AI interviewer-based intention recognition method and device, electronic equipment and medium

Abstract

The invention provides an intention recognition method and device based on an AI interviewer, electronic equipment and a medium. The method comprises the steps of obtaining an interview utterance text of an interview candidate; performing text similarity matching on the interview utterance text and an example utterance set of a preset intention tag, and identifying at least one initial intention and determining a first matching score of each initial intention based on a similarity matching result; splitting the interview utterance text to obtain a keyword associated with each initial intention, and calculating a second matching score of each initial intention based on the vocabulary meaning of the keyword; identifying an association intention of the interview utterance text through a knowledge graph engine; determining a third matching score of each initial intention based on the relevance between the associated intention and each initial intention; and determining a comprehensive score based on the first matching score, the second matching score and the third matching score of each initial intention, and determining a winning intention based on the comprehensive score. According to the invention, the accuracy and reliability of intention recognition are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent interview technology, and in particular to an intention recognition method, device, electronic device and medium based on an AI interviewer. Background Art

[0002] In today's digital age, virtual assistants have been widely used in many fields such as AI (Artificial Intelligence) interviews, customer service, and smart offices. Today, companies are increasingly relying on virtual assistants, and their requirements for performance and accuracy are becoming more stringent. For example, in an AI interview scenario, the virtual interviewer must accurately grasp the company's recruitment profile and accurately determine whether the candidate's answer matches the recruitment profile, which requires the virtual interviewer to accurately identify the candidate's answer intention.

[0003] However, at present, AI interviewers can only identify the intentions of candidates from the superficial meaning of their answers, resulting in low accuracy and precision of intention recognition, which in turn leads to a low match between the interviewee and the company's recruitment profile. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide an intention recognition method, device, electronic device and medium based on AI interviewer to improve the accuracy of intention recognition and thereby improve the matching degree between interviewers and enterprises.

[0005] In a first aspect, an intention recognition method based on an AI interviewer is provided, which is applied to an intention recognition engine. The intention recognition engine includes a machine learning engine, a basic meaning engine, and a knowledge graph engine. The method includes: Obtain the interview speech text of the interview candidate; The interview speech text is matched with the sample speech set of preset intent labels by the machine learning engine for text similarity, and at least one initial intent is identified based on the similarity matching result and a first matching score of each initial intent is determined; the preset intent labels are predefined based on the recruitment needs of the enterprise; The interview discourse text is disassembled by the basic meaning engine to obtain the keywords associated with each initial intent, and the second matching score of each initial intent is calculated based on the lexical meaning of the keywords; Identify the associated intent of the interview discourse text through the knowledge graph engine; determine the third matching score of each initial intent based on the relevance of the associated intent to each initial intent; A comprehensive score is determined based on the first matching score, the second matching score, and the third matching score of each initial intent, and a winning intent is determined based on the comprehensive score.

[0006] Optionally, when the number of intentions identified based on the similarity matching results is multiple, the method further includes: The initial intent is filtered based on the preset intent exclusion rule to obtain the filtered initial intent.

[0007] Optionally, the preset intention exclusion rule includes at least a first exclusion rule and a second exclusion rule: The first exclusion rule is to prioritize the exclusion of intents based on entity value matching; entity values ​​include at least dates and numbers; The second exclusion rule is to determine the intent match type based on the first match score of the initial intent, where the intent match type includes possible match and confirmed match; When multiple initial intents include both types, exclude the intent matching type as the initial intent that may be matched.

[0008] Optionally, calculating a second matching score for each initial intent based on the lexical meaning of the keyword includes: Determine the original role score of the keyword, the original role represents the role or function of the word in its original form; Determine the role score of the keyword in the sentence. The role in the sentence represents the role of the word in the entire sentence structure. Determine the word status score of the keyword after preprocessing; the preprocessing includes synonym replacement, word form restoration and standardization; The second matching score of each initial intent is calculated based on the original word role score of the keyword, the role score in the sentence, the preprocessed word status score, the weight value corresponding to each score, and the preset bonus items and penalty items.

[0009] Optionally, identifying the associated intent of the interview discourse text by a knowledge graph engine; and determining a third matching score of each initial intent based on the relevance of the associated intent to each initial intent includes: Extracting term sets from interview discourse text; Based on the term set, search for a set of paths that can associate these terms in the pre-built knowledge graph of the target industry field; Determine the association intention based on the direction of the path set; A third matching score for each initial intent is determined based on a similarity between the associated intent and each initial intent.

[0010] Optionally, determining a comprehensive score based on the first matching score, the second matching score, and the third matching score of each initial intent, and determining a winning intent based on the comprehensive score includes: Determining a first confidence level for each initial intent based on a machine learning engine; determining a second confidence level for each initial intent based on the base meaning engine; Determine the third confidence level of each initial intent based on the knowledge graph engine; Calculating a comprehensive score based on the first matching score, the second matching score, the third matching score, the first confidence level, the second confidence level, the third confidence level, and a preset weight of each engine; The comprehensive score is compared with the preset score threshold. If it exceeds the preset score threshold, it is determined to be a winning intention.

[0011] Optionally, after determining the winning intention, the method further includes: Determine the match type of the winning intent based on the comprehensive score of each winning intent, where the match type includes possible match and confirmed match; When the number of winning intentions is 1, if the matching type of the winning intention is a confirmed match, the feedback information of the winning intention is directly output; if the matching type of the winning intention is a possible match, the winning intention is output in the form of a preset dialog box for confirmation by the user; When the number of winning intentions is 2 or more, if the matching type of the winning intention is multiple confirmed matches, all the winning intentions are output for the user to select; if the matching type of the winning intention is multiple possible matches, the winning intention is output in the form of a preset dialog box for user confirmation; if the matching type of the winning intention includes possible matches and confirmed matches, feedback information of the winning intention corresponding to the confirmed matches is output.

[0012] In the second aspect, an intention recognition system based on an AI interviewer is provided, the system comprising: An acquisition unit, used for acquiring the interview speech text of the interview candidate; A machine learning engine is used to perform text similarity matching between the interview utterance text and a set of example utterances with preset intent labels, and to identify at least one initial intent based on the similarity matching result and determine a first matching score for each initial intent; the preset intent labels are predefined based on the recruitment needs of the enterprise; The basic meaning engine is used to disassemble the interview discourse text to obtain keywords associated with each initial intent, and calculate the second matching score of each initial intent based on the lexical meaning of the keywords; A knowledge graph engine is used to identify the associated intent of the interview discourse text; determine the third matching score of each initial intent based on the relevance of the associated intent to each initial intent; The determination unit is used to determine a comprehensive score based on the first matching score, the second matching score and the third matching score of each initial intention, and determine a winning intention based on the comprehensive score.

[0013] In a third aspect, an electronic device is provided, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory, used to store computer programs; The processor is used to implement any method step described in the first aspect when executing a program stored in the memory.

[0014] According to a fourth aspect, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any method step described in the first aspect is implemented.

[0015] The embodiment of the present invention provides an intention recognition method, device, electronic device and medium based on AI interviewer, by obtaining the interview speech text of the interview candidate; by using a machine learning engine to match the interview speech text with the sample speech set of preset intention labels for text similarity, and based on the similarity matching result, identify at least one initial intention and determine the first matching score of each initial intention; the preset intention label is predefined based on the recruitment needs of the enterprise; the interview speech text is disassembled by the basic meaning engine to obtain the keywords associated with each initial intention, and the second matching score of each initial intention is calculated based on the lexical meaning of the keywords; the associated intention of the interview speech text is identified by the knowledge graph engine; the third matching score of each initial intention is determined based on the correlation between the associated intention and each initial intention; the comprehensive score is determined based on the first matching score, the second matching score and the third matching score of each initial intention, and the winning intention is determined based on the comprehensive score. The present invention closely cooperates with three engines in the AI ​​interviewer scenario, is associated with each other, and is progressive, processes user answers from different levels and angles, and gradually digs into user intentions, significantly improves the accuracy and reliability of intention recognition, and lays a solid foundation for the efficient and accurate evaluation of AI interviewers.

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 A flowchart of an AI interviewer-based intention recognition method provided by an embodiment of the present invention is shown; Figure 2 A schematic diagram of the structure of an AI interviewer-based intention recognition device provided by an embodiment of the present invention is shown; Figure 3 A schematic structural diagram of an electronic device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0020] Considering that currently AI interviewers can only identify the intentions of candidates from the superficial meaning of their answers, the accuracy and precision of intent recognition are low, which in turn leads to a low match between the interviewee and the company's recruitment profile.

[0021] Based on this, an embodiment of the present invention provides an intention recognition method and system based on an AI interviewer, which is described below through an embodiment.

[0022] An embodiment of the present invention provides an intention recognition method based on an AI interviewer. The method is applied to an intention recognition engine, which consists of three parts: a machine learning engine, a basic meaning engine, and a knowledge graph engine. These three engines are tested and optimized in advance to achieve the required intention recognition accuracy before being deployed and applied.

[0023] The following is an explanation of the test optimization process of the intent recognition engine: First, log in to the pre-built test and tuning platform, select the intent recognition engine to be tested on the test and tuning platform, and then click Test-Utterance Test in the left menu. Then select the engine corresponding to the test utterance as needed.

[0024] Taking machine engine learning as an example, first select the machine learning engine in the engine selection menu drop-down item, and type the user utterance to be tested in the input user utterance box, such as "Are you familiar with the Java language?" The machine learning engine will output the test results. The test results will be presented as single, multiple, or no matching intent.

[0025] When the test result is a single matching intent, the single intent that matches the user utterance is displayed below the input user utterance field. If the tester determines that the match is correct, the tester can continue to test to improve the intent matching score. If it is incorrect, the tester can mark and select the correct intent and continue testing.

[0026] When the test result is multiple matching intents, the tester can implement training by clicking the radio button of one of the matching intents from the multiple matching intents.

[0027] When the test result is no matching intent, you need to select an intent and train it to match the user's words.

[0028] The machine learning model is continuously tested and optimized through a large number of test utterances, so that the matching score of the intent it outputs is continuously improved until it reaches the preset matching score.

[0029] The basic meaning engine and knowledge graph engine can be tested in the same way.

[0030] Among them, when training the knowledge graph engine, for AI interviewers, common interview questions and their answers can be used as part of the knowledge graph. For example, for the common question "Please introduce your strengths and weaknesses", by setting terms (such as "strengths", "disadvantages"), term configuration or categories from the FAQ page, and training the knowledge graph, when the interviewee answers similar questions, it can better match the intent and make a reasonable evaluation. Or add the interviewee's answer as an alternative question to the common questions selected on the knowledge graph page, further train the knowledge graph, and improve the model's ability to recognize the intent of various answers in interview scenarios.

[0031] In addition, when testing user utterances, an intent analysis box is also provided on the test tuning platform, in which a quick overview of the shortlisted intents, the called engines, the corresponding intent matching scores, and the final winning intent can be obtained.

[0032] Next, the intent recognition engine optimized by testing is used to perform intent recognition. The embodiment of the present invention provides an intent recognition method based on an AI interviewer, such as Figure 1 As shown, the method includes: Step S101: Obtain the interview speech text of the interview candidate.

[0033] In this step, in the AI ​​interview scenario, in a virtual interview room, the candidate accesses the interview through a video link, and the AI ​​interviewer's virtual image is displayed on the screen. The interface is simple and intuitive, with a timer, question display area, answer input box, etc. The interview begins, and the AI ​​interviewer opens with: "Welcome to this software development engineer position interview. Next, we will start the professional skills assessment session. Please be prepared." The interview candidate begins to answer questions based on the AI ​​interviewer's questions.

[0034] In one example, the voice of the interview candidate or directly input text information may be obtained. If it is voice, the voice may be further converted into text information to obtain the interview utterance text.

[0035] In order to improve the quality of the interview text, the interview text can be preprocessed before intent recognition, such as removing irrelevant information: First, all information irrelevant to the interview content needs to be removed, such as background noise, non-verbal communication (such as coughing, laughing), etc. If it is in text form, any unnecessary comments or tags need to be deleted. Another example is to standardize the text format: unify the text format, including font, capitalization, etc. For example, convert all text to lowercase to avoid repeated word counting due to different capitalization.

[0036] Step S102: Perform text similarity matching between the interview utterance text and a set of example utterances with preset intent labels through a machine learning engine, and identify at least one initial intent based on the similarity matching result and determine a first matching score for each initial intent.

[0037] In this step, the preset intent tags are predefined based on the recruitment needs of the enterprise, such as the intent tags of "introducing work experience" and "elaborating project details".

[0038] In the embodiment of the present invention, the interview speech text and the sample speech set are firstly vectorized using the TF-IDF algorithm, and then similarity matching is performed using cosine similarity, and the first matching score of each initial intent is determined based on the cosine similarity score.

[0039] In a specific example, suppose the user input utterance is , the preset intent tag set is ; The sample utterance set is , first for all example utterances and user input utterance Perform vectorization. Is the user input utterance The TF-IDF vector, Is the default intent tag Middle Example utterances The TF-IDF vector.

[0040] Calculate the cosine similarity score between the user input utterance and all example utterances under one of the preset intent labels, and take the maximum similarity score as the matching score of the preset intent recognition label : (1); in, is the formula for calculating the cosine similarity of two vectors, It is the code name for the machine learning engine.

[0041] After calculating the scores of the preset intent tags, by comparing with the preset first matching score threshold, the intent corresponding to the preset intent tag that is greater than the first matching score threshold is determined as the initial intent.

[0042] In a specific example, when the AI ​​interviewer asked, "Please describe a product you are most familiar with, and analyze its core value and target user group," and the interview candidate answered with Douyin as an example, the machine learning engine used the TF-IDF algorithm to vectorize the user's answer and the sample utterances corresponding to a large number of predefined intent tags and calculated the similarity. It quickly scans through the massive sample utterance data, preliminarily screens out the intent directions that may match the user's answer, obtains preliminary identification intents, such as "product description", "core value analysis", "target user positioning", etc., and gives the corresponding matching score.

[0043] Step S103: Decompose the interview discourse text through the basic meaning engine to obtain keywords associated with each initial intent, and calculate the second matching score of each initial intent based on the lexical meaning of the keywords.

[0044] The machine learning engine can only judge the surface similarity of the text, and it is difficult to deeply explore the complex relationships and meanings behind the semantics. Therefore, this step further analyzes the semantics of the user's speech through the basic meaning engine, and judges the user's intention at a closer level to improve the accuracy of intent recognition.

[0045] For example, when the machine learning engine initially determines that the user's answer is related to the "core value analysis" intention, the basic meaning engine further analyzes the close connection between the original words mentioned by the user, such as "convenience, richness, and creativity", and the core value description of "short video creation and sharing platform", to confirm the accuracy of the core value analysis intention.

[0046] The specific implementation process will be described in the following embodiments and will not be repeated here.

[0047] Step S104: Identify the associated intent of the interview discourse text through the knowledge graph engine; determine the third matching score of each initial intent based on the relevance of the associated intent to each initial intent.

[0048] In this step, we build a knowledge graph related to the interview, including company information, job requirements, industry knowledge, etc. When the interviewee mentions "I am familiar with big data technology and used Hadoop to process data in previous projects", we can extract terms (such as "big data technology" and "Hadoop") from the answer and map them with nodes in the knowledge graph (such as the company's requirements for big data skills and Hadoop-related knowledge modules). We find the knowledge graph path that matches the interviewee's answer, get the associated intent, and then Figure 1 On the one hand, it can verify the accuracy of the initial intent identified by the machine learning engine, and on the other hand, it can make the initial intent more clear.

[0049] Step S105: determining a comprehensive score based on the first matching score, the second matching score and the third matching score of each initial intention, and determining a winning intention based on the comprehensive score.

[0050] In this step, by balancing the scores of the three dimensions, it is possible to avoid over-fitting of the intent recognition results due to excessively high scores in one dimension or too low scores in one dimension. For example, when an interviewee answers a question about technical ability, the machine learning engine may give a higher matching score based on technical keyword matching, the basic meaning engine may give a lower score based on sentence structure and vocabulary accuracy, and the knowledge graph may give a score rate between the two based on the association between technology and job requirements. The final intent matching degree is determined by the comprehensive score, which can make the intent matching more accurate, provide accurate judgment for AI interviewers, and better assist in interview evaluation.

[0051] The present invention identifies the user's intention through three dimensions: machine learning engine, basic meaning engine and knowledge graph engine. These three engines are interrelated and progressive, and jointly help to accurately understand the user's answers and judge the intention. Through this progressive collaboration, the three engines work closely together in the AI ​​interviewer scenario, process user answers from different levels and angles, and gradually dig deeper into the user's intentions, significantly improving the accuracy and reliability of intention recognition, and laying a solid foundation for the efficient and accurate evaluation of AI interviewers.

[0052] Based on the above embodiment, when the number of intentions identified based on the similarity matching results is multiple, the method further includes: Step S106: Filter the initial intent based on the preset intent exclusion rule to obtain the filtered initial intent.

[0053] The preset intention exclusion rule includes at least a first exclusion rule and a second exclusion rule: The first exclusion rule is to prioritize excluding intents based on entity value matching; entity values ​​include at least dates and numbers.

[0054] Since information such as dates and numbers have relatively clear intentions and no ambiguity, they can be directly excluded without further intent recognition, saving computing resources. Only those unclear intentions can be retained for further recognition.

[0055] The second exclusion rule is to determine the intent match type based on the first match score of the initial intent, where the intent match type includes possible match and confirmed match.

[0056] In this step, the first matching score is compared with a preset threshold range, such as 60-90 points. If it is between 60-90, it is a possible match, and if it is above 90, it is a confirmed match. Of course, the threshold range can also be mapped to between 0 and 1 according to specific computing requirements.

[0057] Possible matches are those with high scores but not meeting the confidence requirements for exact matches. Definite matches are those with high confidence and are judged to be exact matches with the user's utterance.

[0058] When multiple initial intents include both types, exclude the intent matching type as the initial intent that may be matched.

[0059] If there are both possible matches and definite matches, the possible matches can be directly excluded and the definite matches can be retained. On the one hand, this can reduce the processing load of subsequent engines, save computing resources, and improve computing efficiency. On the other hand, it can help quickly screen out intentions that are closer to the user's real thoughts.

[0060] Of course, other exclusion rules can be set according to needs. For example, if the user speech contains two intentions (assuming that intentions i and j are two intentions in the user speech), and a definite match has been found before (assuming that intention i is the definite matching intention that has been found), then the definite matching intention j that appears later is excluded.

[0061] Based on the above embodiment, calculating the second matching score of each initial intent based on the lexical meaning of the keyword includes the following steps: Step S103A: Determine the original word role score of the keyword, where the original word role represents the role or function of the word in its original form.

[0062] In an example, the original word role of a keyword is a noun, a verb, an adjective, and the like.

[0063] In this step, firstly, keywords associated with each intent are extracted. For example, if the interview candidate's answer includes multiple intents such as "product description", "core value analysis", and "target user positioning", then keywords associated with the intent of "product description" include, for example, "beautiful, practical", etc. The basic meaning engine is used to analyze the original word roles of these keywords, the roles in the sentence, and other meanings to determine their relevance to the intent of "product description". The greater the relevance, the higher the second matching score, and the smaller the relevance, the smaller the second matching score.

[0064] Step S103B: Determine the role score of the keyword in the sentence. The role in the sentence represents the role of the word in the entire sentence structure.

[0065] In this step, the roles in the sentence, such as subject, object, and predicate, can reflect their position and function in the entire sentence structure.

[0066] Step S103C: Determine the word status score of the keyword after preprocessing; the preprocessing includes synonym replacement, word form restoration and standardization.

[0067] Step S103D: Calculate the second matching score for each initial intent based on the keyword's original word role score, its role score in the sentence, the preprocessed word status score, the weight value corresponding to each score, and the preset bonus items and penalty items.

[0068] Among them, the bonus items are scores calculated based on sentence structure, word position, sequence rewards, role rewards, extension rewards, etc., and the penalty items are scores calculated based on the number of conjunctions in the user's speech, etc.

[0069] In a feasible implementation, the range of the original word role score, the role score in the sentence, and the pre-processed word status score can be normalized and limited to [0, 1].

[0070] In one example, the intent Score The calculation formula is: (2); in, Score the original word role of the keyword; Score the role of the keyword in the sentence; The word status score of the keyword after preprocessing; Score points for bonus items; Score points for penalty items; is the weight of the original word role; score for the role in the sentence; is the word state weight after preprocessing; The code name for the basic meaning engine.

[0071] Based on the above embodiment, identifying the associated intention of the interview discourse text through the knowledge graph engine; determining the third matching score of each initial intention based on the relevance of the associated intention to each initial intention includes: Step S104A: extracting a term set from the interview discourse text.

[0072] In this step, a pre-trained entity recognition model (such as a BERT model) can be used to extract a set of terms from the interview discourse text. For example, terms such as "TikTok", "short video creation", "algorithm recommendation", "young people", and "middle-aged and elderly users" can be extracted from user answers.

[0073] Step S104B: Based on the term set, search for a path set that can associate these terms in the pre-constructed knowledge graph of the target industry field.

[0074] In this step, the knowledge graph engine searches for associated paths in its own network based on these term sets, and then The associated paths of as a path collection.

[0075] In a specific example, suppose the user input utterance is ,from The term set extracted from ; Terms in the knowledge graph The associated path set is ; Based on preset thresholds , filter out terms with more than A collection of paths : (3); in, A collection of paths One of the associated paths in ; is the preset term quantity threshold; Is one of the terms in the term set.

[0076] Step S104C: Determine the association intention based on the direction of the path set.

[0077] For example, there is a path in the knowledge graph that connects "Tik Tok" with key information such as "the core value of the short video creation platform lies in satisfying users' desire for expression and creativity" and "attracting users of all ages, mainly young people". The associated intention of the interviewee's answer can be inferred from this path.

[0078] Step S104D: Determine a third matching score for each initial intent based on the similarity between the associated intent and each initial intent.

[0079] In a feasible implementation, cosine similarity may be used to calculate the similarity between the associated intent and the initial intent, and the third matching score may be determined based on the similarity score.

[0080] Based on the above embodiment, determining a comprehensive score based on the first matching score, the second matching score, and the third matching score of each initial intention, and determining a winning intention based on the comprehensive score includes: A first confidence level for each initial intent is determined based on a machine learning engine.

[0081] In this step, the confidence can be determined by calculating the standard deviation or variance of the first matching scores of each initial intent. Taking the standard deviation as an example, the lower the standard deviation, the higher the consistency of the machine learning engine in identifying the specific intent, and the higher the confidence. In order to convert it into a more intuitive confidence, one of the following conversion methods can be used: Confidence = − σ ; If you want to get a confidence value between 0 and 1, you can also use a normalization function to limit it to between 0 and 1.

[0082] The engine determines a second confidence level for each initial intent based on the base meaning.

[0083] In this step, the second confidence level may also be determined based on the standard deviation or variance of the second matching score, which will not be described in detail herein.

[0084] The third confidence level of each initial intent is determined based on the knowledge graph engine.

[0085] In this step, the second confidence level may also be determined based on the standard deviation or variance of the third matching score, which will not be described in detail herein.

[0086] A comprehensive score is calculated based on the first matching score, the second matching score, the third matching score, the first confidence level, the second confidence level, the third confidence level, and a preset weight of each engine.

[0087] In a specific example, the composite score The calculation is as follows: (4); in, is the weight of each engine, and ; For Intention The first match score of For Intention The second match score of For Intention 's third match score; For Intention The first confidence level of For Intention The second confidence level; For Intention The third confidence level.

[0088] The comprehensive score is compared with the preset score threshold. If it exceeds the preset score threshold, it is determined to be a winning intention.

[0089] In this step, the number of winning intentions may be one or more.

[0090] Based on the above embodiment, after determining the winning intention, the method further includes: Step S107: determining the matching type of the winning intention based on the comprehensive score of each winning intention, where the matching type includes possible matching and confirmed matching.

[0091] In this step, the comprehensive score is compared with a preset threshold range, such as 60-90, or the range is limited to between 0-1, and the preset threshold range is 0.6-0.9. If the comprehensive score is between 0.6-0.9, it is a possible match, and if it exceeds 0.9, it is a definite match.

[0092] Step S108: When the number of winning intentions is 1, if the matching type of the winning intention is a confirmed match, the feedback information of the winning intention is directly output; if the matching type of the winning intention is a possible match, the winning intention is output in the form of a preset dialog box for user confirmation.

[0093] In one example, the preset dialog box is, for example, "Do you mean XX?", so that the interview candidate can confirm, so as to determine the user's intention and give corresponding feedback according to the determined intention.

[0094] Step S109: When the number of winning intentions is 2 or more, if the matching type of the winning intention is multiple confirmed matches, all the winning intentions are output for user selection; if the matching type of the winning intention is multiple possible matches, the winning intention is output in the form of a preset dialog box for user confirmation; if the matching type of the winning intention includes possible matches and confirmed matches, feedback information of the winning intention corresponding to the confirmed matches is output.

[0095] By further confirming the uncertain winning intention, it is possible to more accurately match the interviewee's intention and give more precise answers or feedback based on the interviewee's intention.

[0096] Based on the same inventive concept, an AI interviewer-based intention recognition system is provided, such as Figure 2 As shown, the system includes: The acquisition unit 201 is used to acquire the interview speech text of the interview candidate.

[0097] The machine learning engine 202 is used to perform text similarity matching between the interview speech text and a set of example speeches with preset intent labels, and to identify at least one initial intent based on the similarity matching results and determine a first matching score for each initial intent; the preset intent labels are predefined based on the company's recruitment needs.

[0098] The basic meaning engine 203 is used to disassemble the interview discourse text to obtain keywords associated with each initial intent, and calculate the second matching score of each initial intent based on the lexical meaning of the keywords.

[0099] The knowledge graph engine 204 is used to identify the associated intent of the interview discourse text; and determine the third matching score of each initial intent based on the relevance of the associated intent to each initial intent.

[0100] The determination unit 205 is configured to determine a comprehensive score based on the first matching score, the second matching score, and the third matching score of each initial intention, and determine a winning intention based on the comprehensive score.

[0101] Based on the same technical concept, an embodiment of the present invention further provides an electronic device, such as Figure 3 As shown, it includes a processor 301 , a communication interface 302 , a memory 303 and a communication bus 304 , wherein the processor 301 , the communication interface 302 , and the memory 303 communicate with each other via the communication bus 304 .

[0102] Memory 303, used for storing computer programs; The processor 301 is used to implement the steps of the AI ​​interviewer-based intention recognition method when executing the program stored in the memory 303.

[0103] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0104] The communication interface is used for communication between the above electronic device and other devices.

[0105] The memory may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0106] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0107] The computer program product for performing the AI ​​interviewer-based intention recognition method provided in the embodiment of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the method described in the previous method embodiment. The specific implementation can be found in the method embodiment, which will not be repeated here.

[0108] The device for AI interviewer intention recognition provided in the embodiment of the present invention can be specific hardware on the device or software or firmware installed on the device. The device provided in the embodiment of the present invention has the same implementation principle and technical effects as the aforementioned method embodiment. For the sake of brief description, the parts not mentioned in the device embodiment can refer to the corresponding contents in the aforementioned method embodiment. Technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can all refer to the corresponding processes in the aforementioned method embodiment, and will not be repeated here.

[0109] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0110] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0111] In addition, each functional unit in the embodiment provided by the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0112] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0113] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0114] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. They should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. An intention recognition method based on AI interviewer, characterized in that: Applied to an intention recognition engine, the intention recognition engine includes a machine learning engine, a basic meaning engine and a knowledge graph engine, the method includes: Obtain the interview speech text of the interview candidate; The interview speech text is matched with a set of example speech with preset intent labels by a machine learning engine for text similarity, and at least one initial intent is identified based on the similarity matching result and a first matching score of each initial intent is determined; the preset intent labels are predefined based on the recruitment needs of the enterprise; Decomposing the interview speech text by a basic meaning engine to obtain keywords associated with each of the initial intentions, and calculating a second matching score for each initial intention based on the lexical meaning of the keywords; Identify the associated intention of the interview speech text through the knowledge graph engine; determine the third matching score of each initial intention based on the relevance of the associated intention to each initial intention; A comprehensive score is determined based on the first matching score, the second matching score, and the third matching score of each initial intent, and a winning intent is determined based on the comprehensive score.

2. The method according to claim 1, characterized in that When the number of intentions identified based on the similarity matching results is multiple, the method further includes: The initial intent is filtered based on the preset intent exclusion rule to obtain the filtered initial intent.

3. The method according to claim 2, characterized in that The preset intention exclusion rule includes at least a first exclusion rule and a second exclusion rule: The first exclusion rule is to prioritize the exclusion of intents based on entity value matching; the entity value includes at least a date and a number; The second exclusion rule is to determine the intent match type based on the first match score of the initial intent, and the intent match type includes possible match and confirmed match; When multiple initial intents include both types, exclude the intent matching type as the initial intent that may be matched.

4. The method according to claim 1, characterized in that: Calculating a second match score for each initial intent based on the lexical meaning of the keyword includes: Determine the original role score of the keyword, the original role represents the role or function of the word in its original form; Determine the role score of the keyword in the sentence. The role in the sentence represents the role of the word in the entire sentence structure. Determine the word status score of the keyword after preprocessing; the preprocessing is synonym replacement, word form restoration and standardization; The second matching score of each initial intent is calculated based on the original word role score of the keyword, the role score in the sentence, the preprocessed word status score, the weight value corresponding to each score, and the preset bonus items and penalty items.

5. The method according to claim 1, characterized in that The related intention of the interview speech text is identified by the knowledge graph engine; Determining a third matching score for each initial intent based on the relevance of the associated intent to each initial intent includes: Extracting a term set from the interview discourse text; Based on the term set, searching for a path set that can associate these terms in a pre-built knowledge graph of the target industry field; Determining the association intention based on the orientation of the path set; A third matching score for each initial intent is determined based on a similarity between the associated intent and each initial intent.

6. The method according to claim 1, characterized in that The step of determining a comprehensive score based on the first matching score, the second matching score, and the third matching score of each initial intent, and determining a winning intent based on the comprehensive score includes: Determining a first confidence level for each initial intent based on a machine learning engine; determining a second confidence level for each initial intent based on the base meaning engine; Determine the third confidence level of each initial intent based on the knowledge graph engine; Calculating a comprehensive score based on the first matching score, the second matching score, the third matching score, the first confidence level, the second confidence level, the third confidence level, and a preset weight of each engine; The comprehensive score is compared with a preset score threshold, and if it exceeds the preset score threshold, it is determined as a winning intention.

7. The method according to claim 6, characterized in that After determining the winning intention, the method further includes: Determine a match type of the winning intention based on the comprehensive score of each winning intention, where the match type includes a possible match and a confirmed match; When the number of winning intentions is 1, if the matching type of the winning intention is a confirmed match, the feedback information of the winning intention is directly output; if the matching type of the winning intention is a possible match, the winning intention is output in the form of a preset dialog box for confirmation by the user; When the number of winning intentions is 2 or more, if the matching type of the winning intention is multiple confirmed matches, all the winning intentions are output for user selection; if the matching type of the winning intention is multiple possible matches, the winning intention is output in the form of a preset dialog box for user confirmation; if the matching type of the winning intention includes possible matches and confirmed matches, feedback information of the winning intention corresponding to the confirmed matches is output.

8. An intention recognition system based on AI interviewer, characterized in that: The system comprises: An acquisition unit, used for acquiring the interview speech text of the interview candidate; A machine learning engine is used to perform text similarity matching between the interview utterance text and a set of example utterances with preset intent labels, and to identify at least one initial intent based on the similarity matching result and determine a first matching score for each of the initial intents; the preset intent labels are predefined based on the recruitment needs of the enterprise; A basic meaning engine, used for disassembling the interview speech text to obtain keywords associated with each of the initial intentions, and calculating a second matching score for each initial intention based on the lexical meaning of the keywords; A knowledge graph engine, configured to identify the associated intent of the interview utterance text; and determine a third matching score of each initial intent based on the association between the associated intent and each initial intent; The determination unit is used to determine a comprehensive score based on the first matching score, the second matching score and the third matching score of each initial intention, and determine a winning intention based on the comprehensive score.

9. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, for implementing the method steps described in any one of claims 1 to 7 when executing a program stored in a memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps described in any one of claims 1 to 7 are implemented.

Images