Training sample generation method and device, storage medium and program product

Abstract

The embodiment of the invention provides a training sample generation method and device, a storage medium and a program product. In the embodiment of the invention, multiple pieces of job-seeking identity information distributed according to a set proportion among different categories under multiple service dimensions are acquired in advance, multiple pieces of resume information are generated by adopting a first agent, and interview interaction between an interviewer and a job seeker is simulated through a second agent and a third agent. Obtaining interview record information corresponding to each piece of resume information; and generating a training sample of the artificial intelligence interview model according to interview problem information included in the interview record information. The job seeking identity information is distributed according to the set proportion among different categories under multiple service dimensions, so that the balance of the job seeker identity information is ensured, in addition, the second agent and the third agent simulate interview interaction, generated question information and answer information are strongly associated, high-quality training data can be provided for model training, and the training efficiency is improved. And the model precision is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a training sample generation method, device, storage medium and program product. Background Art

[0002] Applying artificial intelligence (AI) technology to the interview field has become a research direction for many people. At present, employers use AI interview models to interview job seekers to screen suitable job seekers and improve interaction efficiency.

[0003] At present, the training data for AI interview models is mainly the existing conversation data from the interview process. The AI ​​interview model trained using the above training data has low accuracy and is difficult to meet the interview needs of employers. Summary of the invention

[0004] Multiple aspects of the present application provide a training sample generation method, device, storage medium and program product to improve the accuracy of the AI ​​interview model.

[0005] An embodiment of the present application provides a training sample generation method, including: obtaining multiple job-seeking identity information, wherein the multiple job-seeking identity information is distributed according to a set ratio among categories under multiple service dimensions; generating multiple first prompt words based on the multiple job-seeking identity information and a first prompt word template, wherein the first prompt words include first task description information for resume generation; calling a first intelligent agent respectively according to the multiple first prompt words to generate multiple resume information corresponding to the multiple job-seeking identity information according to the multiple first task description information included in the multiple first prompt words; using a second intelligent agent and a third intelligent agent to respectively simulate an interviewer and a job seeker to perform an interview interaction to obtain multiple interview record information corresponding to the multiple resume information, wherein the interview record information includes at least two interview question information; generating at least two training samples based on the at least two interview question information, and using the at least two training samples corresponding to each of the multiple interview record information as training samples for generating a target artificial intelligence interview model, wherein the target artificial intelligence interview model is used to generate interview question information for job seekers during the interview process.

[0006] An embodiment of the present application also provides an electronic device, comprising: a memory and a processor; the memory is used to store a computer program; the processor is coupled to the memory and is used to execute the computer program to implement each step in the training sample generation method provided in the embodiment of the present application.

[0007] The embodiment of the present application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor implements each step in the training sample generation method provided in the embodiment of the present application.

[0008] The embodiment of the present application also provides a computer program product, including a computer program / instruction. When the computer program / instruction is executed by a processor, the processor is caused to implement each step in the training sample generation method provided in the embodiment of the present application.

[0009] In an embodiment of the present application, multiple job-seeking identity information distributed in a set ratio among different categories under multiple service dimensions is obtained in advance, and multiple resume information is generated by a first agent. The second agent and the third agent respectively simulate the interviewer and the job seeker to interact with each other in the interview to obtain the interview record information corresponding to each resume information; the training sample of the artificial intelligence interview model is generated according to the interview question information included in the interview record information. Since the job-seeking identity information is distributed in a set ratio among different categories under multiple service dimensions, the balance of the job seeker's identity information is guaranteed. In addition, by simulating the interview interaction through the second agent and the third agent, the generated question information and answer information are strongly correlated, which can provide high-quality training data for model training and improve the accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0011] Figure 1 A flowchart of a training sample generation method provided for an exemplary embodiment of the present application;

[0012] Figure 2 A schematic diagram of a flow chart of interaction between a first agent, a second agent and a third agent provided for an exemplary embodiment of the present application;

[0013] Figure 3 A schematic diagram of a use process of a target artificial intelligence interview model provided for an exemplary embodiment of the present application;

[0014] Figure 4 A schematic diagram of the structure of a training sample generating device provided for an exemplary embodiment of the present application;

[0015] Figure 5 A schematic structural diagram of an electronic device provided for an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0017] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0018] The various models involved in this application (including but not limited to language models or large models) are in compliance with relevant laws and standards.

[0019] At present, in the field of AI interviews, interview question information is usually generated by large language models (LLM). However, there are a series of cost issues in calling the large language model method. For example, external large language models may need to be paid for use, which will incur higher calling costs. For another example, if an offline large language model is called, the fine-tuning cost will increase. For another example, large language models have high requirements for hardware resources and increase operating costs. In addition, the calling time of large language models is long, which reduces the interview experience.

[0020] In this embodiment, through the pre-trained neural network model, question information is selected from the question library and provided to the job seeker based on the output information of the job seeker. The pre-trained neural network model consumes less resources than the large language model, and does not need to be paid for use. Considering that selecting question information from the question library is essentially a text classification problem, text classification technology can be used to optimize request latency. However, it is very difficult to generate sample data required for model training without real sample data. Even if there is sample data in the real scene, there is a problem of category imbalance. For example, samples of different job types are unevenly distributed in quantity, there are a lot of samples of live-in nannies, and there are very few sample data of hourly workers. For another example, the sample data of the real scene has a single conversation content, such as only including name and age. The model trained based on sample data with imbalanced categories has the problem of low precision.

[0021] In response to the above technical problems, in an embodiment of the present application, multiple job-seeking identity information distributed in a set ratio among different categories under multiple service dimensions is obtained in advance, and a first agent is used to generate multiple resume information. The second agent and the third agent are used to simulate the interviewer and the job seeker to interact with each other in the interview, so as to obtain the interview record information corresponding to each resume information; the training sample of the artificial intelligence interview model is generated according to the interview question information included in the interview record information. Since the job-seeking identity information is distributed in a set ratio among different categories under multiple service dimensions, the balance of the job seeker's identity information is guaranteed. In addition, by simulating the interview interaction through the second agent and the third agent, the generated question information and answer information are strongly correlated, which can provide high-quality training data for model training and improve the accuracy of the model.

[0022] A solution provided by an embodiment of the present application is described in detail below in conjunction with the accompanying drawings.

[0023] Figure 1 The following is a flow chart of a training sample generation method provided by an exemplary embodiment of the present application. Figure 1 As shown, the method includes:

[0024] 101. Acquire multiple job-seeking identity information, and distribute the multiple job-seeking identity information among various categories under multiple service dimensions according to a set ratio;

[0025] 102. Generate multiple first prompt words according to multiple job-seeking identity information and a first prompt word template, where the first prompt words include first task description information for resume generation;

[0026] 103. Invoke the first agent respectively according to the plurality of first prompt words to generate a plurality of resume information corresponding to the plurality of job-seeking identity information according to the plurality of first task description information included in the plurality of first prompt words;

[0027] 104. Using the second agent and the third agent to respectively simulate the interviewer and the job seeker to perform interview interaction, so as to obtain a plurality of interview record information corresponding to the plurality of resume information, wherein the interview record information includes at least two interview question information;

[0028] 105. Generate at least two training samples based on at least two interview question information, and use at least two training samples corresponding to each of the multiple interview record information as training samples for generating a target artificial intelligence interview model. The target artificial intelligence interview model is used to generate interview question information for job seekers during the interview process.

[0029] In this embodiment, job-seeking identity information refers to the identity information of job seekers. Among them, the identity information of job seekers may include category information under multiple service dimensions. Among them, the service dimensions may include but are not limited to: age, service type, service content, service duration, etc. Depending on the service dimension, the category information is also different. For example, if the service dimension is age, the category information may include but is not limited to: 18-25, 26-35, 36-45, 46-55, 56-60 and other age groups. If the service dimension is service type, the category information may include but is not limited to: live-in nanny, day shift nanny, hourly worker, confinement nanny, childcare nanny, etc. If the service dimension is service content, the category information may include but is not limited to: cooking and housework, caring for the elderly, caring for children, etc.

[0030] Among them, multiple job-seeking identity information is distributed according to a set ratio among various job-seeking categories under multiple service dimensions. Distribution according to the set ratio can be uniform distribution. For example, if there are a total of 100 job-seeking identity information, in the case of uniform distribution, the number of samples for each service type should be 100 / 5=20 (taking 5 service types as an example), the number of samples for each service content should also be about 100 / 3=33 (taking 3 service contents as an example), and the number of samples for each age group should be 100 / 5=20 (taking 5 age groups as an example). The combination of 100 job-seeking identity information in the three dimensions of service type, service content and age is evenly distributed. Of course, in addition to uniform distribution, other set ratios can also be used. For example, when the service content includes 3 categories, the distribution ratio of 100 job-seeking identity information can be 1:2:3, etc.

[0031] In this embodiment, a resume generation agent is provided, which can generate resume information based on job-seeking identity information. For the convenience of distinction and description, the resume generation agent is referred to as the first agent. Among them, the first agent is an agent based on artificial intelligence (AI Agent). The first agent is an intelligent entity that can perceive the environment, make decisions and perform actions. Unlike traditional artificial intelligence, the first agent has the ability to gradually complete a given goal through independent thinking and calling tools, without the need to manually specify each step of the operation. In other words, the first agent is used as a representative of a person or an organization to perform a certain specific behavior, reduce the complexity of a person or organization's work, and reduce workload and communication costs. Among them, the large language model (LLM) can be used as the core computing engine of the first agent to provide the first agent with powerful text generation and comprehension capabilities. On this basis, the first agent can show a wider range of functions, such as dialogue, task completion, reasoning, and a certain degree of autonomous behavior. In the embodiment of this application, the focus is on the function of the first agent to generate resume information based on job-seeking identity information.

[0032] In this embodiment, the first agent corresponds to a first prompt word template. The first prompt word template is used to assist the first agent in generating resume information according to the job-seeking identity information. The first prompt word template includes first task description information for resume generation. The first task description information is used to describe the task that the first agent needs to perform. Based on this, multiple first prompt words can be generated according to multiple job-seeking identity information and the first prompt word template. For example, a job-seeking identity information is filled into the first prompt word template to obtain a first prompt word.

[0033] Optionally, the first prompt word template further includes: role information, target information, requirement information and constraint information. The job seeker identity information can be filled into the constraint information to obtain the first prompt word. Alternatively, the first prompt word template also reserves blank information, and the job seeker identity information is filled into the blank information to obtain the first prompt word.

[0034] For example, the first task description information includes target information and requirement information. The target information refers to the main purpose or expected effect of the first agent. For example, the target information can be implemented as generating a complete resume information. Among them, the role information can be implemented as a resume expert. Among them, the requirement information refers to the requirements for the first agent to generate resume information, and the requirement information includes but is not limited to: 1) According to the constraint information, a complete resume information is randomly generated; 2) The generated resume information needs to include the following fields: name, age, native place, years of employment, job type, job content, order address, relevant skill certificates, work experience, etc. The constraint information may include but is not limited to: 1) The name must meet the Chinese requirements and conform to the Chinese naming format; 2) The age can be [**age**] years old; 3) The native place can be a certain city; 4) The length of time engaged in the industry is between one month and 20 years; 5) The job type is [**workType**]; 6) The job content is [**workContent**]; 7) The order address is a certain city or a certain block.

[0035] Among them, [**age**], [**workType**] and [**workContent**] are placeholders corresponding to age, work type and work content respectively. When the job-seeking identity information is filled into the constraint information, the job-seeking identity information includes: age information, work type information and work content information. [**age**] can be replaced with the age information, [**workType**] can be replaced with the work type information, and [**workContent**] can be replaced with the work content information.

[0036] Optionally, the first prompt word template also includes: background information. In the case where the job-seeking identity information includes at least service type information and service content information, an implementation method of generating multiple first prompt words based on multiple job-seeking identity information and the first prompt word template includes: for any job-seeking identity information, obtaining service type background information and service content background information respectively according to the service type information and service content information; filling the job-seeking identity information, service type background information and service content background information into the first prompt word template to obtain the first prompt word corresponding to any job-seeking identity information. For example, the background knowledge of the first prompt word template includes: work type features [**workTypeFeature**] and work content features [**workContentFeature**], and [**workTypeFeature**] can be replaced with service type background information, and [**workContentFeature**] can be replaced with service content background information.

[0037] In this embodiment, the first agent is called respectively according to the multiple first prompt words to generate multiple resume information corresponding to the multiple job-seeking identity information according to the multiple first task description information included in the multiple first prompt words. Among them, since the multiple job-seeking identity information is distributed according to the set proportion among the categories under the multiple service dimensions, the first agent can generate multiple balanced resume information, lay the foundation for the subsequent interview interaction process, improve the quality of the answer information output by the third agent simulating the job seeker, and increase the number of hits in the question library.

[0038] In this embodiment, the second agent and the third agent respectively simulate the interviewer and the job seeker to perform interview interactions, so as to obtain multiple interview record information corresponding to multiple resume information. Specifically, the second agent simulates the interviewer in a real situation, and the third agent simulates the job seeker in a real situation. The second agent selects question information from the interview question library based on the job seeker's reply information and provides it to the third agent. The third agent replies to the interviewer's question information based on the resume information it maintains, and provides the reply information to the second agent. For example, for each resume information, the interview record information corresponding to each resume information is obtained through the process of simulating the interview interaction between the second agent and the third agent. For example, in Figure 2In the example, the first agent generates resume information, and the second agent and the third agent respectively simulate the interviewer and the job seeker to interact with each other in the interview. The process is demonstrated, but not limited to this. Among them, by having the second agent and the third agent respectively simulate the interviewer and the job seeker to interact with each other in the interview, the interview interaction in the real world can be simulated as much as possible, and a variety of interview scenarios can be created, such as different types of positions, industry backgrounds, and identity characteristics of job seekers, etc., to provide rich and balanced training data for subsequent model training, and significantly improve the AI ​​interview model's ability to understand complex interview situations.

[0039] In this embodiment, the interview record information includes at least two interview question information, which may be question information output by the interviewer simulated by the second agent. Further, the interview record information also includes job application answer information, which may be answer information output by the job seeker simulated by the third agent.

[0040] The second agent and the third agent may also be an agent based on artificial intelligence (AIAgent). For the introduction of the second agent and the third agent, please refer to the introduction of the first agent. It should be noted that in this embodiment, the focus is on the ability of the second agent to simulate the interviewer to select question information from the interview question library, and the focus is on the ability of the third agent to simulate the job seeker to output answer information based on the resume information to the interviewer's question information.

[0041] In this embodiment, each interview record information includes at least two interview question information, and at least two training samples are generated based on the at least two interview question information; the at least two training samples corresponding to each of the multiple interview record information are used as training samples for generating the target artificial intelligence interview model.

[0042] Among them, each interview record information includes at least two interview question information. In the process of generating at least two training samples, training samples can be generated based on all the interview question information in the interview record information, or some representative interview questions can be selected therefrom to ensure that a wider range of sample features are covered as much as possible. High-quality training samples can be effectively generated so that the target artificial intelligence interview model can learn diverse questioning patterns, which can not only improve the interview efficiency, but also ensure the consistency and fairness of the interview process.

[0043] The target AI interview model is used to generate interview question information for job seekers during the interview process. The target AI interview model is not limited in its implementation. For example, the target AI interview model can be implemented as a text classification model, which can include but is not limited to: Bidirectional Encoder Representations from Transformers, Support Vector Machine (SVM), and Text Convolutional Neural Network (Text CNN).

[0044] In an embodiment of the present application, multiple job-seeking identity information distributed in a set ratio among different categories under multiple service dimensions is obtained in advance, and multiple resume information is generated by a first agent. The second agent and the third agent respectively simulate the interviewer and the job seeker to interact with each other in the interview to obtain the interview record information corresponding to each resume information; the training sample of the artificial intelligence interview model is generated according to the interview question information included in the interview record information. Since the job-seeking identity information is distributed in a set ratio among different categories under multiple service dimensions, the balance of the job seeker's identity information is guaranteed. In addition, by simulating the interview interaction through the second agent and the third agent, the generated question information and answer information are strongly correlated, which can provide high-quality training data for model training and improve the accuracy of the model.

[0045] In an optional embodiment, the implementation method of obtaining multiple job-seeking identity information is not limited. In an optional embodiment, multiple job-seeking identity information is pre-maintained locally, and multiple job-seeking identity information can be directly obtained locally. In another optional embodiment, multiple job-seeking identity information can be generated. Among them, the following exemplarily provides an implementation method for generating multiple job-seeking identity information, obtaining multiple service dimensions corresponding to job seekers, the service dimensions include at least two job-seeking category information, and the number of category information included in different service dimensions can be the same or different. At least two category information included in the multiple service dimensions are combined to obtain at least two groups. Based on the at least two groups and their set ratios, multiple job-seeking identity information is generated, and one job-seeking identity information corresponds to one group. The above method can effectively generate multiple job-seeking identity information that meets the set ratio, thereby providing rich and accurate data support for subsequent model training.

[0046] Among them, the detailed introduction of service dimensions and category information can be found in the aforementioned embodiments, which will not be repeated here.

[0047] Each group includes one category information under each service dimension. For example, if there are two service dimensions, one service dimension includes five category information, and the other service dimension includes three category information, then 5×3=15 groups can be obtained, each group includes one category information under each service dimension, that is, includes category information under two different service dimensions.

[0048] Wherein, at least two groups correspond to a set ratio, and the set ratio refers to the ratio of the number of groups corresponding to the at least two groups. For example, at least two groups are implemented as 2, and the set ratio can be 1:1, 1:3 or 1:2, etc. For another example, at least two groups are implemented as 5, and the set ratio can be that the number of the 5 groups is equal, that is, 1:1:1:1:1, or the set ratio can be 1:1:2:1:2. Wherein, considering the category balance, the number ratio of at least two groups is the same. Wherein, the total number of job-seeking identity information to be generated can be obtained, and the number can be determined according to actual needs, for example, the total number can be 500, 1000 or 5000, etc., and this is not limited. According to the set ratio of at least two groups and the total number of job-seeking identity information to be generated, the number of job-seeking identity information corresponding to each group is calculated, and the corresponding number of job-seeking identity information is created for each group. For example, if the total number of job-seeking identity information to be generated is 100 and the groups are 20, then 5 job-seeking identity information are generated for each group, and the 5 job-seeking identity information all include the category information in the group. Subsequently, the first agent can generate resume information based on each job-seeking identity information. Different job-seeking identity information can generate different resume information. In addition to the information included in the job-seeking identity information, the resume information can also include other personalized information of the job seeker. For example, the job-seeking identity information includes: age, job content and job type, and the resume information can also include: name, place of origin, length of employment, service address range, skill certificate, work experience, etc.

[0049] Optionally, when the number of at least two groups is balanced, a random number generation method can be used to determine the group and generate multiple job-seeking identity information to ensure data balance. The number of at least two groups is balanced, which means that the number of each group is the same or approximately the same, and approximately the same means that the absolute value of the difference between the numbers of each group is less than or equal to a set number threshold, which can be 1, 2, or 5, etc., without limitation.

[0050] The following is an example of how to generate random numbers. For example, the service dimensions include job type and job content. Job content includes three categories: cooking and housework, caring for the elderly, and caring for children. Job types include five categories: live-in nanny, day shift nanny, hourly worker, confinement nanny, and childcare nanny. For each group, if the group includes three service dimensions: age, job content, and job type, random number generation is used to determine the group and generate multiple job-seeking identity information, including:

[0051] 1) Randomly generate a value from the preset age range as the age value of the job applicant.

[0052] 2) Generate a job type random number, and perform a modulo 3 operation on the job content random number (the job content includes 3 categories). If the result of modulo 3 on the job content random number is 0, it is determined that the job content of the job applicant is cooking and doing housework; if the result of modulo 3 is 1, it is determined that the job content of the job applicant is taking care of children; if the result of modulo 3 on the job content random number is 2, it is determined that the job content of the job applicant is taking care of the elderly.

[0053] 3) Generate a random number for job content, and perform a modulo 5 (job types include 5 categories) operation on the random number for job type. If the modulo 5 result is 0, the job type of the job applicant is determined to be a live-in nanny; if the modulo 5 result is 1, the job type of the job applicant is determined to be a day shift nanny; if the modulo 5 result is 2, the job type of the job applicant is determined to be an hourly worker; if the modulo 5 result is 3, the job type of the job applicant is determined to be a confinement nanny; if the modulo 5 result is 4, the job type of the job applicant is determined to be a childcare nanny.

[0054] That is, for each service dimension, a random number is generated, and according to the random number corresponding to each service dimension, a category information is selected from the category information of the service dimension, and multiple category information selected from multiple service dimensions are combined into a job-seeking identity information.

[0055] In an optional embodiment, the second agent and the third agent respectively simulate the interviewer and the job seeker to interact in an interview, so as to obtain multiple interview record information corresponding to multiple resume information. The implementation method is not limited. A specific implementation method is provided below. For any resume information, in any round of interview interaction, a second prompt word is generated according to the job-seeking answer information of the previous round of interview interaction and the second prompt word template, and the second prompt word template includes second task description information for generating interview question information; wherein the job-seeking answer information of the first round of interview interaction is preset information; a second agent is called according to the second prompt word to select interview question information of any round of interview interaction from an interview question library according to the second task description information; when the interview question information of any round of interview interaction does not indicate the end of the interview, a third prompt word is generated according to the interviewer question information of any round of interview interaction, any resume information and a third prompt word template, and the third prompt word template includes third task description information for generating job-seeking answer information; a third agent is called according to the third prompt word to generate job-seeking answer information of any round of interview interaction according to the third task description information and any resume information, and trigger the next round of interview interaction; when the interview question information of any round of interview interaction indicates the end of the interview, at least part of the interview interaction information of the round is used as the interview record information corresponding to any resume information. Among them, for each resume information, the second agent and the third agent can simulate multiple rounds of interaction, and a question information and the answer information to the question information can be regarded as one round of interaction.

[0056] Among them, the second agent corresponds to a second prompt word template, and the second prompt word template includes second task description information for generating interview question information, and the second task description information is used to describe the task that the second agent needs to perform. For example, the second task description information includes: target information and task information. Among them, the target information refers to the main purpose or expected effect of the first agent. For example, the target information can be to instruct the second agent to select a question information that best suits the current situation from the interview question library. Task information refers to a specific description of the operation or action that the second agent needs to perform. For example, an example of task information may be: "Please select appropriate question information to interview the job applicant based on the answer information, and you can output the question number and question content."

[0057] Optionally, the second prompt word template also includes role information, requirement information, context information, capability information and output format information. Role information refers to the role played by the second agent, for example, the role information may be an interviewer. Requirement information refers to the requirement for the second agent to select question information from the interview question library. For example, the requirement information may include but is not limited to: 1) a set number of question information (e.g., 1) can be output each time; 2) the next question information is selected from the unselected question information; 3) the question information is selected from the interview question library; 4) the question information is set with a sequence number, and the sequence numbering is performed. Context information refers to the scene in which the role played by the second agent is located. An example of context information can be implemented as: this is a scene of a job seeker interview. The job seeker can be interviewed through video, and the second agent is the interviewer. Each answer information of the job seeker can be obtained through voice-to-text conversion, and pay attention to identifying typos. Capability information refers to the ability of the second agent. An example of capability information can be implemented as: accurately understanding the answer of the job seeker obtained through voice-to-text conversion, identifying the wrong information included in the text, and correcting it. In addition, the second agent has the ability to evaluate the next optimal question of the current dialogue. The output format refers to the format in which the second agent outputs the question information. An example of an output format can be implemented as: question number + [question content]. For example, question number + [question content] can be implemented as: "1. Personal information [Let's do a simple interview now. You are looking for a job such as a nanny, confinement nanny, or childcare nanny, right? Please give a brief self-introduction, including name, age, address, etc.]".

[0058] For any resume information, in each round of interview interaction, a second prompt word is generated according to the job application answer information and the second prompt word template of the previous round of interview interaction. The job application answer information of the first round of interview interaction is preset information, for example, the preset information may include but is not limited to: "start interview" or "hello".

[0059] Among them, the implementation method of generating the second prompt word according to the job-seeking answer information of the previous round of interview interaction and the second prompt word template is not limited. For example, the second prompt word template includes: a blank prompt word, and the job-seeking answer information of the previous round of interview interaction can be filled into the blank prompt word to obtain the second prompt word. For another example, the second task description information includes placeholder information corresponding to the "answer information", and the job-seeking answer information of the previous round of interview interaction can be filled into the placeholder information to obtain the second prompt word. Among them, the second prompt word template also includes: placeholder information corresponding to the interview question library, and multiple question information in the interview question library can be added to the placeholder information. On this basis, the second prompt word can also include the interview question library. Based on this, the second agent is called according to the second prompt word to select the interview question information of any round of interview interaction from the interview question library according to the second task description information.

[0060] There are two main types of question information included in the interview question database: one is question information that does not indicate the end of the interview, and the other is question information that indicates the end of the interview. Question information that indicates the end of the interview refers to information that can trigger the end of the interview interaction. For example, question information that indicates the end of the interview may include but is not limited to: "Thank you very much for participating in this interview", "We will reply to you as soon as possible" or "If you have any questions, please feel free to contact us". Question information that does not indicate the end of the interview is a series of question information raised by the interviewer, which does not clearly convey the signal that the interview is about to end. Such questions usually do not directly lead job seekers to realize that the interview is coming to an end, but continue to discuss the position, company or job seeker's background.

[0061] Among them, when the interview question information of any round of interview interaction does not indicate the end of the interview, the third intelligent agent can generate job application answer information based on the interviewer question information of any round of interview interaction.

[0062] Specifically, the third intelligent agent corresponds to a third prompt word template, and the third prompt word template includes third task description information for generating job application answer information; the third prompt word is generated according to the interviewer's question information of any round of interview interaction, any resume information and the third prompt word template; the third intelligent agent is called according to the third prompt word to generate the job application answer information of any round of interview interaction according to the third task description information and any resume information, and trigger the next round of interview interaction.

[0063] For example, the third task description information may include but is not limited to: 1) Please answer the interviewer's questions based on the job applicant's resume information; 2) Output the answer information corresponding to the question information; 3) When introducing yourself, you can introduce your name, age, place of origin, etc.

[0064] For example, the interviewer question information of any round of interview interaction and any resume information can be filled into the third prompt word template to obtain the third prompt word. For example, in addition to the third task description information, the third prompt word template also includes: first placeholder information corresponding to the resume information and second placeholder information corresponding to the interviewer question information. The any resume information is filled into the first placeholder information, and the interviewer question information is filled into the first placeholder information to obtain the third prompt word. For another example, the interviewer question information of any round of interview interaction and any resume information can be filled into the blank prompt word of the third prompt word template to obtain the third prompt word.

[0065] Wherein, when the interview question information of any round of interview interaction indicates that the interview is over, at least part of the interview interaction information of the rounds is used as the interview record information corresponding to any resume information. For example, all the interview interaction information of the rounds can be used as the interview record information corresponding to any resume information, or, according to preset keywords, part of the interview interaction information of the rounds with preset keywords can be selected as the interview record information corresponding to any resume information. Wherein, the preset keywords can be information that the interviewer is concerned about or the job seeker is concerned about, such as name, age, job type, job content, skill certificate, years of experience, etc.

[0066] Optionally, the third agent corresponds to a target parameter, which is used to control the matching degree between the generated job application answer information and the interview question information. The target parameter can control the randomness and creativity of the job application answer information generated by the third agent. For example, the value range of the target parameter can be between the first parameter value and the second parameter value. For example, the value range of the target parameter can be realized as 0 to 1. If the target parameter is closer to the second parameter value (such as 1), it means that the randomness of the job application answer information output by the target parameter is higher, making the generated job application answer information more diverse and unpredictable, and the third agent is more likely to select low-probability job application answer information, which may make the generated job application answer information more creative, but may also reduce the accuracy or matching degree. If the target parameter is closer to the second parameter value (such as 0), the randomness of the job application answer information output by the third agent is reduced, making the generated job application answer information more certain and focused, so that the third agent tends to select high-probability job application answer information, which can improve the accuracy of the generated job application answer information.

[0067] Based on this, an implementation method of generating job-seeking answer information for any round of interview interaction based on the third task description information and any resume information includes: selecting a target value within the value range of the target parameter in a preset manner; generating job-seeking answer information for any round of interview interaction based on the third task description information, any resume information and the target parameter. For example, by generating a random number, a random number is generated as a target value within the value range of the target parameter, so as to meet the need for diversity of job-seeking answer information. For example, the target parameter can be the temperature parameter of the third agent.

[0068] Optionally, the implementation method of generating the second prompt word according to the job-seeking answer information of the previous round of interview interaction and the second prompt word template is not limited. The second prompt word template includes a task prompt word template, which is a prompt word template for guiding the second agent to perform a specific task. It is generally called each time a specific task is performed to provide continuous guidance. The task prompt word template includes the second task description information. For example, an example of the second task description information is: please select appropriate question information to interview the job seeker based on the historical interview interaction information, and the question sequence number and question content can be output. Based on this, the second prompt word can be generated according to the task prompt word template. Considering that in the case of generating the second prompt word, in addition to considering the job-seeking answer information of the previous round of interview interaction, the interview interaction information before the job-seeking answer information of the previous round of interview interaction can also be considered, and the interview interaction information can include job-seeking answer information and interview question information. In order to facilitate distinction and description, the interview interaction information before any round is called historical interview interaction information, and the historical interview interaction information includes the job-seeking answer information of the previous round of interview interaction, then the second prompt word can be generated according to the historical interview interaction information and the task prompt word template. For example, the historical interview interaction information is filled into the task prompt word template to generate the second prompt word. For example, the second prompt word template includes: placeholder information corresponding to the job application answer information of the previous round of interview interaction, and placeholder information corresponding to the historical answer information before the previous round of job application answer information, and the historical interview interaction information is filled into the above placeholder information to obtain the second prompt word.

[0069] Optionally, during the interaction between the second agent and the third agent, the interview interaction information can be stored in the interview interaction information database. For example, the interview identification information is set for each resume information, and the identification information may include but is not limited to: an identifier (ID), a nickname or a timestamp, etc. In a key-value pair (key-value) manner, the interview identification information is used as the key (key), and the historical interview interaction information is used as the value (value). Among them, after each interview question information is generated, the interview question information and the job-seeking answer information of the previous round can be added to the interview interaction information database. Alternatively, after each job-seeking question information is generated, the job-seeking question information and its corresponding interview question information can be added to the interview interaction information database. In the case of generating a task prompt word, the historical interview interaction information corresponding to the interview identification information can be obtained from the interview interaction information database according to the interview identification information, and the historical interview interaction information is added to the task prompt word template to obtain the task prompt word.

[0070] Further optionally, the second prompt word template also includes a configuration prompt word template, which is used to initialize the configuration of the second agent, for example, to define the dialogue environment, basic rules, background information, behavior, knowledge scope, and response style of the second agent. For example, the configuration prompt word template may include but is not limited to: role information, target information, requirement information, context information, capability information, and output format information.

[0071] Based on this, before generating the second prompt word according to the job-seeking answer information and the second prompt word template of the previous round of interview interaction, it is also possible to: generate the configuration prompt word according to the multiple interview question information in the interview question library and the configuration prompt word template; call the second agent according to the configuration prompt word to initialize the configuration of the second agent. For example, the configuration prompt word template also includes the placeholder information corresponding to the question library, and the multiple interview question information in the interview question library can be added to the placeholder information corresponding to the configuration of the question library to generate the configuration prompt word. For another example, the multiple interview question information in the interview question library can be added to the blank prompt word of the configuration prompt word template to generate the configuration prompt word.

[0072] Among them, in order to solve the situation where training samples are unavailable due to the update of the interview question library, in this embodiment, when the interview question library changes and a new interview question library is generated, the interview question library in the corresponding configuration prompt word template of the second intelligent agent can be replaced with the new interview question library. No other changes are required. The above-mentioned training sample generation method is re-executed to obtain sample data with balanced category distribution under the new interview question library.

[0073] In an optional embodiment, the method of generating at least two training samples according to at least two interview question information is not limited. The following is an example, for any interview question information, the interview interaction information located before any interview question information in the interview record information to which it belongs is used as a sample feature, and any interview question information is used as a sample label to obtain a training sample corresponding to any interview question information.

[0074] For example, the interview record information includes interview question information A1—job-seeking answer information B1, interview question information A2—job-seeking answer information B2, interview question information A3—job-seeking answer information B3, and interview question information A4—job-seeking answer information B4. Among them, the interview question information A1 and the job-seeking answer information B1 are used as sample features, and the interview question information A2 is used as a sample label to generate a sample data; the interview question information A1, the job-seeking answer information B1, the interview question information A2, and the job-seeking answer information B2 are used as sample features, and the interview question information A3 is used as a sample label to generate a sample data; the interview question information A1, the job-seeking answer information B1, the interview question information A2, the job-seeking answer information B2, the interview question information A3, and the job-seeking answer information B3 are used as sample features, and the interview question information A4 is used as a sample label. For another example, the interview record information includes interview question information A1—job-seeking answer information B1, interview question information A2—job-seeking answer information B2, interview question information A3—job-seeking answer information B3, and interview question information A4—job-seeking answer information B4. Among them, interview question information A1 and job application answer information B1 are used as sample features, and interview question information A2 is used as a sample label to generate a sample data; interview question information A2 and job application answer information B2 are used as sample features, and interview question information A3 is used as a sample label to generate a sample data; interview question information A3 and job application answer information B3 are used as sample features, and interview question information A4 is used as a sample label.

[0075] In an optional embodiment, at least two training samples corresponding to each of the multiple interview record information are used to perform model training on the initial artificial intelligence interview model to obtain the target artificial intelligence interview model. Among them, the method of model training is not limited. For example, from the dimension of the interview record information, at least two training samples corresponding to the interview record information are used to perform model training on the initial artificial intelligence interview model, calculate the first loss function of the interview record information dimension, and determine the second loss function of the model dimension based on the first loss function of the interview record information dimension. The model training is performed with the goal of minimizing the second loss function to obtain the target artificial intelligence interview model. For example, the first loss function may include but is not limited to: Mean Squared Error (MSE) function, Cross-Entropy Loss (Cross-Entropy Loss) function, Log Loss (Log Loss) and Mean Absolute Error (MAE), etc. Among them, the first loss function of the interview record information dimension is subjected to weighted averaging or logarithm operations to obtain the second loss function. For another example, the initial artificial intelligence interview model can be trained based on at least two training samples corresponding to each of the multiple interview record information, and the third loss function of the model dimension can be calculated. The model training is performed with the goal of minimizing the second loss function to obtain the target artificial intelligence interview model. Among them, the third loss function can include but is not limited to: Mean Squared Error (MSE) function, Cross-Entropy Loss function, Log Loss, and Mean Absolute Error (MAE), etc.

[0076] In an optional embodiment, in actual application, the target artificial intelligence interview model can simulate the interview interaction between the interviewer and the real job applicant. Figure 3An example of a target artificial intelligence interview model simulating the interaction between an interviewer and a job seeker is provided. At the beginning of the interview, the interviewer starts the interview with preset questions, such as interview question information C1, and the job seeker responds to the preset questions, for example, the job seeker responds to job search answer information D1. In the first round of interaction, the target artificial intelligence interview model can output the interview question information C1 and the job search answer information D1, perform text classification on the interview question information C1 and the job search answer information D1, and obtain category information. The category information corresponds to the question information in the interview question library. It can be considered that each question information corresponds to a category information. Therefore, based on the category information output by the target artificial intelligence interview model, the question information corresponding to the category information can be matched from the interview question library, such as interview question information C2. In the second round of interaction, the interview question information C1, job application answer information D1, interview question information C2 and job application answer information D2 are input into the target artificial intelligence interview model for text classification to obtain category information, and the interview question information D3 is matched from the interview question library based on the category information, and this process is repeated until the nth round (n is a positive integer, and n≥1), when the interview question information Cn indicates the end of the interview, and the interview ends. Further, the interview interaction information can be provided to the recruiter.

[0077] Regarding the embodiments of the present application Figure 1 The detailed implementation and beneficial effects of each step in the method have been described in detail in the aforementioned embodiments and will not be elaborated here.

[0078] It should be noted that the execution subject of each step of the method provided in the above embodiment can be the same device, or the method can be executed by different devices. For example, the execution subject of steps 101 to 103 can be device A; for another example, the execution subject of steps 101 and 102 can be device A, and the execution subject of step 103 can be device B; and so on.

[0079] In addition, in some of the processes described in the above embodiments and the accompanying drawings, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel, and the sequence numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the sequence numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0080] Figure 4A structural diagram of a training sample generating device provided by an exemplary embodiment of the present application is shown in FIG. Figure 4 As shown, the device includes: an acquisition module 41, a generation module 42, a calling module 43, a simulation module 44 and a processing module 45.

[0081] An acquisition module 41 is used to acquire multiple job-seeking identity information, where the multiple job-seeking identity information is distributed among various categories under multiple service dimensions according to a set ratio;

[0082] A generating module 42, configured to generate a plurality of first prompt words according to a plurality of job-seeking identity information and a first prompt word template, wherein the first prompt words include first task description information for resume generation;

[0083] A calling module 43, configured to call the first agent respectively according to the plurality of first prompt words, so as to generate a plurality of resume information corresponding to the plurality of job-seeking identity information according to the plurality of first task description information included in the plurality of first prompt words;

[0084] The simulation module 44 is used to simulate the interviewer and the job seeker to perform interview interaction through the second agent and the third agent respectively, so as to obtain a plurality of interview record information corresponding to the plurality of resume information, wherein the interview record information includes at least two interview question information;

[0085] A generating module 42, configured to generate at least two training samples according to at least two interview question information;

[0086] The processing module 45 is used to use at least two training samples corresponding to each of the multiple interview record information as training samples for generating a target artificial intelligence interview model, and the target artificial intelligence interview model is used to generate interview question information for job seekers during the interview process.

[0087] In an optional embodiment, the acquisition module is specifically used to: obtain multiple service dimensions corresponding to the job seeker, the service dimensions include at least two job search category information, and different service dimensions include the same or different numbers of category information; combine at least two category information included in the multiple service dimensions to obtain at least two groups, and the groups include one category information under each service dimension; generate multiple job search identity information based on the at least two groups and their set ratios, and one job search identity information corresponds to one group.

[0088] Optionally, the simulation module is specifically used to: for any resume information, in any round of interview interaction, generate a second prompt word according to the job-seeking answer information of the previous round of interview interaction and the second prompt word template, the second prompt word template including second task description information for generating interview question information; wherein, the job-seeking answer information of the first round of interview interaction is preset information; call a second agent according to the second prompt word to select interview question information of any round of interview interaction from the interview question library according to the second task description information; if the interview question information of any round of interview interaction does not indicate the end of the interview, generate a third prompt word according to the interviewer question information of any round of interview interaction, any resume information and a third prompt word template, the third prompt word template including third task description information for generating job-seeking answer information; call a third agent according to the third prompt word to generate job-seeking answer information of any round of interview interaction according to the third task description information and any resume information, and trigger the next round of interview interaction; if the interview question information of any round of interview interaction indicates the end of the interview, use at least part of the interview interaction information of the round as the interview record information corresponding to any resume information.

[0089] Optionally, the third intelligent agent corresponds to a target parameter, which is used to control the matching degree between the generated job application answer information and the interview question information; the simulation module is specifically used to: select a target value within the value range of the target parameter in a preset manner; generate job application answer information for any round of interview interaction based on the third task description information, any resume information and the target parameter.

[0090] Optionally, the second prompt word template includes a task prompt word template, and the task prompt word template includes second task description information; the simulation module is specifically used to: generate a second prompt word based on historical interview interaction information and the task prompt word template; wherein the historical interview interaction information is interview interaction information before any round, and the historical interview interaction information includes job search answer information of the previous round of interview interaction.

[0091] Further optionally, the second prompt word template also includes a configuration prompt word template; the generation module is also used to: generate configuration prompt words according to multiple interview question information in the interview question library and the configuration prompt word template; the calling module is also used to: call the second intelligent agent according to the configuration prompt words to initialize the configuration of the second intelligent agent.

[0092] In an optional embodiment, the job-seeking identity information includes at least service type information and service content information; the generation module is specifically used to: for any job-seeking identity information, obtain service type background information and service content background information respectively according to the service type information and service content information; fill the job-seeking identity information and service type background information and service content background information into the first prompt word template to obtain the first prompt word corresponding to any job-seeking identity information.

[0093] In an optional embodiment, the generation module is specifically used to: for any interview question information, use the interview interaction information located before any interview question information in the interview record information to which it belongs as a sample feature, and use any interview question information as a sample label to obtain a training sample corresponding to any interview question information.

[0094] Regarding the embodiments of the present application Figure 4 The detailed implementation and beneficial effects of each step in the illustrated device have been described in detail in the aforementioned embodiments and will not be elaborated on here.

[0095] Figure 5 A schematic diagram of the structure of an electronic device provided by an exemplary embodiment of the present application is shown in FIG. Figure 5 As shown, the device includes: a memory 54 and a processor 55.

[0096] The memory 54 is used to store computer programs and can be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method used to operate on the electronic device.

[0097] The processor 55 is coupled to the memory 54 and is used to execute the computer program in the memory 54, so as to: obtain multiple job-seeking identity information, and the multiple job-seeking identity information is distributed according to a set ratio among various categories under multiple service dimensions; generate multiple first prompt words according to the multiple job-seeking identity information and the first prompt word template, and the first prompt words include first task description information for resume generation; call the first intelligent agent respectively according to the multiple first prompt words to generate multiple resume information corresponding to the multiple job-seeking identity information according to the multiple first task description information included in the multiple first prompt words; simulate the interviewer and the job seeker to perform interview interaction through the second intelligent agent and the third intelligent agent respectively, so as to obtain multiple interview record information corresponding to the multiple resume information, and the interview record information includes at least two interview question information; generate at least two training samples according to the at least two interview question information, and use the at least two training samples corresponding to each of the multiple interview record information as training samples for generating a target artificial intelligence interview model, and the target artificial intelligence interview model is used to generate interview question information for job seekers during the interview process.

[0098] In an optional embodiment, when the processor 55 obtains multiple job-seeking identity information, it is specifically used to: obtain multiple service dimensions corresponding to the job seeker, the service dimension includes at least two job-seeking category information, and the number of category information included in different service dimensions is the same or different; combine at least two category information included in the multiple service dimensions to obtain at least two groups, and the group includes one category information under each service dimension; generate multiple job-seeking identity information according to the at least two groups and their set ratios, and one job-seeking identity information corresponds to one group.

[0099] In an optional embodiment, when the processor 55 simulates the interviewer and the job seeker to perform interview interactions through the second agent and the third agent respectively to obtain multiple interview record information corresponding to multiple resume information, it is specifically used to: for any resume information, in any round of interview interaction, generate a second prompt word according to the job application answer information of the previous round of interview interaction and the second prompt word template, the second prompt word template includes second task description information for generating interview question information; wherein the job application answer information of the first round of interview interaction is preset information; call the second agent according to the second prompt word to select the interview question of any round of interview interaction from the interview question library according to the second task description information; question information; in the case that the interview question information of any round of interview interaction does not indicate the end of the interview, generate a third prompt word according to the interviewer question information of any round of interview interaction, any resume information and a third prompt word template, the third prompt word template including third task description information for generating job application answer information; call a third agent according to the third prompt word to generate job application answer information of any round of interview interaction according to the third task description information and any resume information, and trigger the next round of interview interaction; in the case that the interview question information of any round of interview interaction indicates the end of the interview, use at least part of the interview interaction information as the interview record information corresponding to any resume information.

[0100] Optionally, the third intelligent agent corresponds to a target parameter for controlling the matching degree between the generated job application answer information and the interview question information; when the processor 55 generates the job application answer information for any round of interview interaction based on the third task description information and any resume information, it is specifically used to: select a target value within the value range of the target parameter in a preset manner; generate the job application answer information for any round of interview interaction based on the third task description information, any resume information and the target parameter.

[0101] Optionally, the second prompt word template includes a task prompt word template, and the task prompt word template includes second task description information; when the processor 55 generates the second prompt word based on the job-seeking answer information of the previous round of interview interaction and the second prompt word template, it is specifically used to: generate the second prompt word based on historical interview interaction information and the task prompt word template; wherein the historical interview interaction information is the interview interaction information before any round, and the historical interview interaction information includes the job-seeking answer information of the previous round of interview interaction.

[0102] Further optionally, the second prompt word template also includes a configuration prompt word template; before generating the second prompt word based on the job application answer information of the previous round of interview interaction and the second prompt word template, the processor is also used to: generate configuration prompt words based on multiple interview question information in the interview question library and the configuration prompt word template; call the second intelligent agent according to the configuration prompt word to initialize the configuration of the second intelligent agent.

[0103] In an optional embodiment, the job-seeking identity information includes at least service type information and service content information; when the processor 55 generates multiple first prompt words based on multiple job-seeking identity information and the first prompt word template, it is specifically used to: for any job-seeking identity information, obtain service type background information and service content background information respectively according to the service type information and service content information; fill the job-seeking identity information, service type background information and service content background information into the first prompt word template to obtain the first prompt word corresponding to any job-seeking identity information.

[0104] In an optional embodiment, when the processor 55 generates at least two training samples based on at least two interview question information, it is specifically used to: for any interview question information, use the interview interaction information located before any interview question information in the interview record information to which it belongs as a sample feature, and use any interview question information as a sample label, so as to obtain a training sample corresponding to any interview question information.

[0105] Regarding the embodiments of the present application Figure 5 The detailed implementation and beneficial effects of each step in the illustrated device have been described in detail in the aforementioned embodiments and will not be elaborated on here.

[0106] Further, if Figure 5 As shown, the electronic device also includes: a communication component 56, a display 57, a power component 58, an audio component 59 and other components. Figure 5 Only some components are shown schematically, which does not mean that the electronic device only includes Figure 5 In addition, Figure 5The components in the dashed box are optional components, not mandatory components, and the specific components depend on the product form of the electronic device. The electronic device of this embodiment can be implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone, or an IOT (Internet of Things) device and a smart wearable device (such as a smart watch, a smart bracelet), or a conventional server, a cloud server, or a server array. If the electronic device of this embodiment is implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone, etc., it can include Figure 5 If the electronic device of this embodiment is implemented as a server device such as a conventional server, a cloud server or a server array, it may not include Figure 5 Components within the dashed box.

[0107] The detailed implementation and beneficial effects of the electronic device provided in the embodiments of the present application have been described in detail in the aforementioned embodiments and will not be elaborated here.

[0108] Accordingly, the embodiment of the present application further provides a computer-readable storage medium storing a computer program, which can achieve the above-mentioned Figure 1 The steps in the method embodiment shown may be executed by an electronic device.

[0109] Accordingly, the embodiment of the present application also provides a computer program product, including a computer program / instruction, which, when executed by a processor, causes the processor to implement the above Figure 1 The steps in the method embodiment shown may be executed by an electronic device.

[0110] The above-mentioned memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read only memory (EEPROM), erasable programmable read only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0111] The above-mentioned communication component is configured to facilitate wired or wireless communication between the device where the communication component is located and other devices. The device where the communication component is located can access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G / LTE, 5G and other mobile communication networks, or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wide Band (UWB) technology, Bluetooth (BT) technology and other technologies.

[0112] The above-mentioned display includes a screen, and the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0113] The power supply assembly provides power to various components of the device where the power supply assembly is located. The power supply assembly may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device where the power supply assembly is located.

[0114] The above-mentioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (Microphone, MIC), and when the device where the audio component is located is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in a memory or sent via a communication component. In some embodiments, the audio component also includes a speaker for outputting an audio signal.

[0115] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-readable storage media (including but not limited to disk storage, compact disc read-only memory (Compact Disc Read-Only Memory, CD-ROM), optical storage, etc.) containing computer-usable program code.

[0116] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0117] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0118] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0119] In a typical configuration, a computing device includes one or more processors (Central Processing Unit, CPU), input / output interface, network interface and memory.

[0120] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0121] Computer readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0122] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0123] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A training sample generation method, characterized in that: include: Acquire multiple job-seeking identity information, wherein the multiple job-seeking identity information is distributed among various categories under multiple service dimensions according to a set ratio; Generate a plurality of first prompt words according to the plurality of job-seeking identity information and the first prompt word template, wherein the first prompt words include first task description information for resume generation; Invoke the first agents respectively according to the plurality of first prompt words, so as to generate a plurality of resume information corresponding to the plurality of job-seeking identity information according to the plurality of first task description information included in the plurality of first prompt words; The second agent and the third agent respectively simulate the interviewer and the job seeker to perform interview interaction, so as to obtain a plurality of interview record information corresponding to the plurality of resume information, wherein the interview record information includes at least two interview question information; Based on the at least two interview question information, at least two training samples are generated, and the at least two training samples corresponding to each of the multiple interview record information are used as training samples for generating a target artificial intelligence interview model, and the target artificial intelligence interview model is used to generate interview question information for job seekers during the interview process.

2. The method according to claim 1, characterized in that Get multiple job-seeking identity information, including: Acquire multiple service dimensions corresponding to the job seeker, wherein the service dimensions include at least two job-seeking category information, and different service dimensions include the same or different numbers of category information; Combining at least two pieces of category information included in the multiple service dimensions to obtain at least two groups, each group including one piece of category information under each service dimension; A plurality of job-seeking identity information is generated according to the at least two groups and the set ratio thereof, and one job-seeking identity information corresponds to one group.

3. The method according to claim 1, characterized in that The second agent and the third agent respectively simulate the interviewer and the job seeker to perform interview interaction, so as to obtain a plurality of interview record information corresponding to the plurality of resume information, including: For any resume information, in any round of interview interaction, a second prompt word is generated according to the job application answer information of the previous round of interview interaction and the second prompt word template, wherein the second prompt word template includes second task description information for generating interview question information; wherein the job application answer information of the first round of interview interaction is preset information; Invoke a second agent according to the second prompt word to select interview question information of any round of interview interaction from an interview question library according to the second task description information; If the interview question information of any round of interview interaction does not indicate the end of the interview, generating a third prompt word according to the interviewer question information of any round of interview interaction, any resume information and a third prompt word template, wherein the third prompt word template includes third task description information for generating job application answer information; Invoke a third agent according to the third prompt word to generate job-seeking answer information for any round of interview interaction according to the third task description information and any resume information, and trigger the next round of interview interaction; In the case where the interview question information of any round of interview interaction indicates that the interview is over, at least part of the rounds of interview interaction information is used as the interview record information corresponding to any resume information.

4. The method according to claim 3, characterized in that The third agent corresponds to a target parameter for controlling the matching degree between the generated job application answer information and the interview question information; Generating job search answer information of any round of interview interaction according to the third task description information and any resume information, including: Selecting a target value within the value range of the target parameter in a preset manner; The job search answer information of any round of interview interaction is generated according to the third task description information, any resume information and the target parameter.

5. The method according to claim 3, characterized in that: The second prompt word template includes a task prompt word template, and the task prompt word template includes second task description information; Generate a second prompt word according to the job-seeking answer information of the previous round of interview interaction and the second prompt word template, including: A second prompt word is generated according to the historical interview interaction information and the task prompt word template; wherein the historical interview interaction information is the interview interaction information before any round, and the historical interview interaction information includes the job search answer information of the previous round of interview interaction.

6. The method according to claim 5, characterized in that The second prompt word template also includes a configuration prompt word template; Before generating the second prompt word according to the job-seeking answer information of the previous round of interview interaction and the second prompt word template, the method further includes: Generate configuration prompt words according to the multiple interview question information in the interview question library and the configuration prompt word template; The second agent is called according to the configuration prompt word to initialize the configuration of the second agent.

7. The method according to any one of claims 1 to 6, characterized in that: The job-seeking identity information at least includes service type information and service content information; Generate multiple first prompt words according to the multiple job-seeking identity information and the first prompt word template, including: For any job-seeking identity information, respectively obtain service type background information and service content background information according to the service type information and the service content information; The job-seeking identity information, the service type background information and the service content background information are filled into the first prompt word template to obtain a first prompt word corresponding to any job-seeking identity information.

8. The method according to any one of claims 1 to 6, characterized in that: Generating at least two training samples according to the at least two interview question information includes: For any interview question information, the interview interaction information located before any interview question information in the interview record information to which it belongs is used as a sample feature, and the any interview question information is used as a sample label to obtain a training sample corresponding to the any interview question information.

9. An electronic device, characterized in that: include: Memory and processor; The memory is used to store a computer program; the processor is coupled to the memory and is used to execute the computer program to implement the steps in the method according to any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to implement the steps in the method according to any one of claims 1 to 8.

11. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the processor is caused to implement the steps in any one of the methods of claims 1-8.

Images