer result of the candidate according to the post interview question; and generating a target post interview follow-up question according to the post interview question answer result and the large language model in the case that the post interview question answer result does not meet the post demand of the job position. The technical solution of the application solves the problem that the traditional interview question generation in the prior art can only provide static interview questions by using a question bank or a training class and lacks dynamic follow-up ability, or can adopt a specific mode to ask questions in one direction and lacks effective understanding of the answer content of the candidate and targeted follow-up questions, which cannot help the interviewer to dig the depth and breadth of the knowledge of the candidate and finally determine the matching degree of the candidate and the post, resulting in the problem that the questions are not consistent with the post demand and the self situation of the candidate cannot be analyzed in depth. The post knowledge graph and the character portrait knowledge graph are constructed, then the post interview question is generated, dynamic analysis is performed based on the post interview question, the post interview question answer result of the candidate is finally determined based on the post interview question, and the target post interview follow-up question is dynamically generated, which greatly improves the authenticity and immersion of the target post interview follow-up question. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative effort.

[0021] Figure 1 It is a flowchart of the knowledge graph-based interview question generation method provided by the application.

[0022] Figure 2 It is a structural schematic diagram of the knowledge graph-based interview question generation device provided by the application.

[0023] Figure 3 is a structural schematic diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0024] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0025] The present application provides a method for generating interview questions based on a knowledge graph, which is described below. Figure 1 The method for generating interview questions based on a knowledge graph provided by the present application can be applied to the generation of interview questions by a simulated interviewer based on artificial intelligence and a knowledge graph. The execution subject of the method can be an electronic device or a knowledge graph-based interview question generation device arranged in the electronic device. The knowledge graph-based interview question generation device can be realized by software, hardware or a combination of both. Figure 1 is a flowchart of the method for generating interview questions based on a knowledge graph provided by the present application, as Figure 1 shown, the method comprises the following steps 101, 102, 103, 104 and 105.

[0026] Step 101, obtaining job position knowledge information and candidate personal job-seeking information.

[0027] In this step, the job position knowledge information represents the recruitment situation, industry standard situation and job ability situation of the job position, and the personal job-seeking information represents the self situation of the candidate when seeking a job.

[0028] The recruitment situation of the job position may, for example, be derived from major recruitment websites and platforms covering recruitment information. The industry standard situation of the job position refers to the standard of the professional position in the industry. The job ability situation refers to the ability requirement of the professional position, which can be determined by the experience of the professional in the professional position, and this embodiment is not limited in this regard.

[0029] The candidate refers to a job seeker who needs to seek a job.

[0030] Specifically, the recruitment situation, industry standard situation and job ability situation of the professional position are obtained as the job position knowledge information, and the personal job-seeking information of the candidate is obtained.

[0031] Step 102, constructing a job knowledge graph according to the job position knowledge information, and constructing a personal portrait knowledge graph according to the personal job-seeking information.

[0032] In this step, the post knowledge graph represents a graph constructed after analyzing the recruitment situation, industry standard situation and post ability situation of the job position; and the person portrait knowledge graph represents a graph constructed after analyzing the candidate's own situation. Specifically, after obtaining the post knowledge information, entity recognition and relationship extraction are performed on the post knowledge information, so as to construct a post knowledge graph according to the results of entity recognition and relationship extraction on the post knowledge information; and after obtaining the person job-seeking information, entity recognition and relationship extraction are performed on the person job-seeking information, so as to construct a person portrait knowledge graph according to the results of entity recognition and relationship extraction on the person job-seeking information.

[0033] For example, entity recognition is performed on the post knowledge information, and the recognized entities can be, for example, post name, post skill requirement, behavior characteristic requirement and psychological quality requirement. Relationship extraction is performed on the post knowledge information, and the extracted relationships can be, for example, that the "data analyst" node is associated with the "structured query language (SQL) skill", the SQL is further associated with the "data modeling" behavior, and the behavior is associated with the "logical reasoning" psychological quality, which is not limited in the embodiment.

[0034] For example, entity recognition is performed on the person job-seeking information, and the recognized entities can be, for example, candidate resume, candidate career experience and candidate psychological test data, and relationship extraction is performed on the person job-seeking information, and the extracted relationships can be, for example, candidate skill label and personality characteristics, which are not limited in the embodiment.

[0035] The advantage of such setting is that the traditional interview questions are often from a fixed question bank, and the questions lack pertinence and flexibility. The post knowledge graph and the person portrait knowledge graph are used as knowledge support for subsequent generation of post interview questions, and through the entities and relationships in the post knowledge graph and the person portrait knowledge graph, it can be ensured that the generated post interview questions are closely related to the post requirements of the job position, such as required skills, behavior characteristics and psychological quality. For example, for the "data analyst" post, the post interview question generation engine can not only ask professional questions about "SQL programming", but also generate behavior and psychological quality questions in combination with the "logical reasoning" and "team cooperation" nodes in the graph. The use of the post knowledge graph and the person portrait knowledge graph makes the subsequent post interview question generation controllable and professional, avoiding the common deviation and randomness of the existing language model in generating questions, which is not limited in the embodiment.

[0036] In a specific embodiment, the post knowledge graph is constructed according to the post knowledge information, including: determining a post knowledge entity recognition result and a post knowledge relationship extraction result according to the post knowledge information and a deep learning model; wherein the deep learning model is a model used for entity recognition and relationship extraction; and constructing the post knowledge graph according to the post knowledge entity recognition result and the post knowledge relationship extraction result.

[0037] In this step, the deep learning model may, for example, include models of a post entity recognition module and a post relationship extraction module. The post entity recognition module may, for example, adopt a Bidirectional Encoder Representations from Transformers-Named Entity Recognition (BERT-NER), which is a Chinese named entity recognition model based on a BERT architecture and can effectively identify entity names. The post relationship extraction module may, for example, adopt a Bidirectional Long Short-Term Memory Conditional Random Field (BiLSTM-CRF), which is a deep learning model used for sequence labeling tasks. The present embodiment does not limit this.

[0038] Specifically, the post knowledge information is input into the post entity recognition module in the deep learning model to obtain a post knowledge entity recognition result output by the post entity recognition module; and the post knowledge information is input into the post relationship extraction module in the deep learning model to obtain a post knowledge relationship extraction result output by the post relationship extraction module; wherein the deep learning model is a model used for entity recognition and relationship extraction; and the post knowledge graph is constructed according to the post knowledge entity recognition result and the post knowledge relationship extraction result.

[0039] The advantage of such a setting is to ensure high accuracy of entities and relationships. The post knowledge graph is updated in an incremental manner, and is iteratively updated after obtaining the latest post data from a recruitment website or platform each time to ensure the real-time nature of the post requirements of the job post.

[0040] In a specific embodiment, the post knowledge graph can also be stored by forming a hierarchical network through a graph database, which may, for example, be a Neo4j, a graph database management system. Neo4j is an open-source graph database specially used for storing, managing and querying graph data. The present embodiment does not limit this.

[0041] In a specific embodiment, the character portrait knowledge graph is constructed according to the character job-seeking information, including: determining character job-seeking entity recognition results and character job-seeking relationship extraction results according to the character job-seeking information and a deep learning model; and constructing the character portrait knowledge graph according to the character job-seeking entity recognition results and the character job-seeking relationship extraction results.

[0042] Specifically, the character job-seeking information is input into the deep learning model to obtain character job-seeking entity recognition results and character job-seeking relationship extraction results output by the deep learning model; and the character portrait knowledge graph is constructed according to the character job-seeking entity recognition results and the character job-seeking relationship extraction results.

[0043] The advantage of such a setting is that the deep learning model can complete the characteristics of the candidate in the psychological dimension by performing sentiment analysis and psychology scale mapping on the input character job-seeking information, thereby forming the character portrait knowledge graph.

[0044] Step 103, generating a job interview question according to the job knowledge graph, the character portrait knowledge graph, and a large language model.

[0045] In this step, the large language model represents a language model with analysis ability, creativity, reasoning ability, and execution ability; the large language model may, for example, be composed of a knowledge graph retrieval and an artificial intelligence language model, and the large language model is designed to understand and generate human language, which is usually trained on a large amount of sample data, and the present embodiment does not limit this.

[0046] The job interview question is the first round question generated by the large language model.

[0047] Specifically, after obtaining the job knowledge graph and the character portrait knowledge graph, the job knowledge graph and the character portrait knowledge graph are input into the large language model, and the job knowledge graph and the character portrait knowledge graph are extracted by the large language model to obtain the job interview question output by the large language model.

[0048] Step 104, determining a job interview question answer result of the candidate according to the job interview question.

[0049] Specifically, after obtaining the job interview question, the candidate is asked according to the job interview question, and a job interview question answer result fed back by the candidate is obtained.

[0050] Step 105, in the case that the job interview question answer result does not meet the job requirements of the job, generating a target job interview follow-up question according to the job interview question answer result and the large language model.

[0051] In this step, the job requirements of the job position refer to the specific responsibilities, skills and experience of the candidate required by the job position, and in the case that the job interview question answer result does not meet the job requirements of the job position, it specifically refers to the case that the job interview question answer result is not complete or deviates from the job requirements of the job position, and the present embodiment does not limit this.

[0052] Specifically, in the case that the job interview question answer result is not complete or deviates from the job requirements of the job position, it is determined that the job interview question answer result does not meet the job requirements of the job position, and the target job interview follow-up question is continued to be generated according to the job interview question answer result and the large language model.

[0053] For example, in the obtained job interview question answer result, the candidate does not mention the result when describing the project experience, and the job requirements require the description of the project result. At this time, the target job interview follow-up question is continued to be generated according to the job interview question answer result and the large language model, and the target job interview follow-up question may be, for example, "Please specify what achievements did the project finally bring?" This mechanism ensures that after the question is generated, the answer result of the candidate is closer to the authenticity and depth of the simulation interview process.

[0054] The advantage of such setting is that in order to solve the problem of the existing one-way question static simulation system, the follow-up mechanism is added, and in the case that the job interview question answer result does not meet the job requirements of the job position, the target job interview follow-up question is continued to be generated, which greatly improves the authenticity and immersion of the simulation question.

[0055] In a specific embodiment, after the target job interview follow-up question is generated according to the job interview question answer result and the large language model, it further includes: determining the target question answer result of the candidate according to the target job interview question; determining the job matching situation of the candidate for the job position according to the target question answer result, the job knowledge graph and the large language model.

[0056] Specifically, the target question answer result of the candidate is determined according to the target job interview question; the job matching situation of the candidate for the job position is determined according to the target question answer result, the job knowledge graph and the large language model.

[0057] In a specific embodiment, the post matching situation of the candidate for the job post is determined according to the target question answer result, the post knowledge graph and the large language model, including: inputting the interview question answer result and the post knowledge graph into the semantic analysis layer in the large language model to obtain the similarity result output by the semantic analysis layer; inputting the interview question answer result into the logic analysis layer in the large language model to obtain the element detection result output by the logic analysis layer; inputting the interview question answer result into the psychological analysis layer in the large language model to obtain the psychological analysis result output by the psychological analysis layer; inputting the similarity result, the element detection result and the psychological analysis result into the post matching analysis layer in the large language model to obtain the post matching situation of the candidate for the job post output by the post matching analysis layer.

[0058] In this step, the semantic analysis layer mainly calculates the cosine similarity of the interview question answer result of the candidate and the post knowledge graph, and the semantic analysis layer can use BERT for example; the logic analysis layer mainly detects whether the STAR (Situation, target, action, result) elements are contained by using pattern matching, and the confirmed part will be recorded and continue to trigger the step of generating the target post interview follow-up question according to the post interview question answer result and the large language model; the post matching analysis layer mainly performs sentiment tendency analysis (positive / neutral / negative) on the target question answer result, and if there is voice input, the features such as speech speed, pause and tone will also be extracted to infer the nervousness, which is not limited in this embodiment.

[0059] Specifically, the interview question answer result and the post knowledge graph are input into the semantic analysis layer in the large language model, the semantic analysis layer calculates the cosine similarity of the interview question answer result and the post knowledge graph by performing semantic matching to obtain the similarity result output by the semantic analysis layer, and the similarity result is used to judge whether the candidate covers the required skills and behaviors of the job post; the interview question answer result is input into the logic analysis layer in the large language model, the logic analysis layer detects whether the key elements are contained by using logic rules to obtain the element detection result output by the logic analysis layer; the interview question answer result is input into the psychological analysis layer in the large language model, the psychological analysis layer combines the emotion classifier or the acoustic feature extractor to infer the psychological state of the candidate, including the confidence, the nervousness and the stress resistance. All results are scored by a weighting formula, and the weights of the skill, the logic and the psychological dimensions can be flexibly adjusted according to the characteristics of the post, so as to obtain the psychological analysis result output by the psychological analysis layer; the similarity result, the element detection result and the psychological analysis result are input into the post matching analysis layer in the large language model to obtain the post matching situation of the candidate for the job post output by the post matching analysis layer.

[0060] In one specific embodiment, after determining the candidate's job matching status based on the target question answers, job knowledge graph, and large language model, the method further includes: generating a mock interview report based on the job matching status; wherein the mock interview report is used to provide feedback on the candidate's job matching status.

[0061] In this step, the mock interview report has quantitative indicators, such as skill coverage, logical completeness, confidence index, etc. This embodiment does not limit these.

[0062] The mock interview report also includes qualitative suggestions (such as "increasing the quantitative expression of data results" and "supplementing role division in teamwork questions"). The mock interview report can be displayed through a visual interface or output via an Application Programming Interface (API) for recruitment platforms to use on a large scale. For example, in financial industry interviews, the weight of stress resistance can be increased to 0.4, while in R&D positions, the weight of skills can be increased to 0.6. This embodiment does not impose any limitations on these aspects.

[0063] Specifically, after determining the candidate's job fit based on the target question answers, job knowledge graph, and large language model, the system determines the candidate's skill coverage, logical integrity, confidence index, etc., and generates a mock interview report based on these metrics. The mock interview report is used to provide feedback on the candidate's job fit.

[0064] In one specific embodiment, the method further includes: determining a job matching completion result when the candidate's answers to the job interview questions meet the job requirements; wherein the job matching completion result is used to indicate that the candidate is matched with a job.

[0065] Specifically, if the candidate's answers to the job interview questions match the job requirements, the job matching result is determined, indicating that the candidate is a good fit for the position.

[0066] This invention provides a knowledge graph-based method for generating interview questions. It involves acquiring job knowledge information about a specific position and candidate job-seeking information. The job knowledge information represents the recruitment situation, industry standards, and job competency requirements for the position, while the candidate job-seeking information represents their own situation during the job application process. A job knowledge graph is constructed based on the job knowledge information, and a candidate profile knowledge graph is constructed based on the candidate job-seeking information. The job knowledge graph is constructed after analyzing the recruitment situation, industry standards, and job competency requirements for the position; the candidate profile knowledge graph is constructed after analyzing the candidate's own situation. Interview questions are generated based on the job knowledge graph, the candidate profile knowledge graph, and a large language model. The large language model represents a language model with analytical, creative, reasoning, and execution capabilities. The candidate's answers to the interview questions are determined based on the interview questions. If the answers do not meet the job requirements, follow-up interview questions for the target position are generated based on the answers and the large language model. The technical solution of this invention addresses the problems in existing technologies where traditional interview question generation typically relies on question banks or training courses that only provide static interview questions, lacking dynamic follow-up questioning capabilities; or it may use specific formats for one-way questioning, lacking effective understanding of the candidate's answers and targeted follow-up questions. These follow-up questions fail to help interviewers uncover the depth and breadth of the candidate's knowledge and ultimately determine the candidate's suitability for the position, resulting in a disconnect between questions and job requirements, and a failure to deeply analyze the candidate's own situation. This invention constructs a job knowledge graph and a candidate profile knowledge graph, then generates job interview questions, and dynamically analyzes these questions. Finally, based on the candidate's answers to the job interview questions, it dynamically generates follow-up questions for the target job interview, significantly improving the realism and immersiveness of the follow-up questions.

[0067] The knowledge graph-based interview question generation device provided by the present invention will be described below. The knowledge graph-based interview question generation device described below can be referred to in correspondence with the knowledge graph-based interview question generation method described above.

[0068] Figure 2 This is a schematic diagram of the knowledge graph-based interview question generation device provided by the present invention, with reference to... Figure 2 As shown, the knowledge graph-based interview question generation device 200 includes: an information acquisition module 201, a knowledge graph construction module 202, an interview question generation module 203, an answer result determination module 204, and a follow-up question generation module 205; wherein, The information acquisition module 201 is used to acquire job knowledge information for workplace positions and job-seeking information for candidates. Among them, the job knowledge information represents the recruitment situation, industry standards, and job competency requirements for workplace positions, while the job-seeking information represents the candidate's own situation when applying for a job. The knowledge graph construction module 202 is used to construct a job knowledge graph based on job knowledge information and a person profile knowledge graph based on job seeker information. The job knowledge graph is a graph constructed after analyzing the recruitment situation, industry standards, and job competency of the job. The person profile knowledge graph is a graph constructed after analyzing the candidate's own situation. The interview question generation module 203 is used to generate job interview questions based on the job knowledge graph, the person profile knowledge graph, and the large language model; among them, the large language model represents a language model with analytical, creative, reasoning, and execution capabilities; The answer result determination module 204 is used to determine the candidate's answer results to the job interview questions based on the job interview questions; The follow-up question generation module 205 is used to generate follow-up questions for the target job interview based on the job interview answers and the large language model when the job interview answers do not meet the job requirements.

[0069] In one example embodiment, the knowledge graph construction module 202 constructs a job knowledge graph based on job knowledge information, specifically for: determining job knowledge entity recognition results and job knowledge relationship extraction results based on job knowledge information and a deep learning model; wherein, the deep learning model is a model used for entity recognition and relationship extraction; and constructing the job knowledge graph based on the job knowledge entity recognition results and job knowledge relationship extraction results.

[0070] In one example embodiment, the knowledge graph construction module 202 constructs a person profile knowledge graph based on the person's job-seeking information, specifically for: determining the person's job-seeking entity recognition result and the person's job-seeking relationship extraction result based on the person's job-seeking information and a deep learning model; and constructing the person profile knowledge graph based on the person's job-seeking entity recognition result and the person's job-seeking relationship extraction result.

[0071] In one example embodiment, the device further includes a job matching module. The job matching module is configured to: after generating follow-up interview questions for the target job based on the candidate's answers to the job interview questions and a large language model, determine the candidate's target question answers based on the target job interview questions; and determine the candidate's job matching status for the workplace position based on the target question answers, the job knowledge graph, and the large language model.

[0072] In one example embodiment, the job matching module determines the candidate's job matching status based on the target question answer results, job knowledge graph, and large language model. Specifically, it is used to: input the interview question answer results and job knowledge graph into the semantic analysis layer of the large language model to obtain the similarity results output by the semantic analysis layer; input the interview question answer results into the logical analysis layer of the large language model to obtain the element detection results output by the logical analysis layer; input the interview question answer results into the psychological analysis layer of the large language model to obtain the psychological analysis results output by the psychological analysis layer; and input the similarity results, element detection results, and psychological analysis results into the job matching analysis layer of the large language model to obtain the candidate's job matching status output by the job matching analysis layer.

[0073] In one example embodiment, the device further includes a report generation module. The report generation module is configured to: after determining the candidate's job suitability based on the target question answers, the job knowledge graph, and the large language model, generate a mock interview report based on the job suitability; wherein the mock interview report is used to provide feedback on the candidate's job suitability.

[0074] In one example embodiment, the device further includes a completion result determination module. The completion result determination module is configured to: determine a job matching completion result if the candidate's answers to the job interview questions meet the job requirements; wherein the job matching completion result indicates that the candidate is a suitable candidate for the job.

[0075] The apparatus of this embodiment can be used to execute the method of any embodiment in the side embodiment of the knowledge graph-based interview question generation method. Its specific implementation process and technical effects are similar to those in the side embodiment of the knowledge graph-based interview question generation method. For details, please refer to the detailed description in the side embodiment of the knowledge graph-based interview question generation method, which will not be repeated here.

[0076] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 3As shown, the electronic device may include: a processor 310, a communications interface 320, a memory 330, and a communications bus 340, wherein the processor 310, the communications interface 320, and the memory 330 communicate with each other through the communications bus 340. The processor 310 can call logic instructions in the memory 330 to execute a knowledge graph-based interview question generation method. This method includes: acquiring job knowledge information about the position and the candidate's job-seeking information; wherein the job knowledge information represents the recruitment situation, industry standards, and job competency requirements of the position, and the job-seeking information represents the candidate's own situation when seeking employment; constructing a job knowledge graph based on the job knowledge information and a person profile knowledge graph based on the person's job-seeking information; wherein the job knowledge graph represents a graph constructed after analyzing the recruitment situation, industry standards, and job competency requirements of the position; and the person profile knowledge graph represents a graph constructed after analyzing the candidate's own situation; generating job interview questions based on the job knowledge graph, the person profile knowledge graph, and a large language model; wherein the large language model represents a language model with analytical, creative, reasoning, and execution capabilities; determining the candidate's answers to the job interview questions based on the job interview questions; and generating follow-up interview questions for the target position based on the job interview questions answers and the large language model if the job interview questions answers do not meet the job requirements.

[0077] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0078] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the knowledge graph-based interview question generation method provided by the above methods. This method includes: acquiring job knowledge information of workplace positions and job-seeking information of candidates; wherein, the job knowledge information represents the recruitment situation, industry standards, and job competency of the workplace position, and the job-seeking information represents the candidate's own situation when seeking employment; constructing a job knowledge graph based on the job knowledge information, and constructing a person profile knowledge graph based on the person-seeking information; The system comprises several components: a job knowledge graph (constructed by analyzing recruitment information, industry standards, and job competency requirements for a given job); a candidate profile knowledge graph (constructed by analyzing the candidate's own qualifications); job interview questions generated based on the job knowledge graph, candidate profile knowledge graph, and a large language model; the large language model representing a language model with analytical, creative, reasoning, and execution capabilities; the candidate's answers to the job interview questions; and follow-up interview questions for the target job generated based on the candidate's answers and the large language model if the candidate's answers do not meet the job requirements.

[0079] Furthermore, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the knowledge graph-based interview question generation method provided by the aforementioned methods. This method includes: acquiring job knowledge information of workplace positions and job-seeking information of candidates; wherein the job knowledge information represents the recruitment situation, industry standards, and job competency requirements of the workplace position, and the job-seeking information represents the candidate's own situation when seeking employment; constructing a job knowledge graph based on the job knowledge information, and constructing a person profile knowledge graph based on the person's job-seeking information; wherein the job knowledge graph represents the... The system constructs a knowledge graph based on an analysis of job postings, industry standards, and job requirements. A personality profile knowledge graph represents a knowledge graph constructed based on an analysis of the candidate's own situation. Job interview questions are generated based on the job knowledge graph, the personality profile knowledge graph, and a large language model. The large language model represents a language model with analytical, creative, reasoning, and execution capabilities. The system determines the candidate's answers to the job interview questions. If the candidate's answers do not meet the job requirements, follow-up interview questions for the target position are generated based on the answers and the large language model.

[0080] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0081] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating interview questions based on knowledge graphs, characterized in that, include: The system acquires job knowledge information for workplace positions and job-seeking information for candidates. The job knowledge information refers to the recruitment status, industry standards, and job competency requirements for the workplace position, while the job-seeking information refers to the candidate's personal circumstances when applying for a job. A job knowledge graph is constructed based on the job knowledge information, and a person profile knowledge graph is constructed based on the job seeker information; wherein, the job knowledge graph represents a graph constructed after analyzing the recruitment situation, industry standards, and job competency of the job; the person profile knowledge graph represents a graph constructed after analyzing the candidate's own situation; Job interview questions are generated based on the job knowledge graph, the person profile knowledge graph, and the large language model; wherein, the large language model represents a language model with analytical, creative, reasoning, and execution capabilities; The candidate's answers to the job interview questions are determined based on the job interview questions. If the answers to the job interview questions do not meet the job requirements, follow-up interview questions for the target job are generated based on the answers to the job interview questions and the large language model.

2. The knowledge graph-based interview question generation method according to claim 1, characterized in that, The step of constructing a job knowledge graph based on the job knowledge information includes: The job knowledge entity recognition result and job knowledge relation extraction result are determined based on the job knowledge information and the deep learning model; wherein, the deep learning model is a model used for entity recognition and relation extraction. The job knowledge graph is constructed based on the job knowledge entity recognition results and the job knowledge relationship extraction results.

3. The knowledge graph-based interview question generation method according to claim 2, characterized in that, The step of constructing a knowledge graph of a person's profile based on the job-seeking information includes: Based on the job-seeking information of the individuals and the deep learning model, the results of the identification of job-seeking entities and the extraction of job-seeking relationships are determined. The person profile knowledge graph is constructed based on the person job seeker entity recognition results and the person job seeker relationship extraction results.

4. The knowledge graph-based interview question generation method according to claim 1, characterized in that, After generating follow-up interview questions for the target position based on the answers to the job interview questions and the large language model, the process further includes: Determine the candidate's answers to the target questions based on the target job interview questions; The job matching status of the candidate for the workplace position is determined based on the answer to the target question, the job knowledge graph, and the large language model.

5. The knowledge graph-based interview question generation method according to claim 4, characterized in that, The step of determining the candidate's job match for the workplace position based on the answer to the target question, the job knowledge graph, and the large language model includes: The interview questions and answers, along with the job knowledge graph, are input into the semantic analysis layer of the large language model to obtain the similarity results output by the semantic analysis layer. The interview questions are answered and the results are input into the logic analysis layer of the large language model to obtain the element detection results output by the logic analysis layer. The interview questions are answered and the results are input into the psychological analysis layer of the large language model to obtain the psychological analysis results output by the psychological analysis layer. The similarity results, the element detection results, and the psychological analysis results are input into the job matching analysis layer in the large language model to obtain the job matching status of the candidate for the workplace position output by the job matching analysis layer.

6. The knowledge graph-based interview question generation method according to claim 4, characterized in that, After determining the candidate's job match for the workplace position based on the answer to the target question, the job knowledge graph, and the large language model, the method further includes: A mock interview report is generated based on the job matching information; wherein, the mock interview report is used to provide feedback on the candidate's match with the job position.

7. The knowledge graph-based interview question generation method according to claim 1, characterized in that, Also includes: If the candidate's answers to the job interview questions meet the job requirements, a job matching completion result is determined; wherein, the job matching completion result indicates that the candidate is a good match for the job.

8. A knowledge graph-based interview question generation device, characterized in that, include: The information acquisition module is used to acquire job knowledge information for workplace positions and job-seeking information for candidates; wherein, the job knowledge information represents the recruitment situation, industry standards, and job competency requirements for the workplace position, and the job-seeking information represents the candidate's personal situation when seeking employment; The knowledge graph construction module is used to construct a job knowledge graph based on the job knowledge information and a person profile knowledge graph based on the job seeker information; wherein, the job knowledge graph represents a graph constructed after analyzing the recruitment situation, industry standards, and job competency of the job; the person profile knowledge graph represents a graph constructed after analyzing the candidate's own situation. The interview question generation module is used to generate job interview questions based on the job knowledge graph, the person profile knowledge graph, and the large language model; wherein, the large language model represents a language model with analytical, creative, reasoning, and execution capabilities; The answer result determination module is used to determine the candidate's answer results to the job interview questions based on the job interview questions. The follow-up question generation module is used to generate follow-up interview questions for the target position based on the answers to the interview questions and the large language model when the answers to the interview questions do not meet the job requirements.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the knowledge graph-based interview question generation method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the knowledge graph-based interview question generation method as described in any one of claims 1 to 7.

Images