Science business matching apparatus, method, and program

Abstract

We provide a science business matching system that efficiently matches researchers with businesses, supports collaboration and employment between the two, and facilitates the smooth establishment of relationships between researchers and businesses. [Solution] The science business matching device includes a machine learning model that causes a computer to calculate a matching score for researchers with businesses registered in a database. The machine learning model is a natural language processing model and is trained using either the researcher's field of specialization information and the business's technology field information, or the matching score between the field of specialization information and technology field information determined by an expert, as input data, and the other as ground truth data.

Description

Technical Field

[0001] The present invention relates to a "Science to Business" system that efficiently matches researchers with business entities (such as companies, universities, research institutions, etc.) by leveraging AI technology, and particularly to a technology that appropriately aligns the expertise of researchers with the technical needs of business entities and maximizes collaboration opportunities.

Background Art

[0002] Currently, collaboration between business entities and researchers is not being carried out efficiently due to the difficulty of information sharing and appropriate matching. It is difficult for researchers to find business entities suitable for their expertise, and business entities also struggle to find researchers optimal for their technical needs. To solve such problems, a platform that efficiently links the expertise of researchers with the technical needs of business entities is required.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

[0004] Patent Document 1 discloses a technique for estimating the business stage of a business conducted by a specific applicant from the document information of the patent gazette of the specific applicant using AI.

[0005] Currently, efficient matching based on the detailed fields of expertise of researchers and the technical needs of business entities has not been carried out. In particular, the development of an automatic and precise matching technology utilizing AI has not advanced.

Summary of the Invention

Problems to be Solved by the Invention

[0006] The objective of this invention is to efficiently match researchers with businesses using AI technology, and to support researchers in collaborating with or applying to businesses best suited to their expertise. Furthermore, it aims to support businesses in quickly finding researchers best suited to their technological needs. Through these efforts, the aim is to support researchers' career development and accelerate the social implementation of research results. [Means for solving the problem]

[0007] This invention provides a "science to business" platform that uses AI algorithms to efficiently match researchers with businesses. It employs the following configuration:

[0008] First aspect: Machine learning model A machine learning model that causes a computer to calculate a matching score for researchers against business entities registered in a database, The aforementioned machine learning model is a natural language processing model. The aforementioned machine learning model is trained using information on researchers' areas of specialization and information on the technical fields of the business entity as input data, and matching scores between the area of ​​specialization information and the technical fields of the business determined by experts as ground truth data.

[0009] The aforementioned specialization information includes at least one of the researcher's current and / or past affiliations, specialization, years of experience, years of service, titles and / or abstracts of publications and / or patent documents, and conference presentation information. The aforementioned technical information includes at least one of the entity's business field, the titles and / or abstracts of papers and / or patent documents, and the specialization information of the desired researcher.

[0010] The aforementioned machine learning model, The characteristic quantities of the researcher's field of specialization determined based on the aforementioned field of specialization information, The matching score is calculated based on the degree of agreement with one of the feature quantities in the multiple technological fields possessed by the business entity.

[0011] Second aspect: Prompt generation program In a prompt generation program that causes a computer to generate prompts, The aforementioned prompt is intended to cause a large-scale language model to calculate a matching score for the researcher against the entities registered in the database. The steps include obtaining information on the aforementioned researchers and information on the aforementioned business entity, The steps include extracting information on the researcher's field of specialization from the researcher's information and information on the technical field from the business entity's information, The steps include obtaining information on multiple relevant matching perspectives from the information of the aforementioned researchers and the information of the aforementioned business entities, The steps include selecting one or more of the aforementioned multiple perspective information to determine the perspective for calculating the matching score, A prompt generation program that performs the steps of generating a prompt including the aforementioned field of specialization information, the aforementioned technical information, and the aforementioned viewpoint.

[0012] The aforementioned specialization information includes at least one of the researcher's current and / or past affiliations, specialization, years of experience, years of service, titles and / or abstracts of publications and / or patent documents, and conference presentation information. The prompt generation program according to claim 4, wherein the technical information includes at least one of the business field of an entity, the titles and / or abstracts of papers and / or patent documents, and the specialization information of a desired researcher.

[0013] Third aspect: Science business matching device An input unit that accepts input information on the researcher's area of ​​specialization, A database containing technical information of business entities, A memory unit for storing the above machine learning model, A processing unit that executes the aforementioned machine learning model, A display data creation unit that creates display data including information on researchers or organizations with high matching scores calculated by the aforementioned machine learning model, A science-business matching device comprising a transmission unit that transmits the display data to the terminal of the researcher or the business entity.

[0014] An input unit that receives input of the researcher's field of specialization information A database having the technical field information of the business entity A storage unit that stores the above prompt generation program A processing unit that executes the prompt generation program to cause the large language model to calculate a matching score A display data creation unit that creates display data including information on researchers or business entities with a high matching score calculated A science-business matching device comprising a transmission unit that transmits the display data to the terminal of the researcher or the business entity.

[0015] The input unit further receives input of the researcher's cooperation desire information and / or desired condition information The display data creation unit creates the display data including the matching score for business entities with a high matching score that match the cooperation desire information and / or the desired condition information.

[0016] The input unit further receives input of the research personnel information and / or cooperation field information required by the business entity It further comprises a researcher information database having input researchers and the researcher's field of specialization information The display data creation unit creates the display data including the matching score for researchers with a high matching score that match the research personnel information and / or the cooperation field information.

[0017] The display data creation unit creates display data including the information that determines the matching score and the matching score between the researcher and the business entity.

[0018] It further includes an information retrieval unit that accesses external databases and retrieves publications and patent information about registered entities and / or researchers. The matching score is calculated using the papers and patent information obtained by the aforementioned information acquisition unit.

[0019] It further includes a contact section that provides online meeting and / or messaging functions between the business entity and the researcher.

[0020] The processing unit notifies researchers who have already entered information of organizations whose matching scores exceed a predetermined threshold among the newly registered organizations.

[0021] The processing unit notifies the business entity that has already entered information of researchers whose matching score exceeds a predetermined threshold among the newly registered researchers.

[0022] Fourth aspect: Science-business matching method A science business matching method performed by the science business matching device described above, To obtain information on researchers' areas of specialization, The aforementioned field of specialization information is compared with the technical information of the business entity stored in the database to calculate a matching score. A science business matching method comprising notifying the aforementioned researchers of information on business entities with high matching scores.

[0023] A science business matching method performed by the science business matching device described above, To obtain information on researchers' desire for collaboration and / or desired conditions, The matching score is calculated by comparing the aforementioned collaboration request information and / or the aforementioned desired conditions information with the technical information of the business entity stored in the database. A science business matching method comprising notifying the aforementioned researchers of information on business entities with high matching scores.

[0024] A science business matching method performed by the science business matching device described above, To obtain information on research personnel and / or collaborative fields sought by the business entity, The process involves comparing the aforementioned research personnel information and / or the aforementioned collaborative field information with the researcher's field of specialization information stored in the researcher information database to calculate a matching score. A science business matching method comprising notifying the business entity of information on researchers with high matching scores.

[0025] Fifth aspect: Computer program A computer program that causes a science business matching device to execute the science business matching method described above. [Effects of the Invention]

[0026] This invention enables efficient matching of businesses and researchers, accelerating the social implementation of research results. Furthermore, businesses can quickly find the most suitable researchers to meet their technological needs, and researchers can gain the opportunity to collaborate with the business best suited to their research. By automating these processes, both parties can significantly reduce time and costs compared to conventional matching methods.

[0027] Furthermore, by using machine learning models to perform highly accurate matching, the success rate of matching is improved compared to conventional technologies, enabling businesses and researchers to build more appropriate collaborative relationships. This technology will be an important means for both businesses and researchers to broaden opportunities for collaboration and to rapidly advance the commercialization and social implementation of research results. [Brief explanation of the drawing]

[0028] [Figure 1] This figure shows a science business matching device according to an embodiment of the present invention. [Figure 2]This diagram shows the workflow for science-business matching for researchers according to an embodiment of the present invention. [Figure 3] This figure shows the operation flow of science business matching for a business entity according to an embodiment of the present invention. [Figure 4] This figure shows the workflow for science-business matching for researchers, according to another embodiment of the present invention. [Modes for carrying out the invention]

[0029] Figure 1 shows a science business matching device 100 according to an embodiment of the present invention, and an example of a terminal connected thereto. The science business matching device 100 is connected to a researcher terminal 200, a business entity terminal 300, and an external database 400 via a network and, if necessary, an API (Application Programming Interface). In addition to these, it may also be connected to other parties' servers, etc.

[0030] The science business matching device 100 includes a database 10 for storing technical information of business entities, a storage unit 20 for storing machine learning models and prompt generation programs (described later), a processing unit 30 for processing using the machine learning models, an input unit 40 for receiving input from researchers and business entities, a display data generation unit 50 for creating display data containing information on researchers or business entities with high matching scores calculated by the machine learning models and prompt generation programs, and a transmission unit 60 for transmitting the display data to the terminals of researchers or business entities. Furthermore, it may also include a researcher information database 70 for storing information on researchers' fields of specialization, an information acquisition unit 80 for accessing an external database and acquiring research field information of registered business entities and / or researchers, and a contact unit 90 for executing online meeting and / or messaging functions between business entities and researchers.

[0031] The basic configuration of the science business matching device 100 may be a known server computer equipped with a processor (CPU, MPU, GPU, or SoC (System on a Chip) in which the processor and other elements are integrated on a single chip), memory, storage device, communication device, display device, input device (keyboard, mouse, etc.). External storage and servers may be used as needed. In addition to machine learning models and prompt generation programs, other necessary programs may be stored in the storage device (storage unit 20).

[0032] Each component of the science business matching device 100 is implemented by appropriately utilizing one or more of the components of the server computer (for example, by utilizing programs, memory, and processors stored in a storage device). In other words, the above components 10 to 90 describe the functions of the science business matching device 100, and do not mean that the above components 10 to 90 exist separately and independently. It is sufficient for them to be implemented using known device configurations and known software and hardware controls.

[0033] The machine learning model is a natural language processing model that is trained using information on researchers' areas of specialization and the technical fields of the organization as input data, and matching scores between the areas of specialization and technical fields determined by experts as ground truth data.

[0034] The machine learning model before training can be any model capable of supervised learning, such as one with a neural network like an RNN or CNN. Alternatively, a trained (known) natural language processing model (e.g., a large-scale language model (LLM)) can be fine-tuned to obtain the trained machine learning model.

[0035] The information on researchers' areas of specialization and the technical fields of organizations used in machine learning may be actual, fictitious, or a combination of both. The same applies to test data. However, actual information on researchers' areas of specialization must be handled appropriately in compliance with the Personal Information Protection Act and other relevant laws, and is presented in this specification on the premise that appropriate consent has been obtained.

[0036] Furthermore, the machine learning model may be subjected to additional training using actual inputs, and the science business matching device 100 may be configured to include an additional training unit for performing this additional training. The configuration of the additional training unit may be the same as that of known units.

[0037] The method for calculating the matching score may be anything, but it can be calculated by determining the degree of agreement between the features of the researcher's field of specialization (e.g., a feature vector) determined based on the researcher's field of specialization information and the features of one of the multiple technical fields possessed by the business entity (e.g., a feature vector). The features of the researcher's field of specialization may also be multiple (e.g., based on input, based on each paper), and the matching score may be calculated using each of the multiple features of the researcher's field of specialization and each of the features of the multiple technical fields possessed by the business entity. In this case, the display data can be created using at least one of the highest, average, or lowest matching scores. The method for creating the display data is not particularly limited and can be any known method, for example, by applying the matching results to a stored display layout.

[0038] When calculating features, data input by researchers or organizations may be used as is, or preprocessing such as summary generation, keyword extraction, and weighting may be performed using large-scale language models or other programs before calculation. The same applies to papers, patent information, etc., obtained by methods other than input.

[0039] Furthermore, if the features are stored in database 10 and researcher information database 70 in association with information about researchers or organizations, the efficiency of calculating new matching scores will be increased. In addition, the features can be automatically calculated and stored in database 10 and researcher information database 70 even if there is no matching request from organizations or researchers when any input information is added.

[0040] The researcher terminal 200 is a terminal that researchers operate to input their information and display matching results, and can take the form of a personal computer, tablet, smartphone, or other similar device.

[0041] Researcher information may include their name, gender, age, work history, educational background, titles, abstracts, and details of papers, patents, and conference presentations, keywords related to their field of study (free-form or selectable), collaborative research history, desired position information (location, salary, benefits, etc.), and scope of disclosure (content of information, who will receive it, etc.). Furthermore, it may also include documents compiled in the form of a resume or work history. In addition, it may include information on citations of papers and patents, citation information, co-researchers, co-authors, co-inventors, and their fields of study, awards received at academic conferences, presentations at international conferences, keynote speeches, and presentations at technical seminars, achievements in obtaining external funding such as Grants-in-Aid for Scientific Research, official qualifications such as professional engineer or pharmacist, and content posted on social media or blogs (e.g., whether they discuss or share information about their research field (e.g., AI) on social media). In the case of programming fields, it may also include publicly available open-source source code (e.g., GitHub).

[0042] The business entity terminal 300 is a terminal on which the business entity's representatives input information about the business entity, information about desired personnel, etc., and where matching results are displayed. It can take the form of a personal computer, tablet terminal, smartphone, etc.

[0043] Information on the business entity may include the entity's name, outline, business activities, specific products and services, location information, titles, summaries, and details of intellectual property such as patents, other technical keywords (free-form or selectable), information on offered positions (work location, salary, benefits, environment, project overview, desired experience and field of specialization, etc.), information on collaboration requests (whether or not they are interested, desired field, etc.), technical needs, whether or not proposals from researchers are accepted, and the scope of disclosure (content of information, recipients, etc.). Furthermore, it may also include information on citations of papers and patents, citation information, collaborative research partners, co-applicants, and their technical fields, track record of obtaining external funding such as Grants-in-Aid for Scientific Research, competitive bidding information from public institutions, press releases, and information disseminated through IR, etc.

[0044] Summaries and details of papers and other literature may be configured to accept file uploads.

[0045] External database 400 is a database containing information that can be used for evaluation in specialized fields and technical fields, such as academic paper information and patent information, and may include those provided by public institutions or non-public institutions, and those provided for a fee or free of charge. In addition, it may also include databases containing corporate information.

[0046] The information acquisition unit 80 can process information such as papers and patent documents (title, patent publication number, etc.) that can identify researchers or organizations into the external database 400. The information acquisition unit 80 can also search the external database 400 for the researcher's name, organization name, etc., and acquire (add to database 10 and researcher information database 70) any papers or patent information that exists. Such processing can be performed at any time, and the system may periodically crawl to check for any newly added information.

[0047] Furthermore, the information acquisition unit 80 can process the information, such as the paper information, to generate a summary using, for example, a large-scale language model (either one provided by the science business matching device 100 or an external one). In addition, the information acquisition unit 80 can process the acquired and generated information to add it to the database 10, the researcher information database 70, etc.

[0048] The added information can be used to calculate the matching score (or to calculate features). This can improve the matching accuracy.

[0049] Furthermore, the science business matching device 100 can be configured to automatically acquire information on job postings at research institutions (salary, workplace, benefits, etc.) from other parties' servers. This allows researchers to easily find matching scores with job postings and broaden their job search.

[0050] Next, we will explain the workflow for researchers searching for businesses using such a science business matching device, with reference to Figure 2.

[0051] When the science business matching device 100 receives information input from a researcher, processing begins (S101).

[0052] The processing unit 30 processes the input from the researchers using a machine learning model and calculates a matching score for each piece of technical information in the database (S102).

[0053] Here, the matching score may be calculated by determining the degree of agreement between the features of the field of specialization (e.g., vectors) processed by a machine learning model and the features of the technical fields possessed by the business entity recorded in the database (e.g., vectors), or by other means.

[0054] The processing unit 30 extracts the business entity with the highest matching score in technical field information (S103).

[0055] The processing unit 30 creates a matching score and display data for displaying information about the extracted business entities (S104).

[0056] The processing unit 30 transmits the display data to the researcher's terminal (S105).

[0057] The researcher's input data may be processed so as to be stored in a database, and in this case, vectorized information may also be included in the storage.

[0058] A single display data entry may contain information on one business entity or multiple entities. Furthermore, multiple display data entries may be created, such as one for each business entity.

[0059] Furthermore, when extracting the business entity with the highest matching score in technical field information, the technical information that was considered important may be included in the displayed data, and the reasons for considering it important may also be included in the displayed data. The reasons for considering it important may be in a format that fits into a standard phrase, or they may be generated by a text generation model.

[0060] When researchers are matching with potential employers, a step may be inserted between S102 and S103 to filter the data using the researcher's entered preferences (income, work location, position, etc.) and the conditions presented by the employers. Furthermore, the preferences can be entered in multiple stages, allowing for the creation of display data with different filtering conditions.

[0061] Furthermore, a step may be inserted between S101 and S102 to process the input papers, patent information, etc., and retrieve details and summaries of the papers, etc., from an external database 400.

[0062] Next, we will explain the workflow for businesses to search for researchers using such a science business matching device, with reference to Figure 3.

[0063] When the science business matching device 100 receives input from a business entity regarding the researcher information it desires, processing begins (S201).

[0064] The processing unit 30 processes the input from the business entity using a machine learning model and calculates a matching score for each of the researcher's specialization information present in the database (S202).

[0065] Here, the matching score may be calculated by determining the degree of agreement between the technology vector, which is obtained by processing the input using a machine learning model, and the researcher's specialization vector, which is recorded in the database, or by other means.

[0066] The processing unit 30 extracts researchers with the highest matching score for the major information (S203).

[0067] The processing unit 30 creates display data for displaying the matching score and information about the extracted researchers (S204).

[0068] The processing unit 30 transmits the display data to the business entity terminal (S205).

[0069] The input from this entity may be processed so as to be stored in a database, and in this case, vectorized information may also be stored.

[0070] A single display data entry may contain one or more researcher information entries. Furthermore, multiple display data entries may be created, such as one for each researcher.

[0071] When an organization matches researchers for recruitment, it may insert a step between S202 and S203 to filter the results using the researcher's pre-entered preferences (income, work location, position, etc.) and the organization's presented criteria. Furthermore, the preferences can be entered in multiple stages, allowing for the creation of display data based on different filtering conditions.

[0072] Furthermore, a step may be inserted between S201 and S202 to process the acquisition of details and summaries of patent information, etc., from an external database 400 using the name of the business entity, etc.

[0073] Next, we will explain the workflow for researchers searching for businesses using a science business matching device equipped with a prompt generation program, with reference to Figure 4.

[0074] When the science business matching device 100 receives information input from a researcher, processing begins (S301).

[0075] The science business matching device 100 acquires the input researcher information and the business entity information stored in the database (S302).

[0076] This information may be standardized in advance. Standardizing the field information, career information, etc., used to calculate the matching score makes it easier to clarify the prompt instructions regarding which researcher information and which organization information to compare to calculate the matching score. It is acceptable for some information in the standard format to be missing, but it may be configured so that it is a required input or that researchers are prompted to input it. The standard format should be one that makes the prompt instructions easy to understand, and is not particularly limited. Such standardization may be performed by a prompt generation program, other programs, large-scale language models, etc.

[0077] The science business matching device 100 extracts information on the researcher's field of specialization from the input researcher's information and information on the technology field from the business entity's information (S303). If the information is in a standard format, it becomes easy to extract only the necessary information.

[0078] Information on multiple relevant matching criteria is obtained from the researchers' information and the information on the organizations (S303).

[0079] Multiple perspectives may be pre-stored or generated. These perspectives may include evaluation axes (e.g., degree of agreement between specialized technical information and technical field information, degree of area agreement, degree of experience agreement), weights for each axis (e.g., axis A: 60%, axis B: 30%, axis C: 10%), score ranges and definitions (e.g., 0-100 points, 80 points or higher indicates a high matching score), and output formats. For example, the weights for each axis may have default values, and if information regarding priorities exists in the researcher or organization's information, these may be changed according to these priorities. Furthermore, the weighting may be adjusted based on factors such as the amount of text and the order of presentation.

[0080] From multiple perspectives, determine one or more perspectives to be used for the prompt (S305). The perspectives may include common perspectives that are always used.

[0081] A prompt is generated that includes information on the researcher's field of specialization, information on the business entity's technology field, and criteria for calculating the matching score, and this prompt is input into a large-scale language model (S306). The prompt may be generated by a proprietary program, or it may be generated using the same or a different large-scale language model, or a finely tuned version of the large-scale language model may be used.

[0082] A large-scale language model calculates the matching score (S307).

[0083] The processing unit 30 extracts the business entity with the highest matching score in technical field information (S308).

[0084] The processing unit 30 creates a matching score and display data for displaying information about the extracted business entities (S309).

[0085] The processing unit 30 transmits the display data to the researcher's terminal (S310).

[0086] This operation flow is the same as in Figure 1, except that steps equivalent to S302-307 are performed instead of S102.

[0087] Furthermore, the operation flow when a business entity searches for researchers using a science business matching device equipped with a prompt generation program is the same as in Figure 2, except that steps S302-307 are performed instead of S202, and therefore its explanation is omitted.

[0088] The displayed data may include the matching score and information about researchers or organizations with high matching scores (name or company name, researcher's age and workplace, field of specialization or technical field, patent and publication information, etc.). Furthermore, it may also include text explaining the reasons for the matching.

[0089] Furthermore, the displayed data may be refined by combining it with keyword searches or other methods performed by researchers or organizations. This allows for the acquisition of highly convenient displayed data.

[0090] The displayed data may include links to online meeting and / or messaging functions (for collaboration or job offers) between researchers and organizations, and such messages may be sent. This makes it easier to communicate with partners who have high matching scores. Furthermore, the messaging function may be configured to allow users to register multiple pre-written templates, generate appropriate messages using a large-scale language model, or send templates with just a click, making it easy to send contact messages. Similarly, the system may be configured to allow users to register pre-written templates for replies to sent messages.

[0091] Furthermore, the system could be made more user-friendly by allowing users to see the status of the current interaction (under review, interview request, accepted / rejected, etc.) at a glance. Additionally, if a collaboration is unsuccessful, the system could be configured to have the AI ​​generate feedback, thereby increasing the success rate of future collaborations.

[0092] Furthermore, data, documents, and resume information related to contracts, such as contract templates, may be stored in database 10 and researcher information database 70, or an AI may be provided to perform these tasks, thereby supporting the application and contract process with simple operations.

[0093] The displayed data may also include other related information, such as researcher information (e.g., current workplace address) and business entity information (e.g., headquarters or research institute address, capital, number of employees, website information), and this data may also be stored in database 10 and researcher information database 70.

[0094] The processing unit 30 can be configured to notify researchers and organizations that have already entered (registered) information about organizations and researchers whose matching score exceeds a predetermined threshold, including those for which new information has been registered (including those for which additional information has been registered). Depending on the registrant's preference, this notification may be sent immediately after registration, periodically (e.g., once a month), or at any time requested. In this way, contact with organizations and researchers (those with high matching scores) based on information that did not exist at the time of matching is facilitated.

[0095] Furthermore, the science business matching device of the present invention may be used to match researchers with each other. In this case, researchers only need to input information such as their desired field of specialization to the person they wish to collaborate with, and the process can be carried out in the same way as when a business uses the matching service.

[0096] Furthermore, the science business matching device of the present invention may be used to match businesses with each other. In this case, it is sufficient to input desired technical field information, etc., to the partner with whom collaboration is desired, and the process should be the same as, for example, when researchers use the matching service. [Industrial applicability]

[0097] This invention enables efficient and low-cost matching of businesses and researchers, thereby accelerating the social implementation of research results and facilitating recruitment activities. Therefore, it serves as an important means for both businesses and researchers to expand collaboration opportunities and rapidly advance the commercialization and social implementation of research results. [Explanation of Symbols]

[0098] 100 Science Business Matching Device 10 Databases 20 Memory section 30 Processing Unit 40 Input section 50 Display Data Creation Section 60 Transmitter 70 Researcher Information Database 80 Information acquisition department 90 Contact area 200 researcher terminals 300 business entity terminals 400 External Databases

Claims

1. A science business matching method performed by a science business matching device, To obtain information on researchers' desire for collaboration, The aforementioned collaboration request information is compared with the technical information of the business entity stored in the database, and a machine learning model is used to calculate a matching score. The system includes notifying the researchers of information on organizations with high matching scores, The aforementioned machine learning model is a natural language processing model. The aforementioned machine learning model is trained using information on researchers' areas of specialization and information on the technology fields of businesses as input data, and matching scores between the area of ​​specialization information and the technology fields determined by experts as ground truth data, in a science business matching method.

2. A science business matching method performed by a science business matching device, To obtain information on areas of collaboration sought by the business entity, The aforementioned collaborative field information is compared with the researcher's field of specialization information stored in the researcher information database, and a machine learning model is used to calculate a matching score. The system includes notifying the business entity of information on researchers with high matching scores, The aforementioned machine learning model is a natural language processing model. The aforementioned machine learning model is trained using information on researchers' areas of specialization and information on the technology fields of businesses as input data, and matching scores between the area of ​​specialization information and the technology fields determined by experts as ground truth data, in a science business matching method.

3. The aforementioned machine learning model, The characteristic quantities of the researcher's field of specialization determined based on the aforementioned field of specialization information, The science business matching method according to claim 1, wherein the matching score is calculated based on the degree of agreement with any of the feature quantities of multiple technological fields possessed by the business entity.

4. The aforementioned machine learning model, The characteristic quantities of the researcher's field of specialization determined based on the aforementioned field of specialization information, The science business matching method according to claim 2, wherein the matching score is calculated based on the degree of agreement with any of the feature quantities of multiple technological fields possessed by the business entity.

5. A science business matching method performed by a science business matching device equipped with a storage unit that stores a plurality of matching perspective information, To obtain information on researchers' areas of specialization, The aforementioned field of specialization information is compared with the technical information of the business entity stored in the database, and a large-scale language model is used to calculate a matching score. The system includes notifying the researchers of information on organizations with high matching scores, The prompts input to the aforementioned large-scale language model are generated by a prompt generation program. The prompt generation program, The steps include obtaining information on the aforementioned researchers and information on the aforementioned business entity, The steps include extracting information on the researcher's field of specialization from the researcher's information and information on the technical field from the business entity's information, The steps include obtaining multiple relevant matching perspective information from the storage unit based on the information of the researcher and the information of the business entity, The steps include selecting one or more of the aforementioned multiple matching perspective information to determine the perspective for calculating the matching score, A science business matching method that performs the step of generating a prompt including the aforementioned field of specialization information, the aforementioned technical field information, and the aforementioned perspective.

6. A science business matching method performed by a science business matching device, To obtain information on researchers' areas of specialization, The aforementioned field of specialization information is compared with the technical information of the business entity stored in the database, and a large-scale language model is used to calculate a matching score. The system includes notifying the researchers of information on organizations with high matching scores, The prompts input to the aforementioned large-scale language model are generated by a prompt generation program. The prompt generation program, The steps include obtaining information on the aforementioned researchers and information on the aforementioned business entity, The steps include extracting information on the researcher's field of specialization from the researcher's information and information on the technical field from the business entity's information, The steps include generating information on multiple matching perspectives from the information of the aforementioned researchers and the information of the aforementioned business entity, The steps include selecting one or more of the aforementioned multiple matching perspective information to determine the perspective for calculating the matching score, A science business matching method that performs the step of generating a prompt including the aforementioned field of specialization information, the aforementioned technical field information, and the aforementioned perspective.

7. To obtain information on researchers' desire for collaboration and / or desired conditions, The matching score is calculated by comparing the aforementioned collaboration request information and / or the aforementioned desired conditions information with the technical information of the business entity stored in the database. The science business matching method according to claim 5 or 6, further comprising notifying the researcher of information on business entities with high matching scores.

8. To obtain research personnel information and / or collaborative field information requested by the business entity, The process involves comparing the aforementioned research personnel information and / or collaborative field information with the researcher's field of specialization information stored in the researcher information database to calculate a matching score. The science business matching method according to claim 5 or 6, further comprising notifying the business entity of information on researchers with high matching scores.

9. The aforementioned specialization information includes at least one of the researcher's current and / or past affiliations, specialization, years of experience, years of service, titles and / or abstracts of publications and / or patent documents, and information on conference presentations. The science business matching method according to any one of claims 1 to 6, wherein the technical information includes at least one of the business field of the business entity, the title and / or abstract of a paper and / or patent document, and the specialization information of the desired researcher.

10. A computer program that causes a science business matching device to execute the science business matching method described in any one of claims 1 to 6.

11. An input section that accepts input from researchers, A database containing technical information of business entities, A storage unit for storing the computer program described in claim 10, A processing unit that executes the aforementioned computer program to calculate a matching score, A display data creation unit that creates display data including information on researchers or organizations with high matching scores, A science business matching device comprising a transmission unit that transmits the aforementioned display data to the terminal of the researcher or the business entity.

12. The science business matching apparatus according to claim 11, wherein the display data production unit produces the display data including the matching score for business entities with a high matching score that match the collaboration request information and / or desired conditions information entered by the researcher.

13. The aforementioned input unit further accepts input from the business entity, Furthermore, it includes a researcher information database containing entered information on researchers and their fields of specialization, The science business matching device according to claim 11, wherein the display data production unit produces the display data including the matching score for researchers with a high matching score that match the research personnel information and / or collaborative field information entered by the business entity.

14. The science business matching apparatus according to claim 11, wherein the display data production unit produces display data that includes information that was the deciding factor in determining the matching score and the matching score between the researcher and the business entity.

15. It further includes an information retrieval unit that accesses external databases and obtains publications and patent information about registered entities and / or researchers. The science business matching apparatus according to claim 11, which calculates the matching score using the papers and patent information obtained by the information acquisition unit.

16. The science business matching device according to claim 11, further comprising a contact unit that performs online meeting and / or messaging functions between a business entity and a researcher.

17. The science business matching device according to claim 12, wherein the processing unit notifies researchers who have already entered information of businesses whose matching scores exceed a predetermined threshold among the newly registered businesses.

18. The processing unit notifies the business entity that has already entered information of researchers whose matching score exceeds a predetermined threshold among the newly registered researchers, as described in claim 12.

Images