Methods and systems for intelligent interview recording conversion

Abstract

This application provides a method and system for intelligent interview recording conversion. The method collects interview data from multiple sources using a data acquisition module, resulting in richer and more comprehensive data. This provides more data for interview evaluation, improving the accuracy of the evaluation results. Through an AI large-scale model and pre-set engineer tags, engineer tag information is extracted from the interview data, thereby tagging the professional skills of the target engineer. This makes the engineer's ability assessment more professional and concise, avoiding the subjective influence and biased evaluation of human interviews, thus making the evaluation results more accurate. By mapping the target engineer's performance during the interview, including reaction time to impromptu questions, satisfaction with responses, and fluency of speech, a two-dimensional performance surface of the interview process is drawn. The interview score is then determined based on the size of this two-dimensional performance surface, making the interview results more intuitive.

Description

Technical Field

[0001] This application relates to the field of intelligent interview technology, and in particular to a method and system for intelligent interview recording conversion. Background Technology

[0002] As businesses grow and industry competition intensifies, the demand for labor and employee turnover are increasing rapidly, making employee recruitment a key factor in business development.

[0003] Traditional interview methods mostly rely on human interviews, which are highly subjective. Furthermore, time constraints and the fact that some interviewers are not good at expressing themselves make it difficult to gain a deep understanding of the candidate's professional skills and abilities. This can lead to biased and inaccurate interview results.

[0004] Therefore, how to solve the current low accuracy of talent interview assessment results has become an urgent technical problem. Summary of the Invention

[0005] This application provides a method and system for intelligent interview recording conversion, aiming to improve the accuracy of talent interview assessment results.

[0006] Firstly, this application also provides an intelligent interview recording conversion method for use in the intelligent interview recording conversion system described above, the intelligent interview recording conversion method comprising:

[0007] Obtain interview data from the target engineer;

[0008] Based on the AI ​​big data model, the interview data is transformed to obtain interview text data;

[0009] Based on the AI ​​big data model and preset engineer tags, keyword recognition is performed on the interview text data to obtain the engineer tag information of the target engineer;

[0010] Based on the engineer's tag information, the corresponding professional assessment questions are retrieved from the assessment database for the target engineer to take the assessment, thereby obtaining professional assessment data.

[0011] Based on the preset scoring criteria, the professional assessment data is converted into scores to obtain the level label of the target engineer. In a quadrant of the coordinate system, the level label is mapped to the first point d.

[0012] Obtain the target engineer's response time t and response satisfaction m to ad hoc questions. Using the formula tx1 + mx2 = f, obtain the ad hoc question responsiveness f. Map the ad hoc question responsiveness f to the second point w in the other quadrant of the coordinate system. Here, x1 and x2 are both coefficient values, where x2 is greater than x1 and x2 is greater than 0.5, x1 + x2 equals 1, and ad hoc questions are those not sent to the target engineer from the question list before the interview.

[0013] Obtain the speech fluency of the target engineer and map the speech fluency to the third point y in another quadrant of the coordinate system;

[0014] Based on the following formula, the two-dimensional performance surface s of the target engineer is derived.

[0015] s = (a × b) ÷ 2, where a is the distance between d and w, and b is the shortest distance between the line connecting d and w and y.

[0016] The target engineer's score is determined based on the size of the two-dimensional representation surface.

[0017] Secondly, this application provides an intelligent interview recording conversion system, the system comprising:

[0018] The interview module is used to acquire interview data of target engineers; based on an AI big data model, the interview data is transformed to obtain interview text data; based on the AI ​​big data model and preset engineer tags, keyword recognition is performed on the interview text data to obtain engineer tag information of the target engineers; based on the engineer tag information, corresponding professional assessment questions are retrieved from the assessment database for the target engineers to conduct assessments to obtain professional assessment data.

[0019] The scoring module is used to convert the professional assessment data into scores based on preset scoring criteria to obtain the target engineer's level label. In one quadrant of the coordinate system, the level label is mapped to the first point d. The module obtains the target engineer's response time t and response satisfaction m to ad-hoc questions. Using the formula tx1 + mx2 = f, the ad-hoc question responsiveness f is obtained and mapped to the second point w in another quadrant of the coordinate system. Here, x1 and x2 are coefficient values, where x2 is greater than x1 and greater than 0.5, and x1 + x2 equals 1. Ad-hoc questions are those not sent to the target engineer from the question list before the interview. The module obtains the target engineer's speech fluency and maps it to the third point y in another quadrant of the coordinate system. Based on the following formula, the target engineer's two-dimensional performance surface s is obtained: s = (a × b) ÷ 2, where a is the distance between d and w, and b is the shortest distance from the line connecting d and w to y. Based on the size of the two-dimensional performance surface, the target engineer's scoring result is determined.

[0020] This application provides an intelligent interview recording conversion method and system. The method collects interview data from multiple sources using a data acquisition module, resulting in richer and more comprehensive data. This provides more data for interview evaluation and improves the accuracy of the evaluation results. Through an AI large-scale model and pre-set engineer tags, the method extracts engineer tag information from the interview data, thereby tagging the professional skills of the target engineers. This makes the engineer ability assessment more professional and concise, avoiding the subjective influence and biased evaluation of human interviews, thus making the evaluation results more accurate. By mapping the target engineer's performance during the interview, including reaction time to impromptu questions, satisfaction with responses, and fluency of speech, a two-dimensional performance surface of the interview process is drawn. The interview score is then determined based on the size of this two-dimensional performance surface, making the interview results more intuitive and accurate. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating the first embodiment of an intelligent interview recording conversion method provided in this application.

[0023] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0026] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0027] This intelligent interview recording conversion system includes: The system includes:

[0028] The interview module is used to acquire interview data of target engineers; based on an AI big data model, the interview data is transformed to obtain interview text data; based on the AI ​​big data model and preset engineer tags, keyword recognition is performed on the interview text data to obtain engineer tag information of the target engineers; based on the engineer tag information, corresponding professional assessment questions are retrieved from the assessment database for the target engineers to conduct assessments to obtain professional assessment data.

[0029] The scoring module is used to convert the professional assessment data into scores based on preset scoring criteria to obtain the target engineer's level label. In one quadrant of the coordinate system, the level label is mapped to the first point d. The module obtains the target engineer's response time t and response satisfaction m to ad-hoc questions. Using the formula tx1 + mx2 = f, the ad-hoc question responsiveness f is obtained and mapped to the second point w in another quadrant of the coordinate system. Here, x1 and x2 are coefficient values, where x2 is greater than x1 and greater than 0.5, and x1 + x2 equals 1. Ad-hoc questions are those not sent to the target engineer from the question list before the interview. The module obtains the target engineer's speech fluency and maps it to the third point y in another quadrant of the coordinate system. Based on the following formula, the target engineer's two-dimensional performance surface s is obtained: s = (a × b) ÷ 2, where a is the distance between d and w, and b is the shortest distance from the line connecting d and w to y. Based on the size of the two-dimensional performance surface, the target engineer's scoring result is determined.

[0030] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a first embodiment of an intelligent interview recording conversion method provided in this application. This intelligent interview recording conversion method can be used in the intelligent interview recording conversion system described above.

[0031] like Figure 1 As shown, the intelligent interview recording conversion method includes steps S101 to S109.

[0032] S101. Obtain the interview data of the target engineer;

[0033] In one embodiment, the interview data includes interview dialogue data, self-assessment data, and peer assessment data.

[0034] In one embodiment, the target engineer can conduct a self-assessment of their abilities during the interview, including but not limited to skill areas, work location, work experience, project experience, education, special skills, and self-evaluation.

[0035] In one embodiment, the interview dialogue data can be the interview dialogue between the interviewer (such as HR or relevant department personnel) and the target engineer, including data such as telephone dialogue, email dialogue, and software chat dialogue, and can include voice dialogue data or text dialogue data.

[0036] In one embodiment, the evaluation data may include, but is not limited to, evaluation data of the target engineer by interviewers, former employees / friends / teachers, etc., such as evaluations of work ability, learning ability, work attitude, professional skills, life skills, communication skills, etc.

[0037] Furthermore, based on preset test tags, the target engineer is guided to select at least one self-test tag; based on the self-test content input by the target engineer according to their respective test tags, the self-test data is obtained.

[0038] In one embodiment, the preset test tags may include, but are not limited to, gender, education level, work experience, working hours, skills, areas of expertise, technical fields, and work location.

[0039] In one embodiment, each preset test label may also include more refined sub-labels. For example, the education label may include sub-labels such as associate degree, bachelor's degree, master's degree, doctoral degree and the institution attended; the work experience label may include sub-labels such as company, duration, position and projects participated in; and the professional skills label may include professional skills in multiple fields, such as professional skills certificates, skill recognition certificates / statements from authoritative institutions, etc.

[0040] In one embodiment, the target engineer can select the corresponding tags according to their actual situation, or input descriptive content, such as a description of the process of participating in the project, the difficulties encountered, and the solutions, or a self-evaluation. The descriptive content can be input in the form of voice or text.

[0041] S102. Based on the AI ​​big data model, the interview data is transformed to obtain interview text data;

[0042] In one embodiment, the large AI model can be an LLM (Large Language Model). A large language model is an artificial intelligence model designed to understand and generate human language. They are trained on massive amounts of text data and can perform a wide range of tasks, including text summarization, translation, sentiment analysis, and more. LLMs are characterized by their massive scale, containing billions of parameters that help them learn complex patterns in language data. These models are typically based on deep learning architectures, such as transformers, which contributes to their impressive performance on various NLP tasks.

[0043] In one embodiment, the large AI model may include a speech conversion sub-model and a text correction sub-model.

[0044] Furthermore, based on the speech conversion sub-model, the interview data is formatted to obtain initial text data; based on the text correction sub-model, the initial text data is corrected to obtain the interview text data.

[0045] In one embodiment, the interview data may include text data and voice data. The AI ​​big data model can convert the voice data into text data, and then process the text data, including but not limited to data cleaning, format conversion, text correction and feature extraction.

[0046] In one embodiment, the speech conversion sub-model performs format conversion on the interview data, such as converting speech data into text data to obtain initial text data; the speech conversion sub-model can also unify text data of different formats, such as translating words into Chinese or Chinese into English.

[0047] In one embodiment, the text correction sub-model can perform text correction on the initial text data. These errors can include spelling errors, grammatical errors, semantic errors, or entity errors. Specifically, spelling correction aims to correct spelling errors in words; grammatical correction aims to correct grammatical errors in sentences; semantic correction aims to correct semantic errors in sentences to make them more consistent with common sense and context; and entity correction aims to correct entity errors in the text, such as names of people, places, and technical terms. For example, the target engineer may have their own descriptive habits for certain technical terms, such as referring to "large language model" as "large model." The text correction sub-model can then correct "large model" to "large language model."

[0048] In one embodiment, the text correction sub-model can perform text semantic reasoning based on contextual information, thereby improving the accuracy of text correction. Theoretically, all text content (including but not limited to images, text, etc.) other than the parsed current sentence can be referred to as context. For a certain type of error, the text correction sub-model learns the vector representation of the target word's context, and then predicts the target word using the context vector. If the prediction result differs from the original target word, the original target word is marked as an error, and the prediction result is used for correction.

[0049] In one embodiment, engineer data of at least one preset engineer is obtained; based on expert prior knowledge, tag keywords are extracted from the engineer data as preset engineer tags corresponding to each preset engineer.

[0050] In one embodiment, engineer data of various preset engineers can be collected on platforms such as the Internet and talent databases. Engineer data may include information such as engineer name, engineer skills, and engineer job functions. Then, based on expert prior knowledge, tag keywords are extracted as tag information corresponding to each engineer, and preset engineer tags corresponding to each preset engineer are obtained.

[0051] In one embodiment, the preset engineer tags can be classified into multi-level tags based on the prior knowledge of experts. For example, the first-level tags corresponding to hardware development engineers may include regional tags, professional tags, education tags, skill tags, etc. The professional tags may include second-level tags such as professional type, professional name, professional level, etc. The second-level tags can be further refined into more refined third-level tags; and so on, to achieve structured management of tag information.

[0052] S103. Based on the AI ​​big data model and preset engineer tags, perform keyword recognition on the interview text data to obtain the engineer tag information of the target engineer;

[0053] In one embodiment, the AI ​​big data model performs keyword matching on interview text data based on preset engineer tags, extracting tag keywords from the interview text data as the engineer tag information for the target engineer. For example, work experience includes the names of companies worked for, job positions, salary, and colleague evaluations; education tags include educational information, school information, campus experience information (such as class work information, club experience information, school research project information, competition information, etc.), and major information.

[0054] In one embodiment, the AI ​​big data model can classify and organize the extracted tag keywords according to preset engineer tags. Each preset engineer tag can contain multiple sub-tags, realizing structured tag management and generating structured engineer tag information for the target engineer.

[0055] S104. Based on the engineer's tag information, retrieve the corresponding professional assessment questions from the assessment database for the target engineer to conduct the assessment and obtain professional assessment data.

[0056] In one embodiment, the target engineer can independently apply for engineer level certification, and undergo skills and level assessments based on the target engineer's current engineer tag information. After successful verification, the selected skills and level information are added to the engineer's tag information.

[0057] In one embodiment, assessment questions can be generated by organizing the assessment content corresponding to different professional skills and different levels based on expert prior knowledge or databases, and the assessment questions can be stored in the assessment database.

[0058] In one embodiment, when applying for an assessment, the target engineer can choose the assessment items they need to be assessed on, such as professional skills assessment; the AI ​​big data model queries the corresponding assessment questions from the assessment database based on the target engineer's professional skills type, name, and current professional skills level.

[0059] In one embodiment, the number of assessment questions can be preset, such as one hundred questions, and then divided into different question types, such as multiple choice questions, fill-in-the-blank questions, and scenario-based assessment questions. The number of questions and point values ​​for each question type can be allocated. The AI ​​big data model randomly selects corresponding assessment questions from the assessment database according to preset question rules, forming an assessment paper, which is then provided to the target engineer for online or offline assessment. The target engineer is prompted to complete the assessment within a specified time limit and submit the assessment results to obtain professional assessment data.

[0060] S105. Based on the preset scoring criteria, the professional assessment data is converted into scores to obtain the grade label of the target engineer. In a quadrant of the coordinate system, the grade label is mapped to the first point d.

[0061] In one embodiment, different test question types and scoring standards can be preset. The AI ​​big data model analyzes and autonomously grades the professional assessment data submitted by the target engineer, converts the professional assessment data into scores, and presents the professional skills assessment results of the target engineer in a scoring manner.

[0062] In one embodiment, a passing threshold can be set, such as a percentage system, requiring a score of 90 or higher to pass the professional skills assessment, and a score below 90 to fail.

[0063] In one embodiment, if the target engineer passes the professional assessment, the level label of the professional skills in which the target engineer participated in the assessment is modified and its level label is upgraded; if the target engineer fails the professional assessment, the original level label is maintained.

[0064] Furthermore, based on the engineer tag information and the level tag, an engineer tag information table corresponding to the target engineer is created; the engineer tag information table is updated with tag information according to a preset period.

[0065] In one embodiment, based on engineer tag information and level information, a unique engineer tag information table can be created for each target engineer, generating an engineer information card. The engineer tag information table may include data such as the engineer's personal information, professional skills information, and professional level information.

[0066] In one embodiment, the data in the engineer tag information table can be changed at any time. When the AI ​​model receives an engineer's assessment application, it can lock the engineer's engineer tag information table and then update the information in the engineer tag information table according to the engineer's assessment results. For example, if the assessment is passed, the corresponding level tag is changed, and the assessment items and passing score can be recorded. If the assessment fails, the assessment items and score can also be recorded.

[0067] In one embodiment, the engineer tag information table can be updated according to a preset period. For example, every month, the AI ​​big data model can assess the engineer's capabilities based on information such as the engineer's work content, work results, and work feedback within that period, and then update the engineer tag information table based on the assessment results.

[0068] In one embodiment, when there is new tag data corresponding to the target engineer, the tag information table is updated based on the new tag data.

[0069] In one embodiment, the engineer tag information table can also be updated based on newly added tag data. For example, each time an engineer completes a work task, the tag information of that work task is obtained, including but not limited to work content, work difficulty, completion result, evaluation feedback, etc. Then, based on this tag information, the engineer's ability is evaluated, and the engineer tag information table is updated based on the evaluation results.

[0070] Further, the level scores of each level tag in the engineer tag information table are obtained; when the level score reaches the preset threshold corresponding to the level tag, a level assessment is initiated; based on the assessment results of the target engineer based on the level assessment, the level tag of the target engineer is reassessed to determine the current level tag of the target engineer; when the current level tag is inconsistent with the level tag, the level tag in the engineer tag information table is updated based on the current level tag.

[0071] In one embodiment, the engineer tag information table may include multiple level tags, such as computer professional skill level, software operation level, hardware equipment operation level, solution design level, and scenario solution level. The engineer level tag can be determined based on the weighted result of multiple skill level tags.

[0072] In one embodiment, when an engineer completes a task, the AI ​​big data model can evaluate the engineer's task performance based on information such as task type, task difficulty, participants, task completion results, and user feedback, score the engineer's various abilities, obtain points for different professional skills, and obtain a level score for each level label.

[0073] In one embodiment, when the grade score corresponding to the grade label reaches the preset threshold corresponding to the grade label, the grade assessment can be initiated automatically to the engineer, who can choose to accept or postpone it; or the engineer can initiate the grade assessment application independently.

[0074] In one embodiment, based on the level label corresponding to the level assessment that the engineer needs to undergo, that is, the level label corresponding to the professional skill assessment, corresponding test questions are generated, and then the assessment results of the target engineer are obtained. If the test is passed, it is considered that the level label of the target engineer's professional skill can be improved, and the level label of the corresponding professional skill in the engineer's label information is changed.

[0075] In one embodiment, if all skill level tags for the target engineer's current engineer level meet the minimum requirements for that level, an engineer level assessment application can be initiated. Correspondingly, if the assessment is passed, the target engineer's engineer level can be upgraded.

[0076] In one embodiment, a two-dimensional coordinate system is created, the first quadrant of the two-dimensional coordinate system is used to represent the grade label, and the grade label of the target engineer is mapped to the first quadrant of the two-dimensional coordinate system as the first point d.

[0077] S106. Obtain the target engineer's response time t and response satisfaction m to the ad hoc question. According to the formula tx1+mx2=f, obtain the ad hoc question responsiveness f. Map the ad hoc question responsiveness f to the second point w in the other quadrant of the coordinate system. Here, x1 and x2 are both coefficient values, where x2 is greater than x1 and x2 is greater than 0.5, x1+x2 equals 1, and the ad hoc question is a question in the question list not sent to the target engineer before the interview.

[0078] In one embodiment, the target engineer's response time and satisfaction with the response to ad hoc problems can reflect the target engineer's professional competence, emergency response capabilities, etc. By assessing the target engineer's response time and satisfaction with the response to ad hoc problems, it is possible to determine which positions the target engineer is suitable for and which positions are not suitable for.

[0079] In one embodiment, the temporary problem responsiveness is calculated according to the formula tx1+mx2=f, and f is mapped to the second quadrant of the two-dimensional coordinate system to obtain the second point w.

[0080] S107. Obtain the speech fluency of the target engineer and map the speech fluency to the third point y in another quadrant of the coordinate system.

[0081] In one embodiment, speech fluency can determine the target engineer's mastery of professional knowledge, language communication ability, and language organization ability. Speech fluency can be evaluated based on speech rate, pause frequency, repetition frequency, and speech clarity, and mapped to the third quadrant in a two-dimensional coordinate system to obtain a third point y.

[0082] S108. Based on the following formula, the two-dimensional performance surface s of the target engineer is obtained.

[0083] s = (a × b) ÷ 2, where a is the distance between d and w, and b is the shortest distance between the line connecting d and w and y.

[0084] In one embodiment, d, w, and y are in the same two-dimensional coordinate system and are located in three different quadrants. The three can form a triangle to obtain the two-dimensional representation surface s of the target engineer.

[0085] S109. Determine the target engineer's score based on the size of the two-dimensional representation surface.

[0086] In one embodiment, the area of ​​the two-dimensional surface s is calculated as the scoring result for the target engineer.

[0087] In one embodiment, the scoring results can be divided into multiple levels based on the area of ​​the two-dimensional surface. For example, if the area is smaller than a certain threshold, the scoring result is E, which means unqualified. The scores above the threshold are qualified. The scores above the threshold can be divided into multiple threshold ranges, corresponding to D (qualified), C (average), B (good), A (excellent), S (extremely excellent), etc.

[0088] Furthermore, based on emotion recognition algorithms, multiple emotional stages throughout the target engineer's interview process are obtained, including happy, calm, conflicted, sad, and depressed stages.

[0089] Obtain the duration c of the happy mood phase, and use the value of duration c as the coordinate point e on the z-axis of the coordinate system;

[0090] The overall interview value k is derived from the following formula.

[0091] Based on the K-clustering algorithm, discrete overall interview values ​​k are obtained, and the lowest overall interview value k is removed to output the recommended engineer results.

[0092] In one embodiment, based on an emotion recognition algorithm, the duration *c* of the target engineer maintaining a happy emotional state during the interview is accumulated, and the overall interview score is calculated based on the two-dimensional performance surfaces *s* and *c*. Using a K-clustering algorithm, the discrete distribution of the overall interview score *k* for all engineers is calculated. Based on this discrete distribution, the evaluation results of all engineers are ranked, and the best-performing engineers are selected for recommendation, while those that do not meet the requirements are removed—that is, the engineers with the lowest discrete distribution of the overall interview score *k*.

[0093] This embodiment provides an intelligent interview recording conversion method. This method collects interview data from multiple sources using a data acquisition module, resulting in richer and more comprehensive data. This provides more data for interview evaluation, improving the accuracy of the evaluation results. Through an AI large-scale model and preset engineer tags, engineer tag information is extracted from the interview data, thereby tagging the professional skills of the target engineer. This makes the engineer's ability assessment more professional and concise, avoiding the subjective influence and biased evaluation of human interviews, thus making the interview evaluation results more accurate. By mapping the target engineer's performance during the interview, including reaction time to impromptu questions, satisfaction with responses, and fluency of speech, a two-dimensional performance surface of the target engineer's interview process is drawn. The interview score is then determined based on the size of this two-dimensional performance surface, making the interview results more intuitive.

[0094] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps:

[0095] Obtain interview data from the target engineer;

[0096] Based on the AI ​​big data model, the interview data is transformed to obtain interview text data;

[0097] Based on the AI ​​big data model and preset engineer tags, keyword recognition is performed on the interview text data to obtain the engineer tag information of the target engineer;

[0098] Based on the engineer's tag information, the corresponding professional assessment questions are retrieved from the assessment database for the target engineer to take the assessment, thereby obtaining professional assessment data.

[0099] Based on the preset scoring criteria, the professional assessment data is converted into scores to obtain the level label of the target engineer. In a quadrant of the coordinate system, the level label is mapped to the first point d.

[0100] Obtain the target engineer's response time t and response satisfaction m to ad hoc questions. Using the formula tx1 + mx2 = f, obtain the ad hoc question responsiveness f. Map the ad hoc question responsiveness f to the second point w in the other quadrant of the coordinate system. Here, x1 and x2 are both coefficient values, where x2 is greater than x1 and x2 is greater than 0.5, x1 + x2 equals 1, and ad hoc questions are those not sent to the target engineer from the question list before the interview.

[0101] Obtain the speech fluency of the target engineer and map the speech fluency to the third point y in another quadrant of the coordinate system;

[0102] Based on the following formula, the two-dimensional performance surface s of the target engineer is derived.

[0103] s = (a × b) ÷ 2, where a is the distance between d and w, and b is the shortest distance between the line connecting d and w and y.

[0104] The target engineer's score is determined based on the size of the two-dimensional representation surface.

[0105] In one embodiment, the processor is used to run a computer program stored in a memory and to implement an emotion recognition algorithm to obtain multiple emotional stages throughout the interview process of the target engineer, wherein the emotional stages include a happy stage, a calm stage, a conflicted stage, a sad stage, and a depressed stage.

[0106] Obtain the duration c of the happy mood phase, and use the value of duration c as the coordinate point e on the z-axis of the coordinate system;

[0107] The overall interview value k is derived from the following formula.

[0108] Based on the K-clustering algorithm, discrete overall interview values ​​k are obtained, and the lowest overall interview value k is removed to output the recommended engineer results.

[0109] In one embodiment, after the processor performs the step of converting the professional assessment data into scores based on a preset scoring standard to obtain the target engineer's level label, it is further configured to:

[0110] Based on the engineer tag information and the level tag, create an engineer tag information table corresponding to the target engineer;

[0111] The engineer tag information table is updated according to a preset period.

[0112] In one embodiment, after implementing the creation of the engineer tag information table corresponding to the target engineer based on the engineer tag information and the level tag, the processor is further configured to implement:

[0113] When new tag data corresponding to the target engineer exists, the tag information of the engineer tag information table is updated based on the new tag data.

[0114] In one embodiment, after updating the tag information in the engineer tag information table, the processor is further configured to:

[0115] Obtain the grade rating of each grade tag in the engineer tag information table;

[0116] When the grade score reaches the preset threshold corresponding to the grade label, a grade assessment is initiated.

[0117] Based on the assessment results of the target engineer based on the level evaluation, the level label of the target engineer is re-evaluated to determine the current level label of the target engineer;

[0118] When the current level label is inconsistent with the level label, the level label of the engineer label information table is updated based on the current level label.

[0119] In one embodiment, the large AI model includes a speech conversion sub-model and a text correction sub-model;

[0120] When the processor performs data transformation on the interview data based on the AI ​​large model to obtain interview text data, it is used to:

[0121] Based on the speech conversion sub-model, the interview data is formatted to obtain initial text data;

[0122] Based on the text correction sub-model, the initial text data is corrected to obtain the interview text data.

[0123] In one embodiment, the interview data includes interview dialogue data, self-assessment data, and peer assessment data.

[0124] In one embodiment, when the processor acquires the interview data of the target engineer, it is configured to:

[0125] Based on preset test tags, the target engineer is guided to select at least one self-test tag;

[0126] The self-test data is obtained based on the self-test content input by the target engineers according to their respective test tags.

[0127] In one embodiment, before implementing the process of performing keyword recognition on the interview text data based on the AI ​​big data model and preset engineer tags to obtain the engineer tag information of the target engineer, the processor is further configured to implement:

[0128] Obtain engineer data for at least one preset engineer;

[0129] Based on expert prior knowledge, tag keywords are extracted from the engineer data and used as the preset engineer tags corresponding to each preset engineer.

[0130] The embodiments of this application also provide a computer-readable storage medium storing a computer program, the computer program including program instructions, and the processor executing the program instructions to implement any of the intelligent interview recording conversion methods provided in the embodiments of this application.

[0131] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMediaCard (SMC), SecureDigital (SD) card, or FlashCard equipped on the computer device.

[0132] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligently converting interview recordings, characterized in that, The intelligent interview recording conversion method includes: Obtain interview data from the target engineer; Based on the AI ​​big data model, the interview data is transformed to obtain interview text data; Based on the AI ​​big data model and preset engineer tags, keyword recognition is performed on the interview text data to obtain the engineer tag information of the target engineer; Based on the engineer's tag information, the corresponding professional assessment questions are retrieved from the assessment database for the target engineer to take the assessment, thereby obtaining professional assessment data. Based on the preset scoring criteria, the professional assessment data is converted into scores to obtain the level label of the target engineer. In a quadrant of the coordinate system, the level label is mapped to the first point d. Obtain the target engineer's response time t and response satisfaction m to ad hoc questions. Using the formula tx1 + mx2 = f, obtain the ad hoc question responsiveness f. Map the ad hoc question responsiveness f to the second point w in the other quadrant of the coordinate system. Here, x1 and x2 are both coefficient values, where x2 is greater than x1 and x2 is greater than 0.5, x1 + x2 equals 1, and ad hoc questions are those not sent to the target engineer from the question list before the interview. Obtain the speech fluency of the target engineer and map the speech fluency to the third point y in another quadrant of the coordinate system; Based on the following formula, the two-dimensional performance surface s of the target engineer is derived. s = (a × b) ÷ 2, where a is the distance between d and w, and b is the shortest distance between the line connecting d and w and y. The target engineer's score is determined based on the size of the two-dimensional representation surface.

2. The intelligent interview recording conversion method according to claim 1, characterized in that, It also includes using emotion recognition algorithms to capture multiple emotional stages throughout the interview process with the target engineer. These emotional stages include happy, calm, conflicted, sad, and depressed stages. Obtain the duration c of the happy mood phase, and use the value of duration c as the coordinate point e on the z-axis of the coordinate system; The overall interview value k is derived from the following formula. Based on the K-clustering algorithm, discrete overall interview values ​​k are obtained, and the lowest overall interview value k is removed to output the recommended engineer results.

3. The intelligent interview recording conversion method according to claim 2, characterized in that, After converting the professional assessment data into scores based on preset scoring standards to obtain the target engineer's level label, the process further includes: Based on the engineer tag information and the level tag, create an engineer tag information table corresponding to the target engineer; The engineer tag information table is updated according to a preset period.

4. The intelligent interview recording conversion method according to claim 3, characterized in that, After creating the engineer tag information table corresponding to the target engineer based on the engineer tag information and the level tag, the method further includes: When new tag data corresponding to the target engineer exists, the tag information of the engineer tag information table is updated based on the new tag data.

5. The intelligent interview recording conversion method according to claim 4, characterized in that, After updating the engineer tag information table, the process further includes: Obtain the grade rating of each grade tag in the engineer tag information table; When the grade score reaches the preset threshold corresponding to the grade label, a grade assessment is initiated. Based on the assessment results of the target engineer based on the level evaluation, the level label of the target engineer is re-evaluated to determine the current level label of the target engineer; When the current level label is inconsistent with the level label, the level label of the engineer label information table is updated based on the current level label.

6. The intelligent interview recording conversion method according to claim 1, characterized in that, The AI ​​big model includes a speech conversion sub-model and a text error correction sub-model; The AI-based large-scale model is used to transform the interview data to obtain interview text data, including: Based on the speech conversion sub-model, the interview data is formatted to obtain initial text data; Based on the text correction sub-model, the initial text data is corrected to obtain the interview text data.

7. The intelligent interview recording conversion method according to claim 1, characterized in that, The interview data includes interview dialogue data, self-assessment data, and peer assessment data.

8. The intelligent interview recording conversion method according to claim 7, characterized in that, The acquisition of the target engineer's interview data includes: Based on preset test tags, the target engineer is guided to select at least one self-test tag; The self-test data is obtained based on the self-test content input by the target engineers according to their respective test tags.

9. The intelligent interview recording conversion method according to claim 1, characterized in that, Before obtaining the engineer tag information of the target engineer by performing keyword recognition on the interview text data based on the AI ​​big data model and preset engineer tags, the method further includes: Obtain engineer data for at least one preset engineer; Based on expert prior knowledge, tag keywords are extracted from the engineer data and used as the preset engineer tags corresponding to each preset engineer.

10. An intelligent interview recording conversion system, characterized in that, The system includes: The interview module is used to acquire interview data of target engineers; based on an AI big data model, the interview data is transformed to obtain interview text data; based on the AI ​​big data model and preset engineer tags, keyword recognition is performed on the interview text data to obtain engineer tag information of the target engineers; based on the engineer tag information, corresponding professional assessment questions are retrieved from the assessment database for the target engineers to conduct assessments to obtain professional assessment data. The scoring module is used to convert the professional assessment data into scores based on preset scoring criteria to obtain the target engineer's level label. In one quadrant of the coordinate system, the level label is mapped to the first point d. The module obtains the target engineer's response time t and response satisfaction m to ad-hoc questions. Using the formula tx1 + mx2 = f, the ad-hoc question responsiveness f is obtained and mapped to the second point w in another quadrant of the coordinate system. Here, x1 and x2 are coefficient values, where x2 is greater than x1 and greater than 0.5, and x1 + x2 equals 1. Ad-hoc questions are those not sent to the target engineer from the question list before the interview. The module obtains the target engineer's speech fluency and maps it to the third point y in another quadrant of the coordinate system. Based on the following formula, the target engineer's two-dimensional performance surface s is obtained: s = (a × b) ÷ 2, where a is the distance between d and w, and b is the shortest distance from the line connecting d and w to y. Based on the size of the two-dimensional performance surface, the target engineer's scoring result is determined.

Images