Information processing apparatus, information processing method, and information processing program

Abstract

To provide an information processing apparatus, an information processing method, and an information processing program for allowing for effective matching between a recruiter and a job seeker.SOLUTION: The information processing apparatus comprises: a first acquisition unit which acquires first information on a first user; a second acquisition unit which acquires second information on a second user; and a generation unit which generates a recommendation reason for recommending the second user to the first user or recommending the first user to the second user based on the first information and the second information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and an information processing program. [Background technology]

[0002] Various inventions have been proposed for matching employers and job seekers. For example, one prior invention (see Patent Document 1) proposes a server device that includes: a related keyword storage means for storing related keywords for job postings in association with each job posting; an application information acquisition means for acquiring application information including user information and job search information from a user terminal device; a text information extraction means for extracting text information from the acquired application information; a determination means for determining whether the extracted text information includes related keywords corresponding to the job posting for which the acquired application information is an application; and a motivation evaluation means for evaluating the user's motivation for applying for the job posting based on the determination result. This invention claims to be able to appropriately evaluate a job seeker's motivation regardless of the length of time the job seeker has used the system. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 5271952 Summary of the Invention [Problem to be solved by the invention]

[0004] However, conventional systems still have room for improvement in matching employers and job seekers.

[0005] The present invention has been made in consideration of the above-mentioned problems, and aims to provide an information processing device, an information processing method, and an information processing program that can effectively match employers and job seekers. [Means for solving the problem]

[0006] In order to solve the above problem, the information processing device of the present invention is characterized by comprising a first acquisition unit that acquires first information of a first user, a second acquisition unit that acquires second information of a second user, and a generation unit that generates a recommendation reason for recommending the second user to the first user, or the first user to the second user, based on the first information and the second information. [Effects of the Invention]

[0007] According to the present invention, it is possible to provide an information processing device, an information processing method, and an information processing program that can effectively match employers and job seekers. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a diagram illustrating an example of a schematic configuration of an information processing system according to an embodiment. [Figure 2] 2 is an example of a hardware configuration of a server (information processing device) according to an embodiment. [Figure 3] 10 is an example of a database stored in a storage device of a server according to an embodiment. [Figure 4] 3 is an example of a functional configuration of a server according to the embodiment. [Figure 5] FIG. 10 is a diagram illustrating an example of structuring by a structuring unit of the server according to the embodiment. [Figure 6] FIG. 10 is a diagram illustrating an example of a prompt generated by a generating unit of the server according to the embodiment. [Figure 7] 3 is an example of a hardware configuration of a terminal according to an embodiment. [Figure 8] 3 is an example of a functional configuration of a terminal according to an embodiment. [Figure 9] 10 is an example of a flowchart illustrating an example of information processing by the information processing system according to the embodiment. [Figure 10] 10 is an example of a flowchart illustrating an example of information processing by the information processing system according to the embodiment. [Figure 11]10 is an example of a flowchart illustrating an example of information processing by the information processing system according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, each embodiment of the present invention will be described with reference to the drawings. "Job seekers" (hereinafter also referred to as first users) include candidates for employment, job changes, part-time work, internships (trainees), and part-time workers, as well as candidates for entering into contract work. "Employers" (hereinafter also referred to as second users) include individuals (for example, sole proprietors), various corporations, associations, organizations, individuals, groups, business entities, and the like.

[0010] [Embodiment] First, the configuration of an information processing system 1 will be described with reference to Fig. 1. The information processing system 1 has a configuration in which a server 2, a first user terminal 3, and a second user terminal 4 are connected via a network 5. Note that the information processing system 1 may include any number of servers 2, first user terminals 3, and second user terminals 4. Network 5 may also be connected to a WWW server that provides information in file formats such as HTML, JSON, and CSV (hereinafter, files in these formats will be simply referred to as HTML files), and server 2 may be configured to acquire website information (e.g., text data, audio data, and image data (including video)). Software such as a crawler or scraper can be used to acquire website information. Note that the type of communications network that network 5 is configured as is optional.

[0011] (Server 2) Fig. 2 is a configuration diagram of the server 2. Fig. 2 shows the main hardware configuration of the server 2, and the server 2 has a configuration in which a communication IF 200A, a storage device 200B, and a CPU 200C are connected via a bus (BUS) or the like. The server 2 may also have other input devices such as a keyboard, a mouse, or a touch panel, and a display device such as an LCD monitor or an organic EL monitor. Note that the device does not have to be a CPU as long as it can execute a program.

[0012] The communication IF 200A is an interface for communicating with other devices (for example, a website, the first user terminal 3, the second user terminal 4, etc.). The storage device 200B is, for example, a hard disk drive (HDD) or a semiconductor storage device (solid state drive (SSD)). The CPU 200C controls the server 2 and includes a read only memory (ROM) and a random access memory (RAM), which are not shown.

[0013] 2, the server 2 is configured to include the storage device 200B, but the server 2 may be configured such that another server connected to the server 2 via the network 5 includes all or part of the storage device 200B. In this case, the server 2 references the storage device 200B included in the other server. The information processing program described below may also be configured to be downloadable from the other server connected to the server 2 via the network 5.

[0014] 3 shows an example of a database stored in the storage device 200B of the server 2. The storage device 200B stores information processing programs, various databases, etc. In this embodiment, multiple databases are provided, but a single database may also be used.

[0015] (Pre-trained model) The trained model is a model that can be trained to input the prompt described below to obtain, for a first user, a recommended second user and the reason for that recommendation, and, for a second user, a recommended first user and the reason for that recommendation. A trained model is a trained neural network, such as a large-scale language model (LLM). By providing a large amount of text data to a model with a large number of parameters and training it, an LLM can learn the context and meaning of words in long sentences, which conventional language models have difficulty handling. Based on the training results, an answer to an input is generated by arranging words and sentences that are considered to have a high probability of following a specific word. This makes it possible to apply the model to a variety of tasks by providing prompts containing instructions on the output format and content as input data. The LLM used in this embodiment may be an LLM designed for general-purpose use using prompting, or an LLM fine-tuned for a specific purpose. Furthermore, the language used in this embodiment may be any language, including Japanese and English. The LLM is broadly composed of the following steps (1-1) to (1-4), but this is merely an example, and the LLM does not necessarily have to be composed of the steps below. (1-1) The process of dividing the input sentence into its smallest units (tokenization) (1-2) A process of predicting the probability distribution of the next token (1-3) A process of sampling the next token from the probability distribution of (1-2). (1-4) Repeat steps (1-2) to (1-3) to output a sequence of tokens (sentences).

[0016] (First user DB1) The first user DB1 stores various information about the first user (hereinafter also referred to as first information). The first information includes, for example, information input by the first user as well as information acquired by the server 2 by browsing websites. The browsing is performed, for example, based on information in the destination DB4 described below. The first information also includes, for example, the first user's name, gender, address, contact information (such as a telephone number or email address), as well as information such as a resume, a curriculum vitae, and a behavioral history. Any other information about the first user may be included. The information about the first user may be acquired by any method. The input of the first information may include, in addition to text, audio, video, images, and input in an interactive format such as chat or voice. The first information also includes requirements for a job offer, such as what the user wants to do and what the user does not want to do. Each requirement may be classified as, for example, a "required requirement (such as an absolutely necessary requirement)" or a "recommended requirement (such as a desirable requirement)."

[0017] (Second user DB2) The second user DB2 stores various information about the second user (hereinafter also referred to as second information). The various information about the second user includes, for example, information input by the second user as well as information acquired by the server 2 by browsing websites. The browsing is performed, for example, based on the information in the destination DB4 described below. The second information includes, for example, a name or surname, gender, address or residence, contact information (telephone number, email address, etc.), as well as information such as a job posting, a scout email, and an interest notification. Any other information about the second user may be included. The information about the second user may be acquired by any method. The input of the second information may include, in addition to text, audio, video, images, and input in an interactive format such as chat or voice. The second information also includes requirements for a job seeker, such as what is necessary (for example, skills such as language or programming) and what is not necessary (for example, skills such as language or programming). Each condition may be given a level of strength, such as "essential condition (condition that is absolutely necessary)" or "recommended condition (condition that is good to have)."

[0018] (Template DB3) The template DB3 stores predetermined character strings (hereinafter also referred to as templates). A template is a character string for generating a prompt to be input to a trained model. In this embodiment, a prompt to be input to a trained model is generated by inserting at least a portion of the first information and the second information (here, any of the following conditions is included: a portion or all of the first information, a portion or all of the second information, and a portion or all of the first information and the second information) into a predetermined position in the template character string.

[0019] (Visiting destination DB4) The destination DB4 stores the URLs (Uniform Resource Locators) of websites that the information processing server 2 visits to acquire information. The URLs of the destinations may be registered in advance in the destination DB4 by an administrator of the information processing server 2 or the like. Alternatively, the information processing server 2 may automatically add the URLs of the links of the destination websites as the destination URLs. Here, the websites are not particularly limited, but are preferably sites that individuals have made public on the network, such as blogs, homepages, and SNS (Social Networking Services). This is because it is easy to acquire information about the first and second users.

[0020] (Server 2 function) Fig. 4 is a functional block diagram of the server 2. As shown in Fig. 4, the server 2 has functions such as a receiving unit 201 (accepting unit), a transmitting unit 202, a storage device control unit 203, an acquiring unit 204 (first acquiring unit, second acquiring unit), a structuring unit 205, and a recommendation reason generating unit 206 (hereinafter simply referred to as the generating unit 206). The functions shown in FIG. 4 are realized by the CPU 200C executing an information processing program stored in a ROM (not shown) of the server 2.

[0021] The receiving unit 201 receives information transmitted from the first user terminal 3, the second user terminal 4, a website, and the like.

[0022] The transmitting unit 202 transmits information to the first user terminal 3 and the second user terminal 4, for example.

[0023] The storage device control unit 203 controls the storage device 200B. For example, the storage device control unit 203 controls the storage device 200B to write and read information.

[0024] The acquisition unit 204 acquires, for example, first information and second information. The acquisition unit 204 acquires, for example, information input by the first user and the second user as the first information and the second information. Furthermore, the acquisition unit 204 accesses a website based on a URL stored in the destination DB4, for example, and reads an HTML file provided by the website to acquire website information (for example, text data, audio data, and image data (including video)) as the first information and the second information. Note that the acquisition unit 204 may also acquire information about websites of URLs linked from the accessed website, for example. The acquiring unit 204 may acquire information about the first user and the second user by any method.

[0025] The structuring unit 205 structures the first information and the second information into a predetermined structure. The structuring by the structuring unit 205 will be described below. For example, the structuring unit 205 extracts data such as those exemplified in (3-1) to (3-5) for each of the classification items (2-1) to (2-7) exemplified below. Note that the classification items and data shown below are merely examples and are not limited to the examples shown below. Examples of categories (2-1) Industry, business type, and business model (2-2) Occupation / position (2-3) Career values ​​(social contribution, challenge, independence, etc.) (2-4) Professional skills (Python programming, natural language processing, etc.) (2-5) Soft skills (logical thinking, leadership, etc.) (2-6) Personality and characteristics (motivation to learn, enthusiasm, etc.) (2-7) Working conditions (compensation, place of work, benefits, holidays, etc.) Example data (3-1) Generate a summary for each classification item (3-2) Extract the description for each classification item (3-3) Extract related keywords (3-4) Extract quantities that indicate the degree of skill and experience (e.g., years of experience) (3-5) Extract importance and priority The structuring unit 205 lists the extracted data for each classification item, and does not extract items that do not have content to be extracted. Here, the structuring unit 205 may extract data using the output of a trained model that uses information about the first or second user and a prompt including an extraction instruction. In this case, the structuring unit 205 may extract data simultaneously with the recommendation reason by including the extraction instruction in the prompt when the generating unit 206 (described later) generates a reason. In this case, the structuring unit 205 can generate a summary for each classification item, for example, by using a prompt that generates a summary of the content of each item. Furthermore, when generating a summary, the structuring unit 205 may extract, as the description location, location information (e.g., item, page, number of lines, etc. in a resume or job posting) of the first information and second information in which the description that served as the basis for the summary is written. Furthermore, the structuring unit 205 may use a named entity extraction method to extract data. In this case, related keywords may be registered in advance as a dictionary, and the structuring unit 205 may extract keywords that appear in a sentence. In this case, the location where the keyword is extracted may be extracted as the description location. Furthermore, it is also possible to limit the number of items extracted by the structuring unit 205. For example, the structuring unit 205 may select and extract some items that are particularly high in importance or priority. The structuring unit 205 may also preferentially extract items that are common to the first information and the second information. The structuring unit 205 may also calculate the priority of each of the first information and the second information, and then calculate the priority of the item by combining these. The calculation may also be based on the importance or the number of occurrences of priority keywords.

[0026] FIG. 5 is a diagram showing an example of structuring by structuring unit 205. FIG. 5(a) is an example of the first information before structuring by structuring unit 205. In other words, FIG. 5(a) is an example of the first information acquired by acquisition unit 204. FIG. 5(b) is an example of the first information after structuring by structuring unit 205. The acquired first information (see FIG. 5(a)) becomes structured data (see FIG. 5(b)) by structuring unit 205. Note that the second information is also structured in the same way, so is not shown in the figure.

[0027] The generator 206 generates a prompt by embedding at least a part of the first information and second information structured by the structuring unit 205 into a predetermined character string that is a template stored in the template DB 3. The prompt is made up of, for example, the following elements (4-1) to (4-4). Note that the elements (4-1) to (4-4) shown below are merely examples. The order of the elements is arbitrary. (4-1) At least a part of the (structured) first information and second information (4-2) Output instructions for recommendation reasons (4-3) Output format instructions (4-4) One or more examples of the expected input or output (few-shot prompts) If the trained model has been fine-tuned in advance using examples of expected input and output as training data, the above elements (4-2) to (4-4) are not necessarily required.

[0028] 6A and 6B are diagrams showing examples of prompts generated by the generation unit 206. Fig. 6A is an example of a prompt generated by the generation unit 206. Fig. 6B is a diagram for explaining the prompt generated by the generation unit 206 (see Fig. 6A), and is not generated by the generation unit 206. Next, the generation unit 206 inputs the generated prompt into the trained model to generate a recommendation reason. The recommendation reason includes, for example, the following reasons (5-1) to (5-8). (5-1) Summary of primary and secondary information (5-2) Hopes and conditions that are consistent between the two parties (5-3) Why should you apply or scout? (5-4) Advice on what to talk about in interviews (e.g., strategies for finding a job or company you are interested in) (5-6) Advice on corrections to the first and second information (5-7) The reason why the first user approached the second user (e.g., applied for a job) (5-8) Why did the second user approach the first user (scout, notify of interest, etc.)? Note that a recommendation level may be output in addition to or instead of the recommendation reason. In this case, for example, by including an instruction to output the recommendation level in the prompt, it can be output simultaneously with the recommendation reason. Also, a recommendation level may be output for each classification item structured by the structuring unit 205 (for example, "match level of professional ability: 8, match level of working conditions: 6"). Note that it is preferable to output the recommendation reason from different perspectives for each classification item structured by the structuring unit 205.

[0029] (First user terminal 3) Fig. 7 is a configuration diagram of the first user terminal 3. The first user terminal 3 is, for example, a desktop PC or a notebook PC, but may also be a smartphone or a tablet terminal. Fig. 7 shows the main hardware configuration of the first user terminal 3, which includes a communication IF 300A, a storage device 300B, an input device 300C, a display device 300D, and a CPU 300E connected via a bus (BUS) or the like. Note that the device does not have to be a CPU as long as it can execute a program.

[0030] The communication IF 300A is an interface for communicating with other devices (for example, the server 2, etc.).

[0031] The storage device 300B is, for example, a HDD or a semiconductor storage device. The storage device 300B stores information processing programs and the like. The storage device 300B also stores an identifier (first user terminal ID) for identifying the first user terminal 3. By assigning the first user terminal ID to information transmitted from the first user terminal 3, the server 2 can recognize which first user terminal 3 transmitted the received information. The ID may be an IP (Internet Protocol) address, a MAC (Media Access Control) address, or the like, and may be assigned to the first user terminal 3 by the server 2.

[0032] The input device 300C is, for example, a keyboard, a mouse, a touch panel, or the like, and the first user can input information by operating the input device 300C.

[0033] The display device 300D is, for example, a liquid crystal monitor, an organic EL monitor, etc. The display device 300D displays a screen necessary for using the information processing system 1.

[0034] The CPU 300E controls the first user terminal 3 and includes a ROM and a RAM (not shown).

[0035] Fig. 8 is a functional block diagram of the first user terminal 3. As shown in Fig. 8, the first user terminal 3 has functions such as an input receiving unit 301, a display device control unit 302, a transmitting unit 303, a receiving unit 304, and a storage device control unit 305. The functions shown in Fig. 8 are realized by CPU 300E executing an information processing program stored in ROM (not shown) of the first user terminal 3.

[0036] The input receiving unit 301 receives, for example, information input by the input device 300C.

[0037] Display device control unit 302 controls display device 300D.

[0038] The transmitting unit 303 transmits, for example, the information accepted by the input accepting unit 301 to the server 2.

[0039] The receiving unit 304 receives, for example, information transmitted from the server 2.

[0040] The storage device control unit 305 controls the storage device 300B. For example, the storage device control unit 305 controls the storage device 300B to write and read information.

[0041] (Second user terminal 4) The second user terminal 4 has the same configuration as the first user terminal 3 described with reference to FIGS. 7 and 8, and therefore a duplicated description will be omitted.

[0042] (Operation of information processing system 1) 9 is a flowchart showing an example of information processing by the information processing system 1. Hereinafter, the processing of the information processing system 1 according to this embodiment will be described with reference to FIG.

[0043] (Recommendation reason generation process) 9 is a flowchart showing an example of the recommendation reason generating process, which will be described below with reference to FIG.

[0044] (Step S101) The acquisition unit 204 acquires the first information and the second information. As described above, the acquisition unit 204 acquires information input by the first user and the second user, information on a website, and the like, as the first information and the second information. As described above, the first information and the second information may include any information relating to the first user and the second user. Furthermore, the acquisition unit 204 may acquire the information relating to the first user and the second user by any method.

[0045] (Step S102) The structuring unit 205 structures the acquired first information and second information into a predetermined structure.

[0046] (Step S103) The generator 206 generates a prompt by embedding at least a part of the first information and the second information in a predetermined character string that is a template stored in the template DB3.

[0047] (Step S104) The generation unit 206 inputs the generated prompt into the trained model.

[0048] (Step S105) The generation unit 206 obtains the recommendation reason from the trained model.

[0049] (Step S106) The transmitting unit 202 transmits the generated recommendation reason to the user terminals 3 and 4.

[0050] The recommendation reason may be output together with source information (such as where the information was obtained) that indicates the basis for the recommendation reason or the part of the first information or second information that serves as a reference. Of the information on the description parts of the classification items of the first information and second information structured by the structuring unit 205, the description parts of the classification items used to output the recommendation reason may be output as exhibit information.

[0051] As described above, the server 2 of this embodiment includes an acquisition unit (first and second acquisition units) that acquires the first information of the first user and the second information of the second user, and a generation unit 206 that generates a recommendation reason for recommending the second user to the first user, or the first user to the second user, based on the first information and the second information. In this way, the reason for the recommendation can be known, which is very convenient.

[0052] The generation unit 206 of the server 2 generates a prompt by embedding at least a part of the first information and the second information in a predetermined character string, and inputs this prompt to the trained model to obtain a recommendation reason. In this way, prompts are generated and input into a trained model, which is expected to generate more accurate recommendation reasons.

[0053] The server 2 includes a structuring unit 205 that structures the first information and the second information into a predetermined structure. In this way, since the first information and the second information are structured into a predetermined structure, it is expected that a more accurate reason for recommendation will be generated.

[0054] [Modification 1 of the embodiment] Furthermore, in the above-described embodiment, the information obtained by structuring the first information and the second information by the structuring unit 205 may be used as registration conditions or search conditions for job applications or recruitment. In this case, as described in the embodiment, input of the first information and the second information may include, in addition to text, audio, video, images, and input in an interactive format such as chat or voice. The first information may also include requirements for a job, such as what the user wants to do and what the user does not want to do. Each condition may be classified as, for example, "essential condition" or "recommended condition." Information entered by the first or second user may also be input into a trained model to generate optimal search conditions. In this case, it is preferable that the data structured by the structuring unit 205 be saved as search conditions or registration conditions, and that the saved data can be presented by the first or second user or used for searches.

[0055] [Modification 2 of the embodiment] In the above embodiment and modification 1, if information is insufficient when generating a recommendation reason, a question for acquiring the missing information may be generated for the first and second users. In modification 2, the server 2 includes a missing information detection unit 207, a question unit 208, etc.

[0056] The missing information detection unit 207 detects information that is missing from the first information and the second information to generate a recommendation reason. For example, the missing information detection unit 207 may detect classification items that could not be extracted when the structuring unit 205 structured the first information and the second information as missing information. Alternatively, the missing information detection unit 207 may detect missing information based on the number of items that the structuring unit 205 was able to extract or the combination conditions. For example, the insufficient information detection unit 207 may detect information that is insufficient for generating a reason for recommendation based on conditions such as "three or more specialized skills are required" or "career values ​​or working conditions are essential." Furthermore, for example, when a certain item exists in the first information (or the second information) but does not exist in the second information (or the first information), the missing information detection unit 207 may detect that information on that item is missing from the first information (or the second information). Furthermore, for example, the insufficient information detection unit 207 may detect information that is insufficient for generating a recommendation reason based on whether the number of characters in the information for each item is equal to or greater than a certain number. Furthermore, the missing information detection unit 207 may detect information that is missing for generating a reason for recommendation based on criteria related to the content of the description (for example, whether specific figures are included in the achievements in the work history, etc.). In this case, the relevance may be calculated by inputting the information of the first user (first information) or the information of the second user (second information) and a prompt including an instruction to determine the missing information into the trained model. In this case, the generation unit 206 may include an instruction to determine the missing information in the prompt when generating a reason, thereby detecting the information that is missing for generating a reason for recommendation simultaneously with the reason for recommendation (note that "simultaneously" does not necessarily mean strictly simultaneously).

[0057] (Question handling) 10 is a flowchart showing an example of the question processing, which will be described below with reference to FIG.

[0058] (Step S201) The missing information detection unit 207 of the server 2 detects whether or not there is information missing for generating a recommendation reason in the first information and second information acquired by the acquisition unit 204. If there is missing information (YES), the server 2 executes the process of step S202. If there is no missing information (NO), the server 2 ends the process.

[0059] (Step S202) The questioning unit 208 of the server 2 instructs the sending unit 202 to send a question required to acquire information that is lacking for generating a recommendation reason. The sending unit 202 sends the question to the first user terminal 3 or the second user terminal 4 based on the instruction.

[0060] (Step S203) The acquisition unit 204 of the server 2 determines whether or not an answer to the question has been received from the first user terminal 3 or the second user terminal 4. If an answer has been received, the acquisition unit 204 acquires the answer.

[0061] In this second variant, the questioning unit 208 may store questions in a question DB in advance and extract them, or may generate questions by inputting the missing information into a trained model that has learned appropriate questions for the missing information.

[0062] [Modification 3 of the embodiment] Furthermore, in the above embodiment and Modifications 1 and 2, the information generated from the first information and the second information may be editable. In this case, in Modification 2, the missing information detected by the missing information detection unit 207 may be displayed, or the display position of the missing item may be changed or highlighted to prompt the user to input the missing information. Editing may also be accepted as an editing instruction from a user. In this case, the first user or the second user issues an editing instruction in natural language (e.g., "Suggest more technically specialized job openings") based on the search conditions, registration conditions, generated recommendation reasons, and recommendation results, and the editing instruction and the information before editing are input, and edited conditions can be generated using a trained model.

[0063] (Editing acceptance processing) 11 is a flowchart showing an example of an edit acceptance process, which will be described below with reference to FIG.

[0064] (Step S301) The receiving unit 201 of the server 2 determines whether or not an editing operation for the first information or the second information has been received from the first user terminal 3 or the second user terminal 4. If an editing operation has been received (YES), the server 2 executes the process of step S302. If an editing operation has not been received (NO), the server 2 ends the process.

[0065] (Step S302) The server 2 reflects the received edited content. The storage device control unit 203 stores the received edited content in the first user DB1 or second user DB2 of the storage device 200B. The stored edited content is used when generating a recommendation reason in the embodiment.

[0066] [Fourth Modification of the Embodiment] In addition, in the third modification, the first information or the second information may be corrected based on the edited content. In this case, the first information or the second information may be corrected based on the edited content by, for example, the methods shown in the following (6-1) to (6-4). Note that the methods shown in the following (6-1) to (6-4) are merely examples, and the first information or the second information may be corrected based on the edited content by other methods. (6-1) Add editing information to the first information and / or second information. (6-2) The portion of the first information and / or the second information detected by the missing information detection unit 207 is replaced with the editing information. (6-3) The edited portion and the first information and / or second information before editing are inputted and the first information and / or second information after editing is obtained by LLM. (6-4) The first information or the second information may be directly corrected, or corrections may be suggested so that the user can make corrections.

[0067] [Fifth Modification of the Embodiment] In addition, in the above embodiment and modifications 1 to 4, the second user may be recommended to the first user, or the first user may be recommended to the second user, taking into consideration related users. In this case, the server 2 may include a relatedness calculation unit 209 that calculates the relatedness between users. In this case, for example, the second user may be recommended to the first user, or the first user may be recommended to the second user, by the methods shown in (7-1) to (7-4) below. Note that the methods shown in (7-1) to (7-4) below are merely examples, and other methods may be used to recommend the second user to the first user, or the first user to the second user. (7-1) Recommend a second user who is highly related to the first user. (7-2) Recommend a first user who is highly related to a second user. (7-3) Recommend second users who are matched (scouted, applied for, accepted, etc.) with other first users who are highly related to the first user. (7-4) Recommend first users who are matched (scouted, applied for, accepted, etc.) with other second users who are highly related to the second user. In addition, any combination of information such as the first information and the second information, the first information and the second information structured by the structuring unit 205, and the behavioral history of the first user and the second user (such as sending scouts and applying) can be used to calculate the relevance. Known methods can be used to calculate the relevance. For example, Euclidean distance of vectorized information, Pearson's product-moment correlation coefficient, Jaccard coefficient, etc. can be used. The number of matching words or existing graph features can also be used to calculate the degree of match. Nodes that are particularly emphasized by the user can be given a higher weight when calculating the degree of match. Relevance calculation and recommendations can also be performed using a matrix decomposition-based method such as LightFM or an emphasis filtering method. Alternatively, the relevance may be calculated by inputting a prompt including information about the first and second users and an instruction to calculate the relevance to the trained model. In this case, the relevance may be output simultaneously with the recommendation reason by including the instruction to calculate the relevance in the prompt when the generation unit 206 generates the reason. Furthermore, the recommendation level described with reference to FIG. 6 may be used as the relevance level. In addition, recommendations may be made for each perspective according to the degree of recommendation for each item (for example, "jobs that match your specialized skills," "jobs that match your working conditions," etc.). In addition, the recommended results may be displayed to the first user or the second user, and feedback on the recommended results (such as a two-point rating of positive / negative or a ten-point rating) may be obtained, which may be used as learning data when calculating the relevance.

[0068] [Modification 6 of the embodiment] In the above embodiment and modifications 1 to 5, the number of recommendation reasons that can be output at one time may be limited. For example, a recommendation reason may be generated only for the combination of the first user and the second user recommended in the fifth modification. This is effective when generation resources (processing power of AI servers, etc.) are limited.

[0069] [Seventh Modification of the Embodiment] Furthermore, in the above embodiment and modifications 1 to 6, the timing for presenting the recommendation reason may be determined. Possible timings include, for example, the following (8-1) to (8-4). Note that the timings shown in the following (8-1) to (8-4) are merely examples, and the recommendation reason may be presented at other timings. (8-1) When the number of unread or unresponsive reasons for recommendation that have already been presented falls below a specified number. (8-2) When the primary information or secondary information is updated. (8-3) When the First User or the Second User becomes motivated to change jobs or become employed. (8-4) When a scout, interest notification, application, message exchange, etc. is conducted between the first user and the second user. By timing the presentation of the recommendation reason in this way, it is possible to provide the recommendation reason at an appropriate time even when resources for generating the recommendation are limited.

[0070] [Eighth Modification of the Embodiment] In addition, in the above embodiments and variants 1 to 7, a strategy for getting a job in a desired job type or company (for example, a plan for acquiring the required skills and work experience) may be presented based on the difference between the skills and work experience currently possessed by the first user and the skills and work experience required in the desired job type or company.

[0071] [Modification 9 of the embodiment] In addition, in the above embodiments and variants 1 to 8, when a scout or interest notification is received from a second user, the reason for the scout or interest notification may be presented, and when an application is received from a first user, the reason for the application may be presented.

[0072] [Modification 10 of the embodiment] In addition, in the above-described embodiment and modifications 1 to 9, it is also possible to present the items (categories) and contents of one's work history that are often cited as reasons for recommendation across multiple job offers. By configuring in this way, there is an advantage in that the user (oneself) can know the points that they (themselves) tend to focus on.

[0073] [Other embodiments] Although the present embodiment and its modifications have been described above, the present invention is not limited to these examples. For example, any combination of features of the above-described embodiments and modifications, as well as any addition, deletion, or design change of components by a person skilled in the art, are also encompassed within the scope of the present invention as long as they include the gist of the present invention. [Explanation of symbols]

[0074] 1. Information Processing Systems 2 Server 3. First user terminal 4. Second user terminal 5. Network 200A communication interface 200B storage device 200C CPU 201 Receiving Department (Reception Department) 202 Transmission Unit 203 Storage device control unit 204 Acquisition Department (1st Acquisition Department, 2nd Acquisition Department) 205 Structuring part 206 Recommendation reason generation unit (generation unit) 207 Missing Information Detection Unit 208 Questions 209 Relevance calculation unit 300A communication IF 300B storage device 300C input device 300D display device 300E CPU 301 Input Reception Unit 302 Display device control unit 303 Transmission Unit 304 Receiving Unit 305 Storage device control unit

Claims

1. a first acquisition unit that acquires first information of a first user; a second acquisition unit that acquires second information of a second user; a generation unit that generates a recommendation reason for recommending the second user to the first user or the first user to the second user based on the first information and the second information; An information processing device comprising:

2. The generation unit generating a prompt by embedding at least a part of the first information and the second information in a predetermined character string, and inputting the prompt to a trained model to obtain the recommendation reason; 2. The information processing apparatus according to claim 1, wherein:

3. a structuring unit that structures the first information and the second information into a predetermined structure; 2. The information processing apparatus according to claim 1, further comprising:

4. a receiving unit that receives editing of the result by the structuring unit; 4. The information processing apparatus according to claim 3, further comprising:

5. The generation unit generating a recommendation reason for recommending the second user to the first user or the first user to the second user based on the first information and the second information after structuring by the structuring unit; 4. The information processing apparatus according to claim 3,

6. an association degree calculation unit that calculates an association degree with the first user or the second user; a recommendation unit that recommends the second user to the first user or the first user to the second user based on the relevance; The generation unit generating a recommendation reason for recommending the second user to the first user by the recommendation unit, or for recommending the first user to the second user by the recommendation unit; 2. The information processing apparatus according to claim 1, wherein:

7. a step of a first acquisition unit acquiring first information of a first user; a step of a second acquisition unit acquiring second information of a second user; a generation unit generating a recommendation reason for recommending the second user to the first user or the first user to the second user based on the first information and the second information; An information processing method comprising:

8. Computer, a first acquisition unit that acquires first information of a first user; a second acquisition unit that acquires second information of a second user; a generation unit that generates a recommendation reason for recommending the second user to the first user or the first user to the second user based on the first information and the second information; An information processing program characterized by causing the program to function as:

Images