Information processing apparatus, information processing method, and program

Abstract

To provide job information desired by job seekers or the like that ensures sufficient matching accuracy even if the job information is acquired by an automatic search in relation to an information processing technology for matching the job information with job-seeking information.SOLUTION: A job posting data formatting unit 102 and a job-seeking condition acquisition unit 104 utilize a generation AI system 122 to format original job posting data and extract job-seeking search conditions for searching the formatted job posting data. Based on this, highly accurate formatted job posting data can be searched for, and recommended job posting data suited to a job-seeking user can be output.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an information processing device, an information processing method, and a program for matching employment information with job-seeking information. [Background technology]

[0002] Conventionally, there is known an information processing technology that matches job information from companies offering jobs with job search information from job seekers (for example, Patent Document 1). This information processing technology includes a first input unit for inputting a plurality of first activity information or a plurality of first task information, a second input unit for inputting a plurality of second activity information or a plurality of second task information, and a matching unit that performs a matching process between the plurality of first activity information and the plurality of second activity information, or between the plurality of first task information and the plurality of second task information. This information processing technology may have an automatic input unit that performs language analysis of job information such as a job posting and automatically inputs multiple second activity information, multiple second task information, or multiple second skill information. In addition, the job information may be published on the Web. In this case, the automatic input unit may automatically search the Web by crawling or scraping, read the job information, perform language analysis of the content of the job information, and input the second activity information or second task information of the job change / transfer candidate. In addition, the job information published on the Web may be stored in a storage unit, and the automatic input unit may automatically input the information using the job information stored in the storage unit. When the mode of reading job information on the Web and inputting the second activity information, second task information, or second skill information is adopted, it is possible to determine matching with job seekers / transfer applicants based on many job information published on the Web, and it is possible to realize matching based on an overwhelming amount of information. In addition, this information processing device may have a questioning unit that provides multiple questions related to multiple activity information, multiple task information, or multiple skill information, and this questioning unit may be an AI chatbot, etc. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-111059 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, in conventional information processing technology for matching, when determining matching with job seekers, etc. based on a large amount of job information acquired by searching the Internet using automatic technology such as crawling and scraping, there are the following problems. Job information acquired by automatic Internet searches is in various formats, and in many cases, for example, the person who fills out the job posting writes it in free text. For this reason, there is variation in the notation of the acquired job information, and even if such data is collected, it often does not have value as a database. For this reason, there is a problem that automatic matching using such job information cannot ensure sufficient matching accuracy, and it is not always possible to provide the job information that job seekers, etc. want.

[0005] Therefore, an object of the present invention is to ensure sufficient matching accuracy even for job information obtained by an automatic search, and to provide job seekers with optimal job information. [Means for solving the problem]

[0006] An information processing device according to one embodiment includes a job posting original data acquisition unit that acquires original job posting data and stores it in a job posting original database; and a job posting data formatting unit that generates first query data for each of the original job posting data stored in the job posting original database, the first query data including one or more first classification instruction prompts that instruct a first classification into one or more first metadata and a first input prompt including text data of the original job posting data, queries a first generative artificial intelligence system using the first query data to obtain first response text data in which the text data of the original job posting data has been categorized and formatted into each of the one or more first metadata, and stores a new record of formatted job posting data in the formatted job posting database, the text data in the first response text data that has been categorized and formatted into each of the first metadata as values ​​of each database field corresponding to each of the first metadata.

[0007] In addition, an information processing device according to another example of the aspect generates second query data including one or more second classification instruction prompts that instruct a second classification into one or more second metadata and a second input prompt that includes text data of input job search information data, and queries a second generative artificial intelligence system using the second query data to obtain second response text data in which the text data of the job search information data is classified into one or more second metadata corresponding to one or more database fields of the formatted job posting database; and The system further includes a formatted job posting data search unit that generates a search condition query that specifies text data corresponding to each piece of second metadata in the text data as a search keyword for a database field in the formatted job posting database corresponding to the second metadata, and extracts from the formatted job posting database records of formatted job posting data that match the search condition query, and a recommended job posting data output unit that outputs recommended job posting data corresponding to the job search information data based on the text data of each database field value of the record of formatted job posting data extracted by the formatted job posting data search unit.

[0008] In addition, an information processing device as yet another example of an aspect further includes a job-seeking user dialogue unit that transmits and receives dialogue text data between a user terminal device operated by a user who wishes to search for a job, and the job search condition acquisition unit inputs the dialogue text data received by the job-seeking user dialogue unit from the user terminal device as job search information data. Effect of the Invention

[0009] According to the present invention, even when job information is obtained by an automatic search, sufficient matching accuracy can be ensured, and by extracting job search conditions from job information based on dialogue with job seekers, it is possible to provide job seekers with optimal job information. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a functional block diagram of an embodiment of an information processing device. [Figure 2A] FIG. 13 is a diagram showing an example of text data (salary) of the original job advertisement data. [Figure 2B] FIG. 13 is a diagram showing an example of text data (workplace) of original job advertisement data. [Figure 2C] FIG. 13 is a diagram showing an example of text data (industry type) of original job advertisement data. [Figure 2D] FIG. 13 is a diagram showing an example of text data (job type) of original job advertisement data. [Figure 2E] FIG. 13 is a diagram showing an example of text data (skills) of original job advertisement data. [Figure 3A] FIG. 13 shows an example of a first system prompt (salary). [Figure 3B] FIG. 13 shows an example of a first system prompt (work location). [Figure 3C] FIG. 13 is a diagram showing an example of a first system prompt (business type). [Figure 3D] FIG. 13 shows an example of a first system prompt (job type). [Figure 3E] FIG. 11 is a diagram showing an example of a first system prompt (skill). [Figure 4A] FIG. 11 shows an example of a first few-shot prompt (salary). [Figure 4B] FIG. 11 shows an example of a first few-shot prompt (work location). [Figure 4C] FIG. 13 shows an example of a first few-shot prompt (industry type). [Figure 4D] FIG. 13 shows an example of a first Few-shot prompt (occupation). [Figure 4E] A figure showing an example of a first few-shot prompt (skill). [Figure 5A] FIG. 13 is a diagram showing an example of first response text data (salary). [Figure 5B] FIG. 13 is a diagram showing an example of first response text data (workplace). [Figure 5C]FIG. 13 is a diagram showing an example of first response text data (business type). [Figure 5D] FIG. 13 is a diagram showing an example of first response text data (occupation). [Figure 5E] FIG. 13 is a diagram showing an example of first response text data (skills). [Figure 6] FIG. 13 is a diagram showing an example of a user's comment history. [Figure 7] FIG. 11 illustrates an example of a second system prompt. [Figure 8] FIG. 11 shows an example of a second few-shot prompt. [Figure 9] FIG. 11 is a diagram showing an example of second response text data. [Figure 10] FIG. 2 is a hardware block diagram showing an example of the configuration of a computer that executes an embodiment of an information processing device. [Figure 11] 13 is a flowchart showing an example of a job advertisement data formatting process. [Figure 12] 13 is a flowchart showing an example of a generation AI API access process. [Figure 13] 13 is a flowchart illustrating an example of a job search condition acquisition process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] An embodiment of the present invention (hereinafter referred to as "the present embodiment") will be described in detail below with reference to the drawings. Fig. 1 is a functional block diagram of an information processing device 100 according to the embodiment of the present invention.

[0012] In FIG. 1, a job advertisement original data acquisition unit 101 acquires job advertisement original data and stores it in a job advertisement original database (hereinafter, referred to as “job advertisement original DB”) 110. Specifically, as shown in FIG. 1, the job posting original data acquisition unit 101 automatically accesses multiple job posting data providing websites (hereinafter referred to as "job posting data providing websites") 121, #1 to #N (N: a natural number greater than or equal to 2), connected to the Internet 120, and performs a web scraping process to acquire job posting original data from each job posting data providing website 121. Web scraping is a well-known computer software technique for extracting information from websites connected to the Internet, for example by implementing low-level HTTP (Hyper Text Transfer Protocol) or by embedding a web browser to obtain World Wide Web (WWW) content. Web scraping typically involves the use of bots or crawlers to automatically crawl multiple websites.

[0013] FIG. 2A, FIG. 2B, FIG. 2C, FIG. 2D, and FIG. 2E are diagrams showing examples of job posting original data acquired by job posting original data acquisition unit 101. In these examples, an example is shown in which job posting original data acquisition unit 101 acquires job posting original data by job posting category, such as "salary," "work location," "industry," "job type," or "skill." Of course, job posting original data acquisition unit 101 may acquire job posting original data without depending on job posting category such as these. As shown in FIG. 2A to FIG. 2E, the format of the acquired job posting original data is not unified for each job posting category and further for each job posting data providing website 121, and may vary.

[0014] The job advertisement original data acquiring unit 101 stores in the job advertisement original DB 110 records in which the job advertisement original data by job category as exemplified in FIGS. 2A to 2E is used as is as the value of, for example, one database field.

[0015] In FIG. 1, the job posting data formatting unit 102 generates first inquiry data including one or more first classification instruction prompts that instruct a first classification into one or more first metadata, and a first input prompt that includes text data of the original job posting data.

[0016] More specifically, the first classification prompt is composed of, for example, a first system prompt and a first few-shot prompt as follows.

[0017] The first system prompt includes at least a specification of one or more first metadata used in the first classification and a specification of a data format of the first response text data.

[0018] Figures 3A, 3B, 3C, 3D, and 3E are figures showing examples of a first system prompt specified to generate first inquiry data from the original job posting data of Figure 2A, Figure 2B, Figure 2C, Figure 2D, or Figure 2E. For example, in FIG. 3A, it is specified that the job posting original data in the salary category in FIG. 2A is to be classified into the first metadata consisting of two items, "minimum_monthly_salary" (indicating the minimum salary amount) and "maximum_monthly_salary" (indicating the maximum salary amount), which are of numeric type or null type (indicating the absence of data). In addition, the following sentence specifies that the first classification is to be structured and output in a JSON (Java Script Object Notation) format consisting of a pair of a key indicating the first metadata and a value corresponding to the first metadata. Furthermore, it is specified that if the field of the first metadata is not found, the metadata is not to be specified. In addition, in order to improve the accuracy of the response, the thought process leading to the response is also output, and an explanation of the contents of the JSON format is also requested. Also, in Fig. 3B, it is specified that the original job posting data in the work location category in Fig. 2B is to be classified into one first metadata, which is a string array type "prefectures" (indicating prefectures). The following sentence also specifies that, as in the case of Fig. 3A, the first classification is to be structured and output in a JSON format consisting of a pair of a key indicating the first metadata and a value corresponding to the first metadata. Furthermore, it is specified that if the first metadata field is not found, the metadata is not to be identified. Also, in Fig. 3C, it is specified that the original job posting data in the industry category in Fig. 2C is classified into one first metadata, which is "industries" (indicating the industry) from which multiple selections can be made from the industrial classification indicated by ${industry} in a string array format. The following text also specifies that, as in the case of Fig. 3A, the first classification is to be output structured in a JSON format consisting of a pair of a key indicating the first metadata and a value corresponding to the first metadata. Furthermore, it is specified that if the field of the first metadata cannot be found, the metadata is not to be identified. Furthermore, in Fig. 3D, it is specified that the original job posting data of the occupation category in Fig. 2D is classified into one first metadata, which is "occupations" (indicating occupations) that can be selected from the occupation classification indicated by ${occupation} in a string array type. The following sentence also specifies that the first classification is to be output structured in a JSON format consisting of a pair of a key indicating the first metadata and a value corresponding to the first metadata, as in the case of Fig. 3A. Furthermore, it is specified that the metadata is not to be identified if the field of the first metadata is not found. In FIG. 3E, it is specified that the original job posting data of the skill category in FIG. 2E is classified into four first metadata consisting of "required_soft_skills" (indicating required soft skills), "required_hard_skills" (indicating required hard skills), "welcomd_soft_skills" (indicating welcome soft skills), and "welcomd_hard_skills" (indicating welcome hard skills), each of which is an array type of string. In addition, the following text specifies that the first classification is to be output in a structured JSON format consisting of a pair of a key indicating the first metadata and a value corresponding to the first metadata, as in the case of FIG. 3A. In addition, it is specified that the metadata is not to be identified if the field of the first metadata is not found.

[0019] Thus, the first system prompt includes at least a specification of one or more first metadata to be used in the first classification and a specification of a data format of the first response text data. It should be noted that the first system prompt may also include other systemic condition specifications for the generative AI system 122 .

[0020] Next, the first few-shot prompt causes generative AI system 122, a first generative artificial intelligence system described below, to learn example inputs for a first classification and example responses to the example inputs.

[0021] Figures 4A, 4B, 4C, 4D, and 4E are figures showing examples of a first few-shot prompt specified to generate first inquiry data from the original job posting data of Figure 2A, Figure 2B, Figure 2C, Figure 2D, or Figure 2E. For example, in Figure 4A, when original job posting data in the salary category as exemplified in Figure 4A(a) is entered, the generative AI system 122 described below is instructed to learn to return the first response text data described below as exemplified in Figure 4A(b) in accordance with the first metadata specification and data format specification by the first system prompt in Figure 3A described above. Also, in Figure 4B, when original job posting data for the work location category as exemplified in Figure 4B(a) is entered, the generation AI system 122 described later is instructed to learn so as to respond with the first response text data described later as exemplified in Figure 4B(b) in accordance with the first metadata specification and data format specification by the first system prompt in Figure 3B described above. Also, in Figure 4C, when original job posting data for the industry category illustrated in Figure 4C(a) is input, the generation AI system 122 described later is instructed to learn to respond with the first response text data described later and illustrated in Figure 4C(b) in accordance with the first metadata specification and data format specification by the first system prompt in Figure 3C described above. Furthermore, in Figure 4D, when original job posting data for the job category illustrated in Figure 4D(a) is input, the generative AI system 122 described later is instructed to learn to respond with the first response text data described later and illustrated in Figure 4D(b) in accordance with the first metadata specification and data format specification by the first system prompt in Figure 3D described above. In Figure 4E, when original job posting data for the industry category illustrated in Figure 4E(a) is input, the generation AI system 122 described later is instructed to learn to respond with the first response text data described later and illustrated in Figure 4E(b) in accordance with the first metadata specification and data format specification by the first system prompt in Figure 3E described above.

[0022] Next, the job posting data formatting unit 102 in Figure 1 generates first inquiry data that includes a first classification instruction prompt including the first system prompt illustrated in Figures 3A to 3E and the first few-shot prompt illustrated in Figures 4A to 4E, and a first input prompt including text data of the job posting original data illustrated in Figures 2A to 2E that is sequentially read out from the job posting original DB 110.

[0023] Then, the job posting data formatting unit 102 queries the generative AI system 122, which is a first generative artificial intelligence system, using the generated first query data to obtain first response text data in which the text data of the original job posting data has been classified and formatted into one or more first metadata items.

[0024] Here, the generation AI system 122 is connected to, for example, the Internet 120, and the job posting data reforming unit 102 accesses the application programming interface (hereinafter, referred to as "API") function of the generation AI system 122 via the Internet 120. As the API of the generation AI system 122, for example, the API service of ChatGPT-3.5 or ChatGPT-4 ("GPT" is a trademark application of OpenAI Opco LLC, USA) can be used. In addition, the generation AI system 122 may be constructed locally within the organization in which the information processing system 100 is operated and accessed via an in-organization network such as a local area network.

[0025] Figures 5A, 5B, 5C, 5D, and 5E are figures showing examples of first response text data obtained as a result of accessing the generation AI system 122 with each of the first inquiry data including, respectively, the first system prompt illustrated in Figure 3A, 3B, 3C, 3D, or 3E, the first few-shot prompt illustrated in Figure 4A, 4B, 4C, 4D, or 4E, and the original job posting data illustrated in Figure 2A, 2B, 2C, 2D, or 2E. In each of the first response text data illustrated in Figures 5A to 5E, it can be seen that the text data of the original job posting data, which was originally in various formats as illustrated in Figures 2A to 2E, has been neatly categorized and reformatted into one or more first metadata items in a format conforming to the first system prompt illustrated in Figures 3A to 3E and the first few-shot prompt illustrated in Figures 4A to 4E, which constitute the first classification instruction prompt.

[0026] Next, the job advertisement data formatting unit 102 in Figure 1 stores a new record of formatted job advertisement data in the formatted job advertisement DB 111, the new record having text data classified and formatted for each of the first metadata in the first response text data as values ​​for each database field corresponding to each of the first metadata. The job posting data formatting unit 102 stores a new record of formatted job posting data in the formatted job posting database, the formatted job posting data having, for example, the text data "190000" and "400000" classified and formatted into the two first metadata "minimum_monthly_salary" and "maximum_monthly_salary" illustrated in FIG. 5A returned from the generation AI system 122 for the original job posting data in the salary category illustrated in FIG. 2A as the values ​​of the database fields "minimum_monthly_salary" and "maximum_monthly_salary" corresponding to the above two metadata. At this time, the new record may store information linking the record of the job posting original data (Figure 2A) of the corresponding salary category on the job posting original DB 110 (for example, a record ID (identifier) ​​on the job posting original DB 110 of the record of the job posting original data) for output processing by the recommended job posting data output unit 106 described below. In addition, the job posting data formatting unit 102 stores a new record of formatted job posting data in the formatted job posting database, the formatted job posting data having, as the value of the database field "prefectures" corresponding to the one piece of metadata, the text data "Chiba Prefecture", "Kanagawa Prefecture"" classified and formatted into one first metadata "prefectures" as illustrated in Figure 5B returned from the generation AI system 122 for the original job posting data of the work location category as illustrated in Figure 2B. At this time, as in the case of Figure 5A, the above new record may store information linking the record of the job posting original data (Figure 2B) of the corresponding work location category on the job posting original DB 110 for output processing by the recommended job posting data output unit 106 described below. In addition, the job posting data formatting unit 102 stores a new record of formatted job posting data in the formatted job posting database, the formatted job posting data having, as the value of the database field "industries" corresponding to the one piece of metadata, text data "IT," "Communications," and "Manufacturing (Electrical, Electronic, and Mechanical)"" classified and formatted into one first metadata "industries" as illustrated in FIG. 5C returned from the generation AI system 122 for the original job posting data of the industry category illustrated in FIG. 2C. At this time, as in the case of Figure 5A, the above new record may store information linking it to a record of the job posting original data (Figure 2C) of the corresponding industry category on the job posting original DB 110 for output processing by the recommended job posting data output unit 106 described below. Furthermore, the job posting data formatting unit 102 stores a new record of formatted job posting data in the formatted job posting database, the formatted job posting data having, as the value of the database field "industries" corresponding to a single piece of metadata, text data ""Sales", "Planning / Management", "IT / Communications Engineer"" classified and formatted into a single first metadata "occupations" as illustrated in FIG. 5D returned from the generation AI system 122 for the original job posting data of the job category illustrated in FIG. 2D. At this time, as in the case of Figure 5A, the above new record may store information linking the record of the job posting original data (Figure 2D) of the corresponding job category on the job posting original DB 110 for output processing by the recommended job posting data output unit 106 described below. Then, the job advertisement data formatting unit 102 classifies and formats each text data into the four first metadata “required_soft_skills”, “welcomd_soft_skills”, “required_hard_skills”, and “welcomd_hard_skills” illustrated in FIG. 5E returned from the generation AI system 122 for the job advertisement original data of the skill category illustrated in FIG. 2E, such as ““someone who can cooperate with others in their work”, “someone who can communicate smoothly”, “someone who can solve problems on their own”, “someone who can see things in a positive way and act on them”, etc. A new record of formatted job posting data having the values ​​of the database fields "required_soft_skills", "welcomd_soft_skills", "required_hard_skills", and "welcomd_hard_skills" corresponding to the above four metadata, respectively, "required_soft_skills", "welcomd_soft_skills", "required_hard_skills", and "welcomd_hard_skills", is stored in the formatted job posting database. At this time, as in the case of Figure 5A, the above new record may store information linking the record of the job posting original data (Figure 2E) of the corresponding job category on the job posting original DB 110 for output processing by the recommended job posting data output unit 106 described below.

[0027] As described above, in this embodiment, all original job posting data obtained from the Internet can be formatted into formatted job posting data classified from each perspective corresponding to one or more first metadata specified by, for example, the first system prompt and the first few-shot prompt that constitute the first classification instruction prompt, and stored in the formatted job posting DB111. In this embodiment, by utilizing the excellent text data formatting capabilities of the generation AI system 122, it is possible to convert the original job advertisement data into uniformly formatted job advertisement data.

[0028] In FIG. 1, the job-seeking user dialogue unit 103 transmits and receives dialogue text data to and from a plurality of user terminal devices 123, #1 to #M (M: a natural number of 2 or more), which are connected, for example, via the Internet 120 and operated by users seeking jobs. The job-seeking user dialogue unit 103 transmits and receives LINE messages to and from a LINE ("LINE" is a registered trademark of LINE Corporation of Japan) application running on the user terminal device 123, for example.

[0029] The job-seeking user dialogue unit 103 may, for example, include executing an artificial intelligence chatbot (hereinafter referred to as "AI chatbot") process that generates dialogue text data responsive to dialogue text data received from the user terminal device 123 by using an artificial intelligence program and returns the data to the user terminal device 123.

[0030] The text data of the dialogue that the job-seeking user dialogue unit 103 transmits to and receives from the user terminal device 123 can be stored in a dialogue content database (hereinafter, referred to as “dialogue content DB”) 112 .

[0031] In FIG. 1, the job search condition acquisition unit 104 inputs, as job search information data, for example, a user's statement history received by the job-seeking user dialogue unit 103 from the user terminal device 123 out of the text data of the dialogue described above.

[0032] Fig. 6 is a diagram showing an example of a user's statement history received by the job-seeking user dialogue unit 103 from the user terminal device 123. As shown in Fig. 6, a user who wishes to seek a job can state his or her job search in a conversational format.

[0033] Next, the job search condition acquisition unit 104 generates second inquiry data including one or more second classification instruction prompts instructing a second classification into one or more second metadata, and a second input prompt including text data of the user's comment history, which is the job search information data to be input.

[0034] More specifically, the second classification prompt is composed of, for example, the following second system prompt and second few-shot prompt.

[0035] The second system prompt includes at least a specification of one or more second metadata used for the second classification and a specification of a data format for the second response text data.

[0036] FIG. 7 is a diagram showing an example of a second system prompt designated for generating second inquiry data from the user's utterance history of FIG. In Figure 7, for example, it is specified that the text data of the user's comment history in Figure 6 is classified into six second metadata items consisting of "desired_occupations" (indicating the desired occupation), which is a string array type and can be selected multiple times only from the occupation classification indicated by ${OCCUPATIONS}, "desired_monthly_salary" (indicating the desired monthly salary) and "desired_annual_salary" (indicating the desired annual salary), which are each a numeric or null type, "desired_prefectures" (indicating the desired prefecture (work location)), which is a string array type that selects a prefecture in Japan, "desired_industries" (indicating the desired industry), which is a string array type and can be selected multiple times only from the industry classification indicated by ${INDUSTRIES}, and "other_confitions" (indicating other conditions), which is a string array type. The following text specifies that the second classification should be output in a structured JSON format consisting of pairs of keys indicating the second metadata and values ​​corresponding to the second metadata, as in the case of Figure 3A, etc. Furthermore, it specifies that in the case of conflicting conditions, the condition displayed later should take precedence. In addition, it requests an explanation of the thought process leading to the response and the contents of the JSON format.

[0037] Thus, the second system prompt includes at least a specification of one or more second metadata to be used in the second classification and a specification of a data format for the second response text data. Additionally, the second system prompt may include other systemic condition specifications for the generative AI system 122.

[0038] Next, the second few-shot prompt causes the generative AI system 122, which is a second generative artificial intelligence system described later, to learn examples of inputs for the second classification and examples of responses to the examples of the inputs. Here, in this embodiment, the second generative artificial intelligence system is the same generative AI system 122 as the first generative artificial intelligence system accessed by the job posting data shaping unit 102 in FIG. 1. This is because, for example, the ChatGPT system that realizes the generative AI system 122 can respond to both the first response text data response to the job posting original data and the second response text data response to the user's remark history. In contrast, in order to obtain each response text data with higher accuracy, the first generative artificial intelligence system and the second generative artificial intelligence system may query different generative AI systems. For example, the job posting data shaping unit 102 in FIG. 1 may query the generative AI system 122 that provides the ChatGPT-3.5 service, and the job search condition acquisition unit 104 may query another generative AI system 122 that provides the ChatGPT-4 service.

[0039] FIG. 8 is a diagram showing an example of a second few-shot prompt designated for generating second inquiry data from the user's comment history in FIG. In Figure 8, for example, when a user's comment history as illustrated in Figure 8(a) is input, the generation AI system 122 described later is instructed to learn so as to return the second response text data described later as illustrated in Figure 8(b) in accordance with the second metadata specification and data format specification by the second system prompt in Figure 7 described above.

[0040] Next, the job search condition acquisition unit 104 in FIG. 1 generates second inquiry data including a second classification instruction prompt including the second system prompt illustrated in FIG. 7 and the second few-shot prompt illustrated in FIG. 8, and a second input prompt including text data of a user's utterance history, which is job search information data input from the job-seeking user dialogue unit 103 in FIG. 1, for example.

[0041] Then, the job search condition acquisition unit 104 in Figure 1 queries the generative AI system 122, which is a second generative artificial intelligence system, using the generated second query data to acquire second response text data in which the text data of the user's comment history is classified into one or more second metadata items corresponding to one or more database fields of the formatted job posting database.

[0042] Here, similar to the case of the job posting data reforming unit 102, the generation AI system 122 is connected to the Internet 120, and the job search condition acquisition unit 104 accesses the API function of the generation AI system 122 via the Internet 120. As described above, the API of the generation AI system 122 can be, for example, the API service of ChatGPT-3.5 or ChatGPT-4.

[0043] Figure 9 is a diagram showing an example of second response text data obtained as a result of accessing the generation AI system 122 with second inquiry data including the second system prompt illustrated in Figure 7, the second few-shot prompt illustrated in Figure 8, and the user's speech history illustrated in Figure 6. In the second response text data illustrated in FIG. 9, for example, the user's speech history illustrated in FIG. 6 is classified into five second metadata items "desired_industries," "desired_occupations," "desired_prefectures," "desired_annual_salary," and "desired_monthly_salary" with the values ​​"None," "IT / communications engineer," "Tokyo" or "Saitama," "3,000,000" yen, and "250,000" yen in accordance with the second system prompt illustrated in FIG. 7 and the second few-shot prompt illustrated in FIG.

[0044] Next, in FIG. 1, the formatted job advertisement data search unit 105 generates a search condition query that specifies text data corresponding to each of the second metadata in the second response text data illustrated in FIG. 9 acquired by the job search condition acquisition unit 104 as search keywords for database fields in the formatted job advertisement DB 111 corresponding to the second metadata.

[0045] The second memory fields "desired_industries", "desired_occupations", "desired_prefectures", and "desired_monthly_salary" illustrated in Fig. 9 correspond to the database fields "industries", "occupations", "prefectures", "minimum_monthly_salary", and "maximum_monthly_salary" in the formatted job posting DB111, respectively (see Figs. 3A to 3E). Note that in the example of Fig. 9, there is no second metadata corresponding to the database field "desired_annual_salary" in the formatted job posting DB111, but of course there may be one.

[0046] According to the above-mentioned correspondence, the formatted job form data search unit 105 extracts from the formatted job form DB 111 records of formatted job form data that match the above-mentioned search condition inquiry. The list of text data classified into each second metadata is obtained by ORing each text data to search condition queries for the corresponding database fields. Also, the search condition queries corresponding to the values ​​classified into the second metadata "desired_monthly_salary" are specified as conditions for searching for records whose values ​​are between the values ​​of the database fields "minimum_monthly_salary" and "maximum_monthly_salary" in the formatted job posting DB111.

[0047] In Figure 1, the recommended job form data output unit 106 outputs recommended job form data corresponding to the job search information data based on text data of the values ​​of each database field corresponding to each of the specified database fields of the records of the formatted job form data extracted by the formatted job form data search unit 105.

[0048] Here, the recommended job form data output unit 106 may combine the original job form data in the original job form DB 110 corresponding to the records of the formatted job form data extracted by the formatted job form data search unit 105 and output it as recommended job form data corresponding to the job search information data.

[0049] Fig. 10 is a hardware block diagram showing an example of the configuration of a computer that executes the functions of the information processing device 100 in Fig. 1. This hardware includes a server computer configuration in which a CPU (Central Processing Unit) 1001, a memory 1002 in which a program is loaded and executed, an external storage device 1003 that stores the program and databases such as the original job posting DB 110, the formatted job posting DB 111, and the dialogue content DB 112 in Fig. 1, and a network interface circuit 1004 that controls access to the Internet 120 in Fig. 1 are all connected to each other by a system bus 1005. Although not shown in particular, input devices such as a keyboard and a mouse and output devices such as a display may also be connected.

[0050] Fig. 11 is a flowchart showing an example of a job advertisement data formatting process that realizes the function of the job advertisement data formatting unit 102 in Fig. 1. This flowchart shows a process of executing a job advertisement data formatting program written in, for example, a Python ("Python" is a registered trademark of the Python Software Foundation, USA) computer programming language that the CPU 1001 in Fig. 10 has loaded from the external storage device 1003 to the memory 1002.

[0051] First, the CPU 1001 sets a library for accessing the generated AI API and API authentication information (step S1101). This library is, for example, a ChatGPT API library provided by Python. This library is, for example, pre-installed in the external storage device 1003 of Fig. 10. In step S1101, the CPU 1001 executes a process of importing the library from the external storage device 1003 of Fig. 10 to the memory 1002, for example, an "import openai" command. Furthermore, in step S1101, the CPU 1001 executes, at the beginning of the program, a command for setting API authentication information data acquired in advance by registering in, for example, the ChatGPT service.

[0052] Next, in step S1102, the CPU 1001 reads the job advertisement original data one record at a time from the job advertisement original DB 110 of Figure 1 stored in the external storage device 1003 of Figure 10, and repeatedly executes the series of processes from step S1102 to step S1108 until it determines in step S1103 that there are no more records to read (NO determination).

[0053] In this repetitive process, the CPU 1001 first sets a first system prompt that includes at least one specification of the first metadata and a specification of the data format of the first response text data (step S1104). This results in the generation of a first system prompt, for example that illustrated in FIGS. 3A-3E above.

[0054] Next, CPU 1001 sets a first few-shot prompt to generative AI system 122 of FIG. 1 to learn examples of input for the first classification and examples of responses to the examples of input (step S1105). This results in the generation of the first few-shot prompt, for example the one illustrated in FIGS. 4A to 4E described above.

[0055] Next, CPU 1001 generates first inquiry data including the first system prompt set in step S1104, the first few-shot prompt set in step S1105, and a first input prompt including text data of the job posting original data read from job posting original DB 110 in step S1102 (step S1106).

[0056] Then, CPU 1001 calls a program for a generated AI API access process by using the first inquiry data generated in step S1106 as an argument (step S1107). This process will be described later with reference to the flowchart in FIG.

[0057] Finally, the CPU 1001 generates insert SQL (Structured Query Language) data (INSERT statement) from the text data classified and formatted for each of the first metadata in the first response text data as illustrated in Figures 5A to 5E described above, and accesses the formatted job posting DB 111 using the insert SQL data to insert a new record (step S1108).

[0058] Thereafter, CPU 1001 returns to the process of step S 1102 , and repeatedly executes the formatting process on the record of the next job advertisement original data read from job advertisement original DB 110 .

[0059] Fig. 12 is a flowchart showing an example of processing of a generated AI API access program, which is called as a subroutine from each process of step S1107 in Fig. 11 and step S1306 in Fig. 13 described later. As with Fig. 11, this flowchart also shows processing for executing a generated AI API access program written in, for example, the Python computer programming language, which the CPU 1001 in Fig. 10 has loaded from the external storage device 1003 to the memory 1002.

[0060] 12, first, the CPU 1001 sets a generation AI model (step S1201). This is, for example, a command to specify either ChatGPT-3.5 (for example, when calling from the flowchart in FIG. 11) or ChatGPT-4 (for example, when calling from the flowchart in FIG. 13) as the generation AI model.

[0061] Next, CPU 1001 sets the query data (the first query data or the second query data described above) passed by the subroutine call from each process of step S1107 in FIG. 11 and step S1306 in FIG. 13 described later to a predetermined variable in memory 1002 (step S1202).

[0062] Then, the CPU 1001 sets, for example, other control parameters for the ChatGPT API to other predetermined variable sets on the memory 1002 (step S1203).

[0063] After the processing of steps S1201 to S1203, CPU 1001 calls a response setting function in the API library loaded in step S1101 of FIG. 11 described above or step S1301 of FIG. 13 described below, using the text data or parameter group set as the variables in steps S1201 to S1203 as arguments (step S1204).

[0064] Next, the CPU 1001 calls a response acquisition function of the API library (step S1205).

[0065] Finally, the CPU 1001 returns the response from the generation AI system 122 (Figure 1, for example, the ChatGPT system) obtained by the call in step S1205 as the first response text data or the second response text data described above, and returns to processing of the flowchart of Figure 11 described above or Figure 13 described below (step S1206).

[0066] Fig. 13 is a flowchart showing an example of a search output process that realizes the functions of the job search condition acquisition unit 104, the formatted job advertisement data search unit 105, and the recommended job advertisement data output unit 106 in Fig. 1. As with Fig. 11 and Fig. 12, this flowchart shows a process in which the CPU 1001 in Fig. 10 executes a search output processing program written in, for example, the Python computer programming language, which the CPU 1001 loads from the external storage device 1003 to the memory 1002.

[0067] First, the CPU 1001 sets a library for accessing the generated AI API and API authentication information (step S1301). This process is the same as the process in step S1101 in FIG.

[0068] Next, CPU 1001 reads text data of the user's comment history, exemplified in FIG. 6, passed from a job-seeking user dialogue process (not shown) that executes the functions of job-seeking user dialogue unit 103 in FIG. 1 (step S1302).

[0069] Next, CPU 1001 sets a second system prompt that includes at least one specification of second metadata and a specification of the data format of second response text data (step S1303). This results in the second system prompt being generated, for example as illustrated in FIG.

[0070] Next, CPU 1001 sets a second few-shot prompt to generative AI system 122 of FIG. 1 to learn examples of input for a second classification and examples of responses to the examples of input (step S1304). As a result, the second few-shot prompt, for example that shown in FIG. 8 above, is generated.

[0071] Next, CPU 1001 generates second inquiry data including the second system prompt set in step S1303, the second few-shot prompt set in step S1304, and a second input prompt including the text data of the user's comment history input from job-seeking user dialogue unit 103 in step S1302 (step S1305).

[0072] Next, CPU 1001 calls a subroutine of the generated AI API access processing program with the second query data generated in step S1305 as an argument (step S1306). This processing is shown in the flowchart of FIG.

[0073] Then, CPU 1001 generates extraction SQL data (SELECT statement) from the second response text data illustrated in FIG. 9, in which the text data of the user's comment history obtained in step S1306 is classified into one or more second metadata corresponding to one or more database fields of formatted job posting DB 111, and uses the extraction SQL data to access formatted job posting DB 111 and extracts formatted job posting data that matches the search criteria (step S1307).

[0074] Furthermore, CPU 1001 extracts original job advertisement data that is associated with each of the extracted formatted job advertisement data from original job advertisement DB 110 (step S1308).

[0075] Finally, CPU 1001 generates recommended job advertisement data from each pair of the extracted formatted job advertisement data and the original job advertisement data, and passes it to a job-seeking user interactive process that executes the function of job-seeking user interactive unit 103 in Fig. 1 (step S1309). This job-seeking user interactive process transmits the recommended job advertisement data passed in step S1309 to the user terminal 103 of the corresponding job-seeking user.

[0076] In the above-described embodiment, the search of the formatted job advertisement DB 111 by the formatted job advertisement data search unit 105 has been described as keyword matching for each predetermined database field. In this case, the keywords of the search conditions can be extracted, for example, based on morphological analysis of the job search information data. In addition to matching based on general morphological analysis, the search process may be performed by converting the natural language words and text sentences of both the job advertisement and the consultation content of the job search information into vectors whose elements are, for example, the number of occurrences or occurrence probability of each phrase, and comparing the vectors to find the similarity between the two.

[0077] As described above, according to this embodiment, not only can the original job advertisement data be formatted, but appropriate search conditions can also be extracted from the job search information data from the user to execute a search process on the formatted job advertisement DB 111, in which case the advanced formatting function and search condition extraction function of the generation AI system 122 can be utilized. This ensures sufficient matching accuracy even for job information acquired by an automatic search, making it possible to provide job information that job seekers and the like desire. [Explanation of symbols]

[0078] 100 Information processing device 101 Job posting original data acquisition department 102 Job posting data formatting department 103 Job Seeker User Dialogue Section 104 Job Search Condition Acquisition Section 105 Formatted Job Posting Data Search Section 106 Recommended Job Listing Data Output Unit 110 Recruitment form original DB 111 Formatted Job Posting DB 112 Dialogue Content DB 120 Internet 121 Job posting data providing website 122 Generative AI System 123 User terminal equipment

Claims

1. A job posting original data acquisition unit that acquires job posting original data and stores it in a job posting original database; a job posting data formatting unit that generates, for each of the original job posting data stored in the original job posting database, first query data including one or more first classification instruction prompts that instruct a first classification into one or more first metadata and a first input prompt including text data of the original job posting data, queries a first generative artificial intelligence system using the first query data to obtain first response text data in which the text data of the original job posting data has been categorized and formatted into one or more of the first metadata, and stores a new record of the formatted job posting data in the formatted job posting database, the new record having the text data in the first response text data categorized and formatted into each of the first metadata as values ​​of each database field corresponding to each of the first metadata; An information processing device comprising:

2. a job search condition acquisition unit that generates second inquiry data including one or more second classification instruction prompts that instruct a second classification into one or more second metadata and a second input prompt that includes text data of input job search information data, and queries a second generative artificial intelligence system using the second inquiry data to acquire second response text data in which the text data of the job search information data is classified into one or more second metadata corresponding to any one or more of the database fields of the formatted job posting database; a formatted job advertisement data search unit that generates a search condition query that specifies text data corresponding to each of the second metadata in the second response text data acquired by the job search condition acquisition unit as a search keyword for a database field in the formatted job advertisement database that corresponds to the second metadata, and extracts a record of the formatted job advertisement data that matches the search condition query from the formatted job advertisement database; A recommended job posting data output unit that outputs recommended job posting data corresponding to the job search information data based on the text data of the values ​​of the database fields of the records of the formatted job posting data extracted by the formatted job posting data search unit; The information processing apparatus according to claim 1 , further comprising:

3. The job-seeking user dialogue unit transmits and receives dialogue text data to and from a user terminal device operated by a user who wishes to seek a job, the job-seeking search condition acquisition unit inputs, as the job-seeking information data, a statement history of the user that is received from the user terminal device by the job-seeking user dialogue unit out of the text data of the dialogue; The information processing device according to claim 2 .

4. The first classification prompt comprises: a first system prompt including at least a designation of one or more of the first metadata to be used in the first classification and a designation of a data format of the first response text data; a first few-shot prompt that causes the first generative artificial intelligence system to learn examples of inputs and examples of responses to the examples of inputs for the first classification; The information processing apparatus according to claim 1 .

5. The second classification prompt comprises: a second system prompt including at least a specification of one or more of the second metadata to be used in the second classification and a specification of a data format of the second response text data; a second few-shot prompt that causes the second generative artificial intelligence system to learn examples of inputs and examples of responses to the examples of inputs for the second classification; The information processing device according to claim 2 .

6. 4. The information processing apparatus according to claim 3, further comprising a dialogue content database for storing text data of the dialogue transmitted to and received from the user terminal device by the job-seeking user dialogue unit.

7. The information processing device according to claim 1, wherein the job posting original data acquisition unit automatically accesses a plurality of job posting data providing websites connected to the Internet and performs a web scraping process to acquire the job posting original data from each of the job posting data providing websites.

8. the first generative artificial intelligence system is connected to the Internet; The job posting data reforming unit accesses an application programming interface function of the first generative artificial intelligence system via the Internet. The information processing device according to claim 1 .

9. the second generative artificial intelligence system is connected to the Internet; The job search condition acquisition unit accesses an application programming interface function of the second generative artificial intelligence system via the Internet. The information processing device according to claim 2 .

10. The information processing device according to claim 2 , wherein the first generative artificial intelligence system and the second generative artificial intelligence system are the same or different generative artificial intelligence systems.

11. The user terminal device is connected to the Internet; the job-seeking user dialogue unit transmits and receives text data of the dialogue between the job-seeking user and the user terminal device via the Internet; The information processing device according to claim 3 .

12. 5. The information processing device according to claim 4, wherein the job-seeking user dialogue unit executes an artificial intelligence chatbot process that generates, by using an artificial intelligence program, text data of the dialogue in response to the text data of the dialogue received from the user terminal device and returns the text data to the user terminal device.

13. The information processing device according to claim 2, wherein the recommended job posting data output unit combines the job posting original data in the job posting original database corresponding to the records of the formatted job posting data extracted by the formatted job posting data search unit and outputs the combined data as recommended job posting data corresponding to the job search information data.

14. A job posting original data acquisition process for acquiring job posting original data and storing it in a job posting original database; a job posting data formatting process for generating, for each of the original job posting data stored in the original job posting database, first query data including one or more first classification prompts instructing a first classification into one or more first metadata and a first input prompt including text data of the original job posting data, and querying a first generative artificial intelligence system using the first query data to obtain first response text data in which the text data of the original job posting data has been formatted and categorized into one or more of the first metadata, and storing a new record of the formatted job posting data in the formatted job posting database, the new record having the text data in the first response text data formatted and categorized into each of the first metadata as values ​​of each database field corresponding to each of the first metadata; An information processing method for performing the above.

15. a job search condition acquisition process for generating second query data including one or more second classification prompts for instructing a second classification into one or more second metadata and a second input prompt including text data of input job search information data, and for acquiring second response text data in which the text data of the job search information data is classified into one or more second metadata corresponding to any one or more of the database fields of the formatted job posting database by querying a second generative artificial intelligence system using the second query data; a formatted job posting data search process that generates a search condition query that specifies text data corresponding to each of the second metadata in the second response text data acquired by the job search condition acquisition process as a search keyword for a database field in the formatted job posting database that corresponds to the second metadata, and extracts a record of the formatted job posting data that matches the search condition query from the formatted job posting database; A recommended job posting data output process that outputs recommended job posting data corresponding to the job search information data based on the text data of the values ​​of each database field of the record of the formatted job posting data extracted by the formatted job posting data search process; The information processing method according to claim 14, further comprising the steps of:

16. further executing a job-seeking user dialogue process for transmitting and receiving dialogue text data between the job-seeking user and a user terminal device operated by the user who desires to seek a job, In the job search condition acquisition process, text data of the dialogue received from the user terminal device in the job-seeking user dialogue process is input as the job search information data. The information processing method according to claim 15.

17. A job posting original data acquisition process for acquiring job posting original data and storing it in a job posting original database; a job posting data formatting process for generating, for each of the original job posting data stored in the original job posting database, first query data including one or more first classification prompts instructing a first classification into one or more first metadata and a first input prompt including text data of the original job posting data, and querying a first generative artificial intelligence system using the first query data to obtain first response text data in which the text data of the original job posting data has been formatted and categorized into one or more of the first metadata, and storing a new record of the formatted job posting data in the formatted job posting database, the new record having the text data in the first response text data formatted and categorized into each of the first metadata as values ​​of each database field corresponding to each of the first metadata; A program for causing a computer to execute the following.

18. a job search condition acquisition process for generating second query data including one or more second classification prompts for instructing a second classification into one or more second metadata and a second input prompt including text data of input job search information data, and for acquiring second response text data in which the text data of the job search information data is classified into one or more second metadata corresponding to any one or more of the database fields of the formatted job posting database by querying a second generative artificial intelligence system using the second query data; a formatted job posting data search process that generates a search condition query that specifies text data corresponding to each of the second metadata in the second response text data acquired by the job search condition acquisition process as a search keyword for a database field in the formatted job posting database that corresponds to the second metadata, and extracts a record of the formatted job posting data that matches the search condition query from the formatted job posting database; A recommended job posting data output process that outputs recommended job posting data corresponding to the job search information data based on the text data of the values ​​of each database field of the record of the formatted job posting data extracted by the formatted job posting data search process; The program according to claim 17, further comprising causing the computer to execute the steps of:

19. further causing the computer to execute a job-seeking user dialogue process for transmitting and receiving dialogue text data between the computer and a user terminal device operated by a user who wishes to seek a job, In the job search condition acquisition process, text data of the dialogue received from the user terminal device in the job-seeking user dialogue process is input as the job search information data.

20. The program of claim 18.

Images