Intelligent question answering method and device based on job application scene, equipment and medium

Abstract

The invention discloses an intelligent question and answer method, device and equipment based on a job application scene and a medium. Comprising the following steps: taking a large language model as an Agent, and identifying a question and answer scene corresponding to each inquiry message under the condition that the inquiry message input by a user is acquired; under the condition that the question and answer scene is a job application scene, multiple job application intention types associated with the job application scene are obtained, the multiple job application intention types comprise job application post search, job application recruitment behavior QA, enterprise consultation QA and resume optimization, and the multiple job application intention types correspond to different job application question and answer tools; performing intention recognition on the inquiry message, and matching a corresponding target job-hunting intention type for the inquiry message from a plurality of job-hunting intention types; and when the target job-hunting intention type is a job-hunting recruitment behavior QA, calling the intelligent BI to convert the inquiry message into an sql statement, and querying a user job-hunting behavior table and an HR recruitment behavior table based on the sql statement to obtain a response result.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, in particular to the field of artificial intelligence technology, and in particular to an intelligent question-answering method, device, equipment and medium based on a job search scenario. Background Art

[0002] With the advent of big models, AI's understanding of natural language is getting better and better. Problems that could only be solved by manual chat in the past can now be solved automatically by big models combined with local knowledge bases and other professional tools. For example, in the e-commerce field, different platforms have launched chat assistants that can help users recommend products that better meet their needs.

[0003] However, currently in the job search and recruitment field, there is no similar chat assistant to answer questions and resolve doubts for job seekers, and it is unable to meet the needs of job seekers well. Summary of the invention

[0004] In view of this, the embodiments of the present application provide an intelligent question-answering method, device, equipment, medium and product based on a job search scenario, which can answer questions and resolve doubts for job seekers in a job search and recruitment scenario, and better meet the needs of job seekers.

[0005] In a first aspect, an embodiment of the present application provides an intelligent question-and-answer method based on a job search scenario, the method comprising: taking a large language model as an agent, and in the case of obtaining an inquiry message input by a user, for each inquiry message, identifying the question-and-answer scenario corresponding to the inquiry message based on the large language model; in the case where the question-and-answer scenario is a job search scenario, obtaining multiple job search intention types associated with the job search scenario, wherein the multiple job search intention types include job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization, and the multiple job search intention types correspond to different job search question-and-answer tools, and the job search question-and-answer tools are used to return a response result for the inquiry message of the corresponding job search intention type; performing intent recognition on the inquiry message, and matching the corresponding target job search intention type for the inquiry message from the multiple job search intention types; in the case where the target job search intention type is job search and recruitment behavior QA, calling the intelligent BI to convert the inquiry message into a SQL statement, and querying the user job search behavior table and the HR recruitment behavior table based on the SQL statement to obtain a response result, and the user job search behavior table and the HR recruitment behavior table are constructed based on the obtained job search user behavior data and recruitment user behavior data.

[0006] In a second aspect, an embodiment of the present application provides an intelligent question-and-answer device based on a job search scenario, the device comprising: a scenario recognition module, which uses a large language model as an agent, and identifies the question-and-answer scenario corresponding to the inquiry message for each inquiry message when an inquiry message input by a user is obtained; an acquisition module, which is used to obtain multiple job search intention types associated with the job search scenario when the question-and-answer scenario is a job search scenario, wherein the multiple job search intention types include job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization, and the multiple job search intention types correspond to different job search question-and-answer tools, and the job search question-and-answer tools are used to return a response result for the inquiry message of the corresponding job search intention type; an intent recognition module, which is used to perform intent recognition on the inquiry message, and match the corresponding target job search intention type for the inquiry message from the multiple job search intention types; a query module, which is used to call the intelligent BI to convert the inquiry message into a SQL statement when the target job search intention type is a job search and recruitment behavior QA, and query the user job search behavior table and the HR recruitment behavior table based on the SQL statement to obtain a response result, wherein the user job search behavior table and the HR recruitment behavior table are constructed based on the obtained job search user behavior data and recruitment user behavior data.

[0007] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the steps of the intelligent question-and-answer method based on a job search scenario as in the first aspect are implemented.

[0008] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the steps of the intelligent question-and-answer method based on a job search scenario as in the first aspect are implemented.

[0009] In a fifth aspect, an embodiment of the present application provides a computer program product, which is stored in a non-volatile storage medium, and when the computer program product is executed by a processor, implements the steps of the intelligent question-and-answer method based on a job search scenario as in the first aspect.

[0010] In a sixth aspect, an embodiment of the present application provides a chip, which includes a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the steps of the intelligent question-and-answer method based on a job search scenario as in the first aspect.

[0011] The present application provides an intelligent question-and-answer method, device, equipment, medium and product based on a job search scenario. In order to accurately respond to user job search questions, the question-and-answer scenarios in the job search scenario are further subdivided into different job search intention types. Different job search intention types correspond to different job search question-and-answer scenarios. Multiple job search intention types may include job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization. Using a large language model as an agent, for each inquiry message sent by a user, the question-and-answer scenario corresponding to the inquiry message is effectively identified based on the large language model. When it is identified as a job search scenario, the inquiry message is intentionally identified, and the job search intention type of the inquiry message in the job search scenario is further identified. Based on this, the present application calls different job search question-and-answer tools to answer the inquiry message according to the different job search intention types identified. After identifying the target job search intention type, the job search question-and-answer tool is matched based on the target job search intention type, and the job search question-and-answer tool corresponding to the target job search intention type is called to query the inquiry message, thereby realizing intelligent scheduling of the job search question-and-answer tool. In this way, each job search inquiry message of the user can be assigned to a matching job search question and answer tool, and then based on the matching job search question and answer tool, the answer result can be accurately queried for the job search inquiry message. For example, when the target job search intention type is job search and recruitment behavior QA, the intelligent BI, user job search behavior table and HR recruitment behavior table are used as job search question and answer tools, and the intelligent BI is called to convert the inquiry message into a SQL statement, and the user job search behavior table and HR recruitment behavior table are queried based on the SQL statement to obtain the answer result. In this way, in the job search question and answer scenario, targeted and effective responses to different job search intention questions of job seekers can be achieved, and the user's multiple needs in the job search field, such as job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization, can be solved in one stop, so as to better meet the needs of job seekers. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following briefly introduces the drawings in the embodiments of the present application.

[0013] Figure 1 It is a flowchart of an intelligent question-answering method based on a job search scenario provided by an embodiment of the present application;

[0014] Figure 2 It is a structural diagram of an intelligent question-and-answer device based on a job search scenario provided in an embodiment of the present application;

[0015] Figure 3 It is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] The principles and spirit of the present application will be described below with reference to several exemplary embodiments. It should be understood that the purpose of providing these embodiments is to make the principles and spirit of the present application clearer and more thorough, so that those skilled in the art can better understand and implement the principles and spirit of the present application. The exemplary embodiments provided herein are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments herein, all other embodiments obtained by ordinary technicians of the art without creative work are within the scope of protection of this application.

[0017] In this document, terms such as first, second, third, etc. are only used to distinguish one entity (or operation) from another entity (or operation), but not to require or imply any order or relationship between these entities (or operations).

[0018] The following is a detailed description of the intelligent question-answering method based on a job search scenario provided by the embodiment of the present application through specific embodiments and their application scenarios in conjunction with the accompanying drawings.

[0019] Figure 1 It is a flow chart of an intelligent question-and-answer method based on a job search scenario provided by an embodiment of the present application. The executor of the intelligent question-and-answer method based on a job search scenario may be a job search chat assistant.

[0020] Exemplarily, the job-seeking chat assistant can be applied to the C-end server of an online recruitment platform, and the C-end can be a job-seeking system for fresh graduates, an online job-seeking system, a job recommendation system, etc. used by job seekers.

[0021] The following uses the execution subject of the intelligent question-and-answer method based on the job search scenario as an example, which is a job search chat assistant, to illustrate the intelligent question-and-answer method based on the job search scenario of the present application. It should be noted that the above execution subject and application scenario do not constitute a limitation on the present application.

[0022] like Figure 1 As shown, the intelligent question-answering method based on a job search scenario provided in an embodiment of the present application may include steps 110 to 140.

[0023] Step 110, using the large language model as an agent, when obtaining a query message input by a user, for each query message, identifying a question-answering scenario corresponding to the query message;

[0024] Step 120, when the question-and-answer scenario is a job-seeking scenario, multiple job-seeking intention types associated with the job-seeking scenario are obtained, the multiple job-seeking intention types including job search, job-seeking and recruitment behavior QA, enterprise consulting QA, and resume optimization, and the multiple job-seeking intention types correspond to different job-seeking question-and-answer tools;

[0025] Step 130, performing intent recognition on the inquiry message, and matching the inquiry message with a corresponding target job-seeking intent type from multiple job-seeking intent types;

[0026] Step 140, when the target job-seeking intention type is job-seeking and recruitment behavior QA, call the intelligent BI to convert the query message into a SQL statement, and query the user job-seeking behavior table and the HR recruitment behavior table based on the SQL statement to obtain the response result.

[0027] The intelligent question-and-answer method based on the job search scenario provided in the embodiment of the present application, in order to accurately respond to the user's job search questions, further subdivides the question-and-answer scenarios in the job search scenario into different job search intention types. Different job search intention types correspond to different job search question-and-answer scenarios. Multiple job search intention types may include job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization. Using the large language model as the Agent, for each inquiry message sent by the user, the question-and-answer scenario corresponding to the inquiry message is effectively identified based on the large language model. When it is identified as a job search scenario, the inquiry message is intentionally identified, and the job search intention type of the inquiry message in the job search scenario is further identified. Based on this, the present application calls different job search question-and-answer tools to answer the inquiry message according to the different job search intention types identified. After identifying the target job search intention type, the job search question-and-answer tool is matched based on the target job search intention type, and the job search question-and-answer tool corresponding to the target job search intention type is called to query the inquiry message, thereby realizing the intelligent scheduling of the job search question-and-answer tool. In this way, each user's job search inquiry message can be assigned to a matching job search question and answer tool, and then based on the matching job search question and answer tool, the job search inquiry message can be accurately queried for the answer result. For example, when the target job search intention type is job search and recruitment behavior QA, the intelligent BI, user job search behavior table and HR recruitment behavior table are used as job search question and answer tools, and the intelligent BI is called to convert the inquiry message into a SQL statement, and the user job search behavior table and HR recruitment behavior table are queried based on the SQL statement to obtain the answer result. In this way, the job search user's questions can be answered and doubts can be resolved in the job search and recruitment scenario, and the user's job search and recruitment behavior QA, enterprise consulting QA, and resume optimization needs in the job search field can be solved in one stop, so as to better meet the needs of job seekers.

[0028] The specific implementation of the above steps will be described in detail below in conjunction with specific embodiments.

[0029] In step 110, the large language model is used as an agent. When a query message input by a user is obtained, for each query message, a question-answering scenario corresponding to the query message is identified.

[0030] In step 110, for each inquiry message, the message keywords in the inquiry message can be extracted through the large language model, and the message keywords are analyzed to determine the question-answering scenario corresponding to the inquiry message. Different question-answering scenarios correspond to different message keyword sets. For example, the question-answering scenario can be a job search scenario, a shopping scenario, etc.

[0031] The large language model can be specified in advance by the developer, for example, set to a large language model (LLMs), a general meaning thousand questions model, etc., which is not specifically limited in this application. Among them, LLMs is a deep learning model with powerful natural language processing capabilities. Based on LLMs, natural language text can be understood and generated for various tasks, such as text generation, question-answering systems, language translation, etc.

[0032] The job search chat assistant may have a corresponding job search chat interface, where users can successively enter multiple inquiry messages. For each inquiry message, the large language model first performs question-answering scene recognition, and identifies the question-answering scene that matches the inquiry message by performing semantic analysis and other processing on the inquiry message. The question-answering scene may be, for example, a job search scene, a shopping scene, etc.

[0033] The core responsibility of the Agent in this application is to understand the user's job search question and answer needs and decompose these job search question and answer needs into a series of executable subtasks. In order to achieve this goal, the large language model in this application needs to have the following key capabilities:

[0034] Natural Language Processing: The task planning agent needs to have strong NLP capabilities to understand the job search question and answer requirements expressed by users verbally or in writing, including speech recognition, text analysis, sentiment recognition, etc.

[0035] Task decomposition and scheduling: After understanding the user's job search and Q&A needs, the task planning agent will decompose the job search and Q&A needs into a series of specific subtasks. These subtasks may require different skills and resources, so the task planning agent also needs to have intelligent scheduling capabilities to assign subtasks to the most appropriate subtask tools.

[0036] Real-time feedback and adjustment: During the task execution, the task planning agent needs to constantly monitor the task progress and make adjustments based on the feedback. For example, if a subtask encounters difficulties, the task planning agent needs to reallocate resources or adjust the task plan in a timely manner.

[0037] Involving step 120, when the question-answering scenario is a job-seeking scenario, a plurality of job-seeking intention types associated with the job-seeking scenario are obtained.

[0038] In step 120, the question-and-answer scenarios in the job search scenario are further subdivided into different job search intention types, and different job search intention types correspond to different job search question-and-answer scenarios. Multiple job search intention types may include job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization. Multiple job search intention types correspond to different job search question-and-answer tools, and the job search question-and-answer tools are used to return answer results for inquiry messages corresponding to the job search intention types, that is, different job search question-and-answer tools provide question-and-answer services for inquiry messages of different job search intention types.

[0039] Job search is used to represent the user's intention to find a job based on the screening criteria, such as "I want to find an algorithm engineer position in Shanghai"; job recruitment behavior QA is used to represent the user's intention to understand the job search behavior of other job seekers and / or the recruitment behavior of recruiters, such as "I want to know what positions job seekers of XX major in XX school are looking at", "I want to know what new positions XX company has released", "I want to know whether XX position in XX company is a hot recruitment"; enterprise consulting QA is used to represent the user's intention to understand enterprise-related information, such as enterprise size, enterprise introduction, enterprise news and other information; resume optimization is used to represent the user's intention to optimize his or her personal resume, such as "I want to modify my resume".

[0040] In step 130, the inquiry message is subjected to intent recognition, and a corresponding target job-seeking intent type is matched for the inquiry message from among a plurality of job-seeking intent types.

[0041] In step 130, the message keywords in the inquiry message are extracted, and semantic analysis is performed based on the message keywords to find the job-seeking intention type with the highest semantic similarity to the inquiry message, that is, the highest matching degree, as the target job-seeking intention type. In this way, after step 130, the inquiry message can be queried by calling the job-seeking question-and-answer tool corresponding to the target job-seeking intention type to obtain the answer result returned by the job-seeking question-and-answer tool.

[0042] For example, if the query message input by the user is "I want to find an algorithm engineer position in Shanghai", the corresponding target job-seeking intention type is job search; if the query message input by the user is "I want to know what positions job seekers majoring in XX school are looking at", the corresponding target job-seeking intention type is job recruitment behavior QA; if the query message input by the user is "Is there any news about XX company yesterday?", the corresponding target job-seeking intention type is enterprise consulting QA.

[0043] In some embodiments of the present application, in order to accurately identify the specific job-seeking intention in the user's job-seeking inquiry message, determine that the large language model is a general meaning thousand-question model, and the above step 130 performs intent recognition on the inquiry message, and matches the inquiry message with a corresponding target job-seeking intention type from multiple job-seeking intention types, which may specifically include the following steps:

[0044] Based on the Tongyi Qianwen model, the inquiry message is used to identify the intent and match the inquiry message with the corresponding target job-seeking intention type.

[0045] In this way, the Tongyi Qianwen model is generated based on a large amount of training data and is suitable for a wide range of knowledge question-and-answer scenarios. Through the logical reasoning ability of the Tongyi Qianwen model, it can perform effective semantic analysis and mining of inquiry messages, and accurately summarize the user's job-seeking intentions expressed by the inquiry messages.

[0046] In some embodiments of the present application, in order to accurately and effectively recommend available positions to users, after matching the query message with a corresponding target job search intention type from multiple job search intention types in step 130, the method may further include the following steps:

[0047] When the target job search intention type is job search, a definition field set configured for the job search question-and-answer tool corresponding to the target job search intention type is obtained, and integrity verification is performed on the inquiry message based on the definition field set to obtain a verification result;

[0048] When the check result indicates that a field is missing, guidance information is generated based on the missing field in the query message and then returned to the user until the received query message passes the integrity check, wherein the guidance information is used to guide the user to complete the missing field.

[0049] The defined field set is pre-configured and contains necessary fields for users to filter available positions.

[0050] In an embodiment of the present application, when the user's job search intention is to search for a job position, based on the set of defined fields corresponding to the job search question and answer tool, it is queried whether useful field information is missing in the user's inquiry message. If it is missing, a sentence asking for relevant information is generated using the big model to guide the user to complete the missing information, so as to more accurately find and filter job information suitable for the user to submit. For example, the search engine requires that the job search location must be entered. If the user's inquiry message is "I want to find an algorithm engineer position", the big model of the present application will determine that the location is missing and will further ask the user where he wants to work. Only after the user completes the missing information will the search engine be called to find relevant positions for the user, ensuring that it can accurately and effectively recommend positions that can be submitted to the user.

[0051] In one example, the guidance information may be used to prompt the user to supplement information such as intended positions, intended industries, skills, salary ranges, job search areas, and which subway is convenient for travel.

[0052] In some embodiments of the present application, the set of defined fields configured for the job search question and answer tool corresponding to the job search may include query, job search location, education level, company size, company nature, job type, years of experience, and upper and lower salary limits.

[0053] In this way, by limiting the necessary job search fields in the defined field set, it is ensured that comprehensive and effective job screening conditions can be extracted from the user's inquiry message in the future. Based on the job screening conditions, the jobs suitable for the user's application can be accurately screened, avoiding the return of low-matching jobs to the user due to broad screening conditions in the inquiry message, thereby improving the user's satisfaction with job search questions and answers.

[0054] In the related art, during the chat between the job search assistant and the user, the chat time may be too long, causing the large model to ignore the inquiry message sent by the user earlier, resulting in problems such as task forgetting and inaccurate intent recognition.

[0055] Based on this, in some embodiments of the present application, after the integrity check of the query message based on the defined field set is performed, the following steps may be specifically included:

[0056] If the user's current inquiry message passes the integrity check, extracting a job-seeking character string from the user's current inquiry message, and using the job-seeking character string to update the user's corresponding job-seeking profile, wherein the user's job-seeking profile is generated based on the user's historical inquiry messages;

[0057] By calling Elasticsearch to query the current inquiry message and user job profile, and performing ES semantic sorting, the response result is obtained.

[0058] Specifically, when receiving a user's inquiry message, the application can construct a user job-seeking portrait based on the job-seeking string in the inquiry message, and continuously update the user job-seeking portrait according to the latest received inquiry message, and the user job-seeking portrait can be used to characterize the user's current job-seeking preference or job-seeking intention. The job-seeking string can be the field value of the job-seeking field extracted from the inquiry message, for example, algorithm engineer, company nature: state-owned enterprise, company size: more than 1,000 people, etc.

[0059] In the embodiment of the present application, in order to solve the problem of Agent forgetting, the user's job-seeking preference portrait is extracted and updated in real time based on the user's chat history, and the portrait is continuously generated and updated in multiple rounds of conversations, which can avoid the large model from forgetting, so as to recommend jobs to users more accurately. Elasticsearch is a distributed search and analysis engine at the core of the Elastic Stack. For the job search intention type of job search, by calling Elasticsearch query messages and performing ES semantic sorting, it can quickly and accurately filter out recommended jobs for users.

[0060] In some embodiments of the present application, when the target job search intention type is job position search, a job search string is extracted from the user's inquiry message, and the job search string is converted into a code acceptable to Elasticsearch, and a response result is obtained by calling the Elasticsearch query code and performing ES semantic sorting.

[0061] For example, the job search string may be converted according to the following conversion rules:

[0062] Query: generally refers to job title, company name, five social insurances and one housing fund, two days off a week, and other welfare word descriptions. For example, when searching for a certain type of company, the user may not directly enter the company name, such as "big factory". At this time, it is necessary to convert "big factory" into at least one specific big factory company name.

[0063] Location: If it is a province or city, remove the words "province and city", such as Shanghai instead of Shanghai. If it is a city or county, include the words "district and county", such as Changning District instead of Changning. In addition, if you input Shanghai Changning, the output is Changning District. Note that it supports provinces, cities, districts and counties, but does not support locations such as landmarks.

[0064] Education: supports the following six character strings: junior high school and below, technical / secondary school, associate degree, bachelor's degree, master's degree, high school, and doctorate.

[0065] Company size: supports the following seven character strings: less than 50 people, 50-150 people, 150-500 people, 500-1000 people, 1000-5000 people, 5000-10000 people, and more than 10000 people. Here, you need to convert the text description into the company size, for example, a large company can be converted into 5000-10000 people or more than 10000 people.

[0066] Company nature: supports the following seven character strings: European and American foreign companies, non-European and American foreign companies, state-owned enterprises, private companies, public institutions, listed companies, and startups.

[0067] Working experience requirement: supports the following six strings: 1 year, 2 years, 3-4 years, 5-7 years, 8-9 years, 10 years and above, no experience required. If the user requires 1-3 years of working experience, it can be mapped to 1 year or 2 years.

[0068] Job Type: Supports the following three strings: Full-time, Part-time, Internship.

[0069] Salary Cap: Supports numbers and extracts the salary cap value.

[0070] Salary lower limit: supports numbers and extracts the salary lower limit value.

[0071] In an embodiment of the present application, when the user's job search intention corresponding to the query message is to search for a job position, a job search string can be extracted from the query message, and the job search string can be converted into a code acceptable to Elasticsearch, so that the accuracy of the job search can be guaranteed. For example, if the user's query message is "I want to find an algorithm engineer in a state-owned enterprise, and the company size must be more than 10,000 people", the large language model will extract the following job search string query from the query message: algorithm engineer, company nature: state-owned enterprise, company size: more than 10,000 people, and then convert these job search strings into codes acceptable to the search engine, so that the search engine can directly query the code for job search, and finally return the recommended positions.

[0072] In the related art, when the user's job search intention is relatively special, by calling the search engine, it may not be possible to find suitable recommended positions, resulting in inaccurate response results. For example, if a user wants to find an algorithm engineer position in a company that can see the Huangpu River, by calling the search engine to directly search for positions, no suitable recommended positions can be found.

[0073] In some embodiments of the present application, in order to solve the problem of inaccurate response results, after obtaining the response results by calling Elasticsearch to query the current query message and the user's job-seeking profile and performing ES semantic sorting, the method may further include the following steps:

[0074] The big language model performs a React operation on the response result of Elasticsearch, so that the big language model determines whether the response result of Elasticsearch meets the preset response condition;

[0075] When the response result of Elasticsearch does not meet the preset response conditions, the job-seeking intention type of the inquiry message is re-determined as real-time search, and the inquiry message is searched again in the preset web page library, and the response result is returned to the user after it is obtained.

[0076] Specifically, the preset response condition may include a preset matching degree threshold value corresponding to the current query message and the response result. The large language model evaluates the matching degree between the current query message and the response result to Elasticsearch. If the matching degree is greater than the preset matching degree threshold value, it can be considered that the response result of Elasticsearch meets the preset response condition; otherwise, it is considered that the preset response condition is not met.

[0077] In the embodiment of the present application, the LLM can be made to react. After the large language model obtains the search results of Elasticsearch, it will reflect on them. When the search results are inaccurate, it will reposition the intention of the inquiry message as a real-time search and search in a pre-set web page, so that job seekers with special and niche job-seeking intentions can also get accurate recommended positions as response results.

[0078] In some examples, when a user searches for a product manager at a large company, the large language model may translate it as "product manager of company 1". When the search engine is called but no suitable position is found, the large language model can reformulatequery to generate "product manager of company 2" and then search again.

[0079] Involving step 140, when the target job-seeking intention type is job-seeking and recruitment behavior QA, the intelligent BI is called to convert the query message into a SQL statement, and based on the SQL statement, the user job-seeking behavior table and the HR recruitment behavior table are queried to obtain the response result.

[0080] In step 140, a user job-seeking behavior table and an HR recruitment behavior table are constructed based on the obtained job-seeking user behavior data and recruitment user behavior data.

[0081] In some embodiments of the present application, before the above-mentioned step 140, the method may further include: parsing the acquired job-seeking user behavior data, extracting field data of multiple first fields to construct a user job-seeking behavior table; parsing the acquired recruitment user behavior data, extracting field data of multiple second fields to construct an HR recruitment behavior table.

[0082] Specifically, the multiple first fields may include but are not limited to user name, school name, school type, click behavior object, delivery behavior object, chat behavior object, job title, company name, company location; the multiple second fields may include but are not limited to: resume download object, HR positive feedback object, job title, company name, company location, user delivery resume, user ID. By summarizing the job search user behavior data of different job search user groups, a user job search behavior table that can characterize the job search preferences of different user groups is obtained; by summarizing the recruitment user behavior data of different recruitment user groups, a user recruitment behavior table that can characterize the recruitment preferences of different user groups is obtained.

[0083] In an embodiment of the present application, the recruitment platform can use the JD, CV data and user behavior data in the platform to generate a user job search behavior table and an HR recruitment behavior table, generate sql for the user job search question in the inquiry message through intelligent BI, and then use the generated sql to query in the wide table, and answer the user job search question according to the query result. For example, userid, school type, school name, click behavior, delivery behavior, chat behavior, JD position name, JD company name, JD company location and other fields can be organized into a fresh graduate job search table. HR review resume, HR positive feedback, position name, company name, company location, user delivery resume, user id can be organized into an HR behavior table. If the user inquiry message is "Which companies do students from Shanghai 985 generally apply to?", you can query the fresh graduate job search table according to the user question, generate sql to query the table, and return it to the big model after getting the answer. If the user inquiry message is "Is a certain company, a certain position, hot recruitment?", the big model can generate sql to query the HR behavior table, and return it to the big model after getting the answer.

[0084] In this way, the user behavior data of job seekers and recruiters on the recruitment platform can be integrated into user job search behavior tables and HR recruitment behavior tables. When faced with job search questions that cannot be answered by the knowledge question and answer database and the large model, the user intention is converted into SQL through intelligent BI, and the user job search behavior table and HR recruitment behavior table are queried based on SQL statements to obtain the answer results, thereby effectively responding to user inquiry messages and providing users with effective employment guidance, solving the problem of being unable to find adapted answers from the question and answer database and being unable to effectively respond to inquiry messages. At the same time, the user behavior data in the recruitment platform is reasonably utilized, improving the degree of data utilization.

[0085] In some embodiments of the present application, the response results of some inquiry messages can be queried by calling the real-time search interface bing.

[0086] In this application, different job search intention types correspond to different job search question and answer tools. In the process of chatting with users, in order to provide targeted and effective responses to inquiry messages of each job search intention type, agents are used to perform intent recognition, path planning, and intelligent calls to job search question and answer tools to provide users with targeted response services.

[0087] In the related art, in the process of replying to user inquiry messages, since the job-seeking questions in the user inquiry messages are diverse, many of them exceed the knowledge reserve of the large model, and it is impossible to find suitable answers in the question and answer database, so it is impossible to effectively reply to the inquiry messages.

[0088] In some embodiments of the present application, after the above step 130, the following steps may also be included:

[0089] When the target job-seeking intent type is resume optimization, the user resume profile is constructed by parsing the inquiry message, and similar high-quality resume data is matched to the user resume profile by calling the high-quality resume database, and resume modification suggestions are generated as the response result based on the high-quality resume data;

[0090] When the target job-seeking intention type is enterprise consulting QA, the inquiry message is queried from the RAG knowledge base to obtain the response result.

[0091] Specifically, the user resume portrait is used to characterize the user's personal career experience and job-seeking preferences. The high-quality resume database may include high-quality resume data with high scores selected based on HR scores, and / or high-quality resume data modified and adjusted by career planning experts. Based on the user resume portrait, high-quality resume data with high similarity to the user's personal career experience and job-seeking preferences is selected. The resume modification suggestions can be the resume adjustment information generated by the large language model by comparing the user resume portrait with the high-quality resume data, or directly using the desensitized high-quality resume data as the resume modification suggestions.

[0092] In this way, for the job search intention type of resume optimization, by calling the high-quality resume database, resume modification suggestions are generated for users, and resume modification guidance is effectively provided. For the job search intention type of corporate consulting QA, by calling the RAG knowledge base to query inquiry messages, corporate inquiries can be responded to quickly and effectively. This application calls the corresponding job search question and answer tool for effective query messages of different job search intention types, and can return accurate response results for inquiry messages of various job search intentions, thereby improving the comprehensiveness of the job search assistant's answers.

[0093] To better understand this application, the following is an example of a job search dialogue between a job seeker User and a job search assistant Assis provided by this application:

[0094] User: Hello, I am looking for an algorithm engineer position.

[0095] Assis: Hello! Of course I can help you. Where is your desired job location? I can provide you with more accurate recruitment information.

[0096] User: Shanghai.

[0097] Assis: I found some relevant job information. For example, there is a job for "Senior Java Engineer XXX" and the job location is in Shanghai. The position requires the person to be responsible for developing and maintaining the company's core business systems and have a solid Java technology foundation and project experience. Specific responsibilities include working closely with colleagues and customers, understanding requirements and converting them into system functions, and using the tools and libraries provided by the system for code analysis and optimization. In addition, it is also required to ensure the security and reliability of the system, participate in system architecture design, and follow up on technological trends to keep technology synchronized. If you are interested, you can view the detailed job description and requirements. This is the job link: [Senior Java Engineer XXX Job Link]. Please note that although this position is a Java engineer, the relevant skills and experience of an algorithm engineer are also highly matched in many technical requirements. If you are specifically looking for a position marked as "Algorithm Engineer", please let me know and I can help you find it further.

[0098] User: Thank you!

[0099] Assis: You’re welcome! Feel free to let me know if you have any other questions or need more help.

[0100] I wish you good luck in your job search!

[0101] It should be noted that the above is only a simple example of a job search dialogue scenario for job search, which is a job search intention type. In addition, it can also be applied to job search dialogue scenarios with other user job search intentions, and the present application does not make any specific limitations on this.

[0102] Corresponding to the method embodiment of the present application, the present application also provides an intelligent question-and-answer device based on a job search scenario.

[0103] Figure 2 Schematic diagram of the structure of an intelligent question-answering device based on a job search scenario provided by an embodiment of the present application. Figure 2 As shown, the intelligent question-answering device 200 based on the job search scenario may include: a scenario recognition module 210 , an acquisition module 220 , an intention recognition module 230 and a query module 240 .

[0104] Among them, the scene recognition module 210 is used to use the large language model as the Agent, and when obtaining the inquiry message input by the user, for each inquiry message, identify the question and answer scene corresponding to the inquiry message; the acquisition module 220 is used to obtain multiple job search intention types associated with the job search scene when the question and answer scene is a job search scene, wherein the multiple job search intention types include job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization, and the multiple job search intention types correspond to different job search question and answer tools; the intention recognition module 230 is used to identify the intention of the inquiry message, and match the corresponding target job search intention type for the inquiry message from the multiple job search intention types; the query module 240 is used to call the intelligent BI to convert the inquiry message into a SQL statement when the target job search intention type is job search and recruitment behavior QA, and query the user job search behavior table and the HR recruitment behavior table based on the SQL statement to obtain the response result, and the user job search behavior table and the HR recruitment behavior table are constructed based on the obtained job search user behavior data and recruitment user behavior data.

[0105] The intelligent question-and-answer device based on the job search scenario provided in the embodiment of the present application, in order to accurately respond to the user's job search questions, further subdivides the question-and-answer scenarios in the job search scenario into different job search intention types. Different job search intention types correspond to different job search question-and-answer scenarios. Multiple job search intention types may include job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization. Using the large language model as the Agent, for each inquiry message sent by the user, the question-and-answer scenario corresponding to the inquiry message is effectively identified based on the large language model. When it is identified as a job search scenario, the inquiry message is intentionally identified, and the job search intention type of the inquiry message in the job search scenario is further identified. Based on this, the present application calls different job search question-and-answer tools to answer the inquiry message according to the different job search intention types identified. After identifying the target job search intention type, the job search question-and-answer tool is matched based on the target job search intention type, and the job search question-and-answer tool corresponding to the target job search intention type is called to query the inquiry message, thereby realizing the intelligent scheduling of the job search question-and-answer tool. In this way, each job search inquiry message of the user can be assigned to a matching job search question and answer tool, and then based on the matching job search question and answer tool, the answer result can be accurately queried for the job search inquiry message. For example, when the target job search intention type is job search and recruitment behavior QA, the intelligent BI, user job search behavior table and HR recruitment behavior table are used as job search question and answer tools, and the intelligent BI is called to convert the inquiry message into a SQL statement, and the user job search behavior table and HR recruitment behavior table are queried based on the SQL statement to obtain the answer result. In this way, in the job search question and answer scenario, targeted and effective responses to different job search intention questions of job seekers can be achieved, and the user's multiple needs in the job search field, such as job search, job search and recruitment behavior QA, enterprise consulting QA, and resume optimization, can be solved in one stop, so as to better meet the needs of job seekers.

[0106] In some embodiments of the present application, it also includes: a verification module, which is used to match the corresponding target job search intention type for the inquiry message from multiple job search intention types, and when the target job search intention type is a job search, obtain the set of defined fields configured for the job search question and answer tool corresponding to the target job search intention type, and perform integrity verification on the inquiry message based on the set of defined fields to obtain a verification result; a generation module, which is used to generate guidance information based on the missing fields in the inquiry message and return it to the user when the verification result indicates that the field is missing, until the received inquiry message passes the integrity verification, wherein the guidance information is used to guide the user to complete the missing fields.

[0107] In some embodiments of the present application, the set of defined fields configured for the job search question and answer tool corresponding to the job search includes query, job search location, education level, company size, company nature, job type, years of experience, and upper and lower salary limits.

[0108] In some embodiments of the present application, the query module is also used to: extract a job search string from the current inquiry message when the user's current inquiry message passes the integrity check, and use the job search string to update the user's corresponding user job search portrait, wherein the user job search portrait is generated based on the user's historical inquiry messages; obtain a response result by calling Elasticsearch to query the current inquiry message and the user job search portrait, and performing ES semantic sorting.

[0109] In some embodiments of the present application, it also includes: a determination module, which is used to perform a React operation on the response result of Elasticsearch through a large language model, so that the large language model determines whether the response result of Elasticsearch satisfies the preset response condition; the intention recognition module is also used to redefine the job search intention type corresponding to the inquiry message as real-time search when the response result of Elasticsearch does not meet the preset response condition, and re-search the inquiry message in the preset web page library, and return the response result to the user after obtaining it.

[0110] In some embodiments of the present application, it also includes: a construction module, which is used to parse the obtained job-seeking user behavior data before calling the intelligent BI to convert the query message into a SQL statement, and extract the field data of multiple first fields to construct a user job-seeking behavior table, wherein the multiple first fields include user name, school name, school type, click behavior object, delivery behavior object, chat behavior object, position name, company name, and company location; the construction module is also used to parse the obtained recruitment user behavior data, and extract the field data of multiple second fields to construct an HR recruitment behavior table, wherein the multiple second fields include resume download objects, HR positive feedback objects, position names, company names, company locations, user submitted resumes, and user IDs.

[0111] In some embodiments of the present application, the query module is specifically used for: when the target job-seeking intention type is resume optimization, constructing a user resume portrait by parsing the inquiry message, and matching similar high-quality resume data for the user resume portrait by calling a high-quality resume database, and generating resume modification suggestions based on the high-quality resume data as a response result; when the target job-seeking intention type is corporate consulting QA, querying the inquiry message from the RAG knowledge base to obtain a response result.

[0112] The intelligent question-answering device based on job-seeking scenarios provided in the embodiment of the present application can achieve Figure 1 The various processes implemented by the service platform in the method embodiment can achieve the same technical effect. To avoid repetition, they will not be described here.

[0113] Figure 3 It is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application.

[0114] like Figure 3 As shown, the electronic device 300 includes a memory 301 , a processor 302 , and a computer program stored in the memory 301 and executable on the processor 302 .

[0115] In an example, the processor 302 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0116] The memory 301 may include a read-only memory (ROM), a random access memory (RAM), a disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer executable instructions, and when the software is executed (e.g., by one or more processors), it can be operated to perform the operations described with reference to the card opening method in the embodiment according to the first aspect of the present application.

[0117] The processor 302 runs a computer program corresponding to the executable program code by reading the executable program code stored in the memory 301, so as to implement the card activation method in the embodiment of the first aspect.

[0118] In some examples, the electronic device 300 may further include a communication interface 303 and a bus 310. Figure 3 As shown, the memory 301, the processor 302, and the communication interface 303 are connected via a bus 310 and communicate with each other.

[0119] The communication interface 303 is mainly used to realize the communication between the modules, devices, units and / or equipment in the embodiment of the present application. The communication interface 303 can also be used to access input devices and / or output devices.

[0120] The bus 310 includes hardware, software, or both, coupling the components of the electronic device 300 to each other. By way of example and not limitation, the bus 310 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a Memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-E) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses or a combination of two or more of these. Where appropriate, the bus 310 may include one or more buses. Although embodiments of the present application describe and illustrate a particular bus, the present application contemplates any suitable bus or interconnect.

[0121] The electronic device provided in the embodiment of the present application can realize Figure 1 The various processes implemented by the electronic device in the method embodiment can achieve the same technical effect, and to avoid repetition, they will not be described here.

[0122] In combination with the intelligent question-answering method based on job search scenarios in the above embodiments, the present application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, the steps of any one of the intelligent question-answering methods based on job search scenarios in the above embodiments are implemented.

[0123] In combination with the intelligent question-answering method based on job search scenarios in the above embodiments, the present application embodiment can provide a computer program product for implementation. The (computer) program product is stored in a non-volatile storage medium, and when the program product is executed by at least one processor, it implements the steps of any one of the intelligent question-answering methods based on job search scenarios in the above embodiments.

[0124] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned embodiment of the intelligent question-and-answer method based on the job search scenario, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0125] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0126] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0127] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), appropriate firmware, plug-in, function card, etc. When implemented in software, the elements of the present application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0128] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0129] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0130] The above is only a specific implementation of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. An intelligent question-answering method based on a job search scenario, characterized in that: include: Taking the large language model as the agent, when obtaining the inquiry message input by the user, for each inquiry message, identifying the question-answering scenario corresponding to the inquiry message based on the large language model; In the case where the question-and-answer scenario is a job-seeking scenario, a plurality of job-seeking intention types associated with the job-seeking scenario are obtained, wherein the plurality of job-seeking intention types include job search, job-seeking recruitment behavior QA, enterprise consulting QA, and resume optimization, and the plurality of job-seeking intention types correspond to different job-seeking question-and-answer tools, and the job-seeking question-and-answer tool is used to return a response result for an inquiry message corresponding to the job-seeking intention type; Performing intent recognition on the inquiry message, and matching a corresponding target job-seeking intention type for the inquiry message from among the multiple job-seeking intention types; When the target job-seeking intention type is job-seeking and recruitment behavior QA, the intelligent BI is called to convert the inquiry message into a SQL statement, and based on the SQL statement, the user job-seeking behavior table and the HR recruitment behavior table are queried to obtain the response result. The user job-seeking behavior table and the HR recruitment behavior table are constructed based on the acquired job-seeking user behavior data and recruitment user behavior data.

2. The method according to claim 1, characterized in that After matching the query message with a corresponding target job-seeking intention type from the multiple job-seeking intention types, the method further includes: In the case where the target job search intention type is the job search position search, obtaining a set of definition fields configured for the job search question-and-answer tool corresponding to the target job search intention type, and performing integrity verification on the inquiry message based on the set of definition fields to obtain a verification result; In the case where the verification result indicates that a field is missing, guidance information is generated based on the missing field in the inquiry message and then returned to the user until the received inquiry message passes the integrity check, wherein the guidance information is used to guide the user to complete the missing field.

3. The method according to claim 2, characterized in that The set of defined fields configured for the job search question and answer tool corresponding to the job search includes query, job search location, education background, enterprise size, enterprise nature, job type, years of experience, and upper and lower salary limits.

4. The method according to claim 2, characterized in that: Also includes: If the current inquiry message of the user passes the integrity check, extracting a job-seeking character string from the current inquiry message, and using the job-seeking character string to update a user job-seeking portrait corresponding to the user, wherein the user job-seeking portrait is generated based on the historical inquiry messages of the user; The response result is obtained by calling Elasticsearch to query the current inquiry message and the user job-seeking profile and performing ES semantic sorting.

5. The method according to claim 4, characterized in that After obtaining the response result, the method further includes: The big language model performs a React operation on the response result of Elasticsearch, so that the big language model determines whether the response result of Elasticsearch meets the preset response condition; When the response result of Elasticsearch does not meet the preset response conditions, the job search intention type corresponding to the inquiry message is re-determined as real-time search, and the inquiry message is searched again in the preset web page library, and the response result is returned to the user after it is obtained.

6. The method according to claim 1, characterized in that Before calling the intelligent BI to convert the query message into a SQL statement, it also includes: Parsing the obtained job-seeking user behavior data, extracting field data of multiple first fields to construct a user job-seeking behavior table, wherein the multiple first fields include user name, school name, school type, click behavior object, delivery behavior object, chat behavior object, position name, company name, and company location; The acquired recruitment user behavior data is parsed, and field data of multiple second fields are extracted to construct an HR recruitment behavior table, wherein the multiple second fields include resume download objects, HR positive feedback objects, position names, company names, company locations, user submitted resumes, and user IDs.

7. The method according to claim 1, characterized in that Also includes: In the case where the target job-seeking intention type is resume optimization, a user resume profile is constructed by parsing the inquiry message, and similar high-quality resume data is matched for the user resume profile by calling a high-quality resume database, and resume modification suggestions are generated based on the high-quality resume data as the response result; In the case where the target job-seeking intention type is enterprise consulting QA, the inquiry message is queried by calling the RAG knowledge base to obtain the response result.

8. An intelligent question-answering device based on a job search scenario, characterized in that: include: A scene recognition module is used to use the large language model as an agent, and when obtaining a query message input by a user, for each query message, identify the question-answering scene corresponding to the query message; an acquisition module, for acquiring, when the question-and-answer scenario is a job-seeking scenario, a plurality of job-seeking intention types associated with the job-seeking scenario, wherein the plurality of job-seeking intention types include job search, job-seeking recruitment behavior QA, enterprise consulting QA, and resume optimization, and the plurality of job-seeking intention types correspond to different job-seeking question-and-answer tools, and the job-seeking question-and-answer tool is used to return a response result for an inquiry message corresponding to the job-seeking intention type; an intention recognition module, configured to perform intention recognition on the inquiry message, and match the inquiry message with a corresponding target job-seeking intention type from among the multiple job-seeking intention types; The query module is used to call the intelligent BI to convert the inquiry message into a SQL statement when the target job-seeking intention type is job-seeking and recruitment behavior QA, and query the user job-seeking behavior table and the HR recruitment behavior table based on the SQL statement to obtain the response result. The user job-seeking behavior table and the HR recruitment behavior table are constructed based on the acquired job-seeking user behavior data and recruitment user behavior data.

9. An electronic device, characterized in that: The electronic device comprises: a processor and a memory storing computer program instructions; when the electronic device executes the computer program instructions, the method according to any one of claims 1 to 7 is implemented.

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

Images