Information processing systems, information processing methods, and programs

Abstract

To enable matching with experts. [Solution] An information processing system comprising: a known information storage unit for storing known information; a provider storage unit for storing information identifying the provider that provided the known information; a search unit for searching the known information storage unit for known information related to a question received from a questioner; and a presentation unit for presenting the provider that provided the retrieved known information to the questioner.

Description

Technical Field

[0001] The present invention relates to an information processing system, an information processing method, and a program.

Background Art

[0002] Patent Document 1 discloses a system that matches an expert suitable for the needs of a prospective customer.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] It is difficult to determine which experts are suitable for a questioner who has asked a question.

[0005] The present invention has been made in view of such a background, and an object thereof is to provide a technology capable of matching experts.

Means for Solving the Problems

[0008] According to the present invention, it is possible to match experts with users. [Brief explanation of the drawing]

[0009] [Figure 1] This figure shows an example of the overall configuration of an information processing system. [Figure 2] This figure shows an example of the hardware configuration of management server 2. [Figure 3] This figure shows an example of the software configuration for management server 2. [Figure 4] This diagram illustrates the operation of management server 2. [Modes for carrying out the invention]

[0010] <System Overview> The following describes an information processing system according to one embodiment of the present invention. The information processing system of this embodiment aims to match a user (questioner) with an expert. In the information processing system of this embodiment, answers to questions are created using the so-called RAG method, and experts who have previously provided the knowledge (hereinafter referred to as known information) used in the RAG answer generation are matched with the questioner. In addition, an incentive is given to the provider of the knowledge used in the RAG answer generation. Furthermore, when a match is made (the questioner requests work from the expert), that expert is given an incentive.

[0011] Figure 1 shows an example of the overall configuration of an information processing system. The information processing system in this embodiment includes a management server 2. The management server 2 is connected to the user terminal 1 via a communication network. The communication network is, for example, the internet and is constructed using public telephone networks, mobile phone networks, wireless communication channels, Ethernet (registered trademark), etc.

[0012] User terminal 1 is a computer operated by the user. User terminal 1 can be, for example, a smartphone, a tablet computer, or a personal computer.

[0013] The management server 2 may be a general-purpose computer such as a workstation or personal computer, or it may be logically implemented through cloud computing.

[0014] <Management Server> Figure 2 shows an example of the hardware configuration of the management server 2. Note that the illustrated configuration is just one example, and other configurations are also possible. The management server 2 includes a CPU 201, memory 202, storage device 203, communication interface 204, input device 205, and output device 206. The storage device 203 stores various data and programs, such as a hard disk drive, solid-state drive, or flash memory. The communication interface 204 is an interface for connecting to a communication network, such as an adapter for connecting to Ethernet®, a modem for connecting to a public telephone network, a wireless communication device for wireless communication, or a USB (Universal Serial Bus) connector or RS232C connector for serial communication. The input device 205 is for inputting data, such as a keyboard, mouse, touch panel, button, or microphone. The output device 206 is for outputting data, such as a display, printer, or speaker. Furthermore, each functional unit of the management server 2, as described later, is realized by the CPU 201 reading programs stored in the storage device 203 into memory 202 and executing them, and each storage unit of the management server 2 is realized as part of the storage area provided by memory 202 and storage device 203.

[0015] Figure 3 shows an example of the software configuration of the management server 2. The management server 2 includes a known information storage unit 231, a provider storage unit 232, a question receiving unit 211, a search unit 212, a generation unit 213, an output unit 214, a presentation unit 215, an evaluation decision unit 216, an incentive granting unit 217, and a request receiving unit 218.

[0016] <Storage section> The known information storage unit 231 stores various types of knowledge as known information. Known information may be, for example, PDF files, word processor files, or text files. Image data and audio data can also be used as known information. The known information storage unit 231 can also be implemented, for example, by a file system.

[0017] In this embodiment, the known information storage unit 231 is assumed to consist of a file system that stores known information as files and a vector database that stores vector data obtained by embedding (part or all) of the contents of the files. By searching the vector database, it becomes possible to efficiently search for known information that has similar content to the search target.

[0018] The provider memory unit 232 stores information about the provider who provided the known information (hereinafter referred to as provider information). The provider information may include information for identifying the provider (e.g., provider ID), the name of the provider (such as a surname and name), and the contact information of the provider (such as an email address, a chat account, a phone number, etc.). The information for identifying the provider may be, for example, the user ID of the user terminal 1 or the identification information of the user terminal 1. In this embodiment, the information for identifying the provider is the user ID. The user ID is, for example, a character string used when the user of the user terminal 1 logs in to this system. Alternatively, the user ID may be the email address of the user of the user terminal 1. Also, the information for identifying the provider may be the identification information of the user terminal 1, and for example, the MAC address, IMEI (International Mobile Equipment Identifier), IP address, etc. of the user terminal 1 may be used.

[0019] <Functional unit> The question reception unit 211 receives questions from questioners.

[0020] The search unit 212 searches the known information storage unit 231 for known information related to the questions received from the questioners.

[0021] Whether it is related to the question can be determined, for example, by whether the distance between the vector data obtained by embedding the question and the vector data stored in the known information storage unit 231 is close. For example, the question can be tokenized using a morphological analyzer, and each token can be converted into a feature vector. For the conversion into a feature vector, distributed representation learning algorithms such as word2vec or GloVe can be used. Using the feature vectors obtained from the question, a question vector is generated. The question vector can be obtained by averaging the feature vectors of each token included in the question. Alternatively, a weighted average may be calculated according to the importance of the tokens. The search unit 212 also vectorizes the known information in the same manner. That is, after tokenizing by a morphological analyzer and converting each token into a feature vector, a known information vector is generated.

[0022] The relevance between the question vector and the known information vector can be determined by calculating the cosine similarity between these vectors. The cosine similarity takes a value from -1 to 1, and the closer it is to 1, the higher the similarity. In this embodiment, when the cosine similarity between the question vector and the known information vector is greater than or equal to a predetermined threshold (for example, 0.7), they are regarded as relevant. When multiple pieces of known information exceed the threshold, a predetermined number (for example, 5 pieces) can be selected as relevant in the order of decreasing cosine similarity.

[0023] The generation unit 213 generates an answer to the question. The generation unit 213 can create an answer by so-called RAG (Retrieval-Augmented Generation). The generation unit 213 can generate an answer by giving a prompt including the known information retrieved by the retrieval unit 212, the question, and an instruction to create an answer to the question based on the known information to the LLM.

[0024] The output unit 214 outputs the answer generated by the generation unit 213. The output unit 214 can send the answer to the user terminal 1 of the questioner who asked the question.

[0025] The presentation unit 215 presents an expert who matches the question to the questioner. In this embodiment, the presentation unit 215 presents the provider who provided the retrieved known information to the questioner. The provider of the known information can be obtained from the provider storage unit 232. The presentation unit 215 identifies the known information used for generating the answer among the retrieved known information, and can obtain the provider of the identified known information from the provider storage unit 232.

[0026] The incentive granting unit 217 grants an incentive to the provider of the known information. The incentive granting unit 217 can grant an incentive to the provider presented to the questioner.

[0027] The incentive provision unit 217 can provide various forms of incentives to providers of known information. For example, one possible incentive is monetary compensation. Monetary compensation may be paid directly to the provider of known information's bank account or credit card, or it may be provided in the form of electronic money or cryptocurrency.

[0028] Another possible incentive is the awarding of points. These points may be units with their own unique value usable within this system, or they may be points usable in other partner services. Providers can use the awarded points within this system or in partner services.

[0029] Another possible incentive is to award providers with ratings and titles. For example, providers who meet certain criteria could be given titles such as "Gold Member" or "Expert," which could then be displayed on their profiles. This would improve the providers' visibility and trustworthiness.

[0030] Furthermore, the methods for providing incentives are not limited to the examples mentioned above; various methods can be adopted to improve the motivation of providers and encourage the provision of better information.

[0031] The request receiving unit 218 receives job requests from questioners to providers. The request receiving unit 218 can accept the selection of one or more providers from those presented by the presentation unit 215. The request receiving unit 218 can receive the questioner's designation of a provider and the details of the request. The details of the request may include, for example, a detailed description of the work, the desired deadline, and the budget. The detailed description of the work may include, for example, the circumstances behind the question, the challenges the questioner is facing, and the specific deliverables the questioner is requesting from the provider. The request receiving unit 218 can notify the designated provider of the information identifying the questioner and the details of the request.

[0032] After notifying the provider of the job request, the Request Reception Unit 218 can receive a response from the provider indicating acceptance or rejection of the request. If the provider accepts the request, the Request Reception Unit 218 can receive an estimate of the price and delivery date from the provider. Subsequently, the Request Reception Unit 218 can present the presented estimate of the price and delivery date to the questioner and receive the questioner's approval. If the questioner approves, the Request Reception Unit 218 can treat this as a concluded agreement between the questioner and the provider.

[0033] The incentive distribution unit 217 can provide an incentive to the provider in response to the occurrence of a job request. If the job request includes a requested amount, the incentive distribution unit 217 may also provide an incentive in accordance with the requested amount.

[0034] The evaluation determination unit 216 can determine an evaluation value for the retrieved known information. The evaluation determination unit 216 can, for example, receive an evaluation value for the known information from an evaluator. The evaluator may be the questioner, or another user who has viewed the question and the known information. Alternatively, an expert other than the provider of the known information may be the evaluator. The evaluation determination unit 216 can calculate a statistical value (e.g., a total value) of the evaluation value for the retrieved known information for each provider, and use the calculated statistical value as the provider's evaluation value.

[0035] The evaluation determination unit 216 can determine the provider's evaluation value according to the number of known pieces of information retrieved that were provided by the provider.

[0036] The evaluation determination unit 216 can determine an evaluation value based on the number of times known information has been used to generate an answer in the past. In this case, a generation history storage unit can be provided to manage the history of information used to generate answers. The generation history may include questions, answers, and the known information used in those answers. The evaluation determination unit 216 can calculate a statistical value of the number of times the information has been used for generation for each provider of known information, and use the calculated statistical value as the provider's evaluation value.

[0037] The incentive granting unit 217 can grant incentives to providers of known information in accordance with its evaluation of the known information and / or the provider.

[0038] <Operation> Figure 4 is a diagram illustrating the operation of the management server 2.

[0039] Management Server 2 receives a question (S301), searches for known information related to the question (S302), provides the retrieved known information to LLM to generate an answer to the question (S303), presents the provider of the known information to the questioner (S304), and provides an incentive to the provider of the known information (S305). When Management Server 2 receives a job request from the questioner to a provider (S306), it notifies the provider of the job request (S307) and provides an incentive to the provider to whom the request was made (S308).

[0040] As described above, the information processing system of this embodiment makes it possible to match the provider of known information used in generating answers by RAG with the questioner.

[0041] Although these embodiments have been described above, they are intended to facilitate understanding of the present invention and are not intended to limit its interpretation. The present invention can be modified and improved without departing from its spirit, and equivalents thereof are also included.

[0042] For example, the processing performed by each functional unit of the management server 2 described above may be executed by any of the functional units. Furthermore, different functional units may be added to perform some of the processing performed by each of the functional units described above. Also, the functional units of the management server 2 may be distributed across multiple computers.

[0043] Furthermore, the information stored in each memory unit of the management server 2 may be stored in any of the memory units. That is, the information stored in the multiple memory units mentioned above may be stored in a single memory unit, or a portion of the information stored in one memory unit may be stored in another memory unit.

[0044] <Example 1> In the above embodiment, the search unit 212 selected the most relevant piece of known information from among the multiple pieces of known information it retrieved and generated an answer using only that piece of known information. However, it is also possible to generate a more appropriate and complete answer by combining multiple pieces of known information related to the question.

[0045] For example, the search unit 212 can select multiple known pieces of information whose relevance to the question is above a predetermined threshold. The generation unit 213 can then combine the selected known pieces of information to generate an answer to the question.

[0046] Specifically, the generation unit 213 can generate an answer by extracting the information necessary to answer the question from multiple selected known pieces of information, and by organizing and integrating the extracted information. For example, if the question is "Please tell me about the application fields of artificial intelligence," the search unit 212 can select known pieces of information such as "Application of artificial intelligence in the medical field," "Application of artificial intelligence in the financial field," and "Application of artificial intelligence in the manufacturing industry." From this known information, the generation unit 213 can extract examples of artificial intelligence applications in medicine, finance, and manufacturing, and generate an answer categorized by field.

[0047] Furthermore, the generation unit 213 can also generate a response that reconciles contradictions among the selected known information. For example, if one piece of known information states that "artificial intelligence will surpass human capabilities within 10 years," and another piece of known information states that "it will still take more than 50 years for artificial intelligence to surpass human capabilities," the generation unit 213 can generate a consistent response that takes both claims into account, such as, "There are various opinions on the rate of development of artificial intelligence; some predict it will surpass human capabilities within 10 years, while others predict it will take more than 50 years. Technological progress is difficult to predict, but there is no doubt that it is steadily developing."

[0048] <Modification 2> Furthermore, the presentation unit 215 may select which providers to present to the questioner based on the provider's evaluation value. For example, the presentation unit 215 may present only providers whose evaluation value, determined by the evaluation determination unit 216, is above a predetermined threshold to the questioner. This makes it possible to limit the providers presented to the questioner to those who can provide high-quality information.

[0049] Furthermore, the presentation unit 215 may present a predetermined number of providers to the questioner in order of their evaluation scores. For example, the presentation unit 215 can select 10 providers in order of their evaluation scores and present the selected providers to the questioner. This allows for limiting the number of providers presented to the questioner while prioritizing the matching of high-quality providers.

[0050] Furthermore, the presentation unit 215 may change the threshold value of the provider's evaluation depending on the content of the question. For example, the presentation unit 215 can determine the difficulty level of a question based on keywords included in the question's content. The presentation unit 215 can then raise the threshold value of the evaluation when the question is difficult, and lower the threshold value of the evaluation when the question is easy. This allows the questioner to be matched with a provider of an appropriate evaluation value according to the difficulty level of the question.

[0051] Furthermore, the presentation unit 215 may change the threshold value of the provider's evaluation value depending on the attributes of the questioner. For example, the presentation unit 215 can determine the questioner's expertise based on the questioner's occupation, age, past question history, etc. Then, the presentation unit 215 can raise the threshold value of the evaluation value if the questioner's expertise is high, and lower the threshold value of the evaluation value if the questioner's expertise is low. This makes it possible to match the questioner with a provider with an appropriate evaluation value according to the questioner's expertise.

[0052] <Variation 3> Alternatively, machine learning models may be used to match questions with known information.

[0053] For example, the search unit 212 uses a machine learning model 250 that takes a question as input and outputs the degree of relevance to each piece of known information to search for known information related to the question. The machine learning model 250 can be constructed, for example, using a neural network.

[0054] The machine learning model 250 can be trained using a dataset of question-and-known information pairs. The dataset can include past questions and pairs of known information selected by the provider as relevant information for those questions. Alternatively, known information used to generate answers to questions may also be used as relevant information for training.

[0055] The input to machine learning model 250 is a vector representation of a question-and-known information pair. Both the question and the known information can be vectorized, as described above. A vector formed by combining the question vector and the known information vector can be used as input.

[0056] The output of the machine learning model 250 can be a score representing the degree of relevance between the input question and the known information. The score may be a real number in the range of 0 to 1, for example.

[0057] The search unit 212 sequentially provides pairs of question vectors and vectors of each known piece of information as input to the machine learning model 250, and can select a predetermined number of highly relevant known pieces of information as search results based on the output score.

[0058] Various neural network architectures can be used to construct the 250 machine learning models, such as CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), LSTM (Long Short-Term Memory), and Transformer. Alternatively, pre-trained language models such as BERT can be fine-tuned and used.

[0059] <Modification 4> In this embodiment, when the provider provides information to the questioner, the information may be exchanged via the management server 2. Specifically, the following configuration can be adopted.

[0060] The management server 2 includes an intermediary unit that mediates communication between the provider terminal and the questioner terminal. The intermediary unit transmits information received from the provider terminal to the questioner terminal and transmits information received from the questioner terminal to the provider terminal.

[0061] The provider terminal transmits the information the provider provides to the questioner to the intermediary unit of management server 2. The questioner terminal presents the information received from the provider via the intermediary unit of management server 2 to the questioner.

[0062] Furthermore, the questioner's terminal transmits the information the questioner wants to send to the provider to the mediation unit of management server 2. The provider's terminal presents the information received from the questioner by management server 2 to the provider.

[0063] Furthermore, the intermediary unit of the management server 2 may identify the provider's personal information contained in the information received from the provider's terminal, delete or anonymize the identified personal information, and then send it to the questioner's terminal.

[0064] Furthermore, the intermediary unit of the management server 2 may identify the questioner's personal information contained in the information received from the questioner's terminal, delete or anonymize the identified personal information, and then send it to the provider's terminal.

[0065] Furthermore, the mediation unit of the management server 2 may be equipped with a recording unit that records the content of communication between the provider terminal and the questioner terminal.

[0066] <Modification 5> In this embodiment of the information processing system, a function can be added to analyze the history of questions and answers and present frequently asked questions and their answers.

[0067] Specifically, the management server 2 further comprises a question history storage unit, an answer history storage unit, a frequently asked questions extraction unit, and an answer presentation unit.

[0068] The question history storage unit stores data of questions received in the past. For example, the question history storage unit can store the text data of the question, the date and time the question was received, the user ID of the questioner, and other related information.

[0069] The answer history storage unit stores data on answers to past questions. The answer history storage unit can store, for example, the text data of the answer, the date and time the answer was generated, the ID of the known information used to generate the answer, and the questioner's evaluation of the answer, all associated with each other.

[0070] The frequently asked questions (FAQ) extraction unit analyzes the question data stored in the question history storage unit and extracts frequently asked questions. Frequently asked questions can be extracted, for example, by analyzing the frequency of keyword occurrences in the text data of the questions. Alternatively, frequently asked questions can be extracted by clustering the text data of the questions to group similar questions and selecting representative questions from each group.

[0071] The answer presentation unit presents the frequently asked questions extracted by the frequently asked questions extraction unit to the questioner, along with their corresponding answers. The answer presentation unit can also present the questioner with frequently asked questions and their answers if there are similar questions to the question received from the questioner.

[0072] Furthermore, the answer presentation section may display a list of frequently asked questions to the questioner before accepting their question. The questioner can easily refer to past answers by selecting one that is similar to their own question from the list.

[0073] <Disclosure Items> Furthermore, this disclosure also includes the following configurations. [Item 1] A known information storage unit that stores known information, A provider storage unit that stores information identifying the provider who provided the aforementioned known information, A search unit retrieves the aforementioned known information related to the question received from the questioner from the known information storage unit, A presentation unit that presents to the questioner the provider who provided the searched known information, An information processing system characterized by comprising the following features. [Item 2] The information processing system described in item 1, The system includes a generation unit that provides a large-scale language model with a prompt containing the aforementioned question and the retrieved known information to generate an answer to the question. An information processing system characterized by the following. [Item 3] The information processing system described in item 1, The system includes an incentive-granting unit that provides an incentive to the provider presented to the questioner. An information processing system characterized by the following. [Item 4] The information processing system described in item 1, The system includes a request reception department that receives job requests from the aforementioned questioner to the aforementioned provider. An information processing system characterized by the following. [Item 5] The information processing method described in item 4, An incentive granting unit that provides an incentive to the provider in response to the aforementioned request, An information processing method characterized by the following. [Item 6] The information processing method described in item 1, An evaluation determination unit that determines an evaluation of the provider of the searched known information, An incentive granting unit that provides an incentive to the provider in accordance with the evaluation, An information processing method characterized by comprising: [Item 7] The information processing system described in item 6, The evaluation determination unit receives the evaluation value for the known information from the evaluator as the evaluation value of the provider of the known information. An information processing system characterized by the following. [Item 8] The information processing system described in item 1, The evaluation determination unit determines the evaluation value according to the number of known pieces of information retrieved that were provided by the provider. An information processing system characterized by the following. [Item 9] The information processing system described in item 1, The system includes a generation unit that provides a large-scale language model with the aforementioned question and a prompt containing the retrieved known information to generate an answer to the question, The evaluation determination unit determines the evaluation value according to the number of times the known information has been used in the past to generate the answer. An information processing system characterized by the following. [Item 10] The steps include storing known information in the known information storage unit, A step of storing information that identifies the provider who provided the aforementioned known information, The steps include: searching the known information storage unit for the known information related to the question received from the questioner; The steps include presenting the provider who provided the searched known information to the questioner, An information processing method characterized by a computer executing the following. [Item 11] The steps include storing known information in the known information storage unit, A step of storing information that identifies the provider who provided the aforementioned known information, The steps include: searching the known information storage unit for the known information related to the question received from the questioner; The steps include presenting the provider who provided the searched known information to the questioner, A program that causes a computer to execute something. [Explanation of Symbols]

[0074] 1 User terminal 2 Management Server

Claims

1. A known information storage unit that stores known information, A provider storage unit that stores information identifying the provider who provided the aforementioned known information, A search unit retrieves the aforementioned known information related to the question received from the questioner from the known information storage unit, A presentation unit that presents to the questioner the provider who provided the searched known information, An information processing system characterized by comprising the following features.

2. The information processing system according to claim 1, The system includes a generation unit that provides a large-scale language model with a prompt containing the aforementioned question and the retrieved known information to generate an answer to the question. An information processing system characterized by the following.

3. The information processing system according to claim 1, The system includes an incentive-granting unit that provides an incentive to the provider presented to the questioner. An information processing system characterized by the following.

4. The information processing system according to claim 1, The system includes a request reception department that receives job requests from the aforementioned questioner to the aforementioned provider. An information processing system characterized by the following.

5. The information processing method according to claim 4, An incentive granting unit that provides an incentive to the provider in response to the aforementioned request, An information processing method characterized by the following.

6. The information processing method according to claim 1, An evaluation determination unit that determines an evaluation of the provider of the searched known information, An incentive granting unit that provides an incentive to the provider in accordance with the evaluation, An information processing method characterized by comprising:

7. The information processing system according to claim 6, The evaluation determination unit receives the evaluation value for the known information from the evaluator as the evaluation value of the provider of the known information. An information processing system characterized by the following.

8. The information processing system according to claim 1, The evaluation determination unit determines the evaluation value according to the number of known pieces of information retrieved that were provided by the provider. An information processing system characterized by the following.

9. The information processing system according to claim 1, The system includes a generation unit that provides a large-scale language model with the aforementioned question and a prompt containing the retrieved known information to generate an answer to the question, The evaluation determination unit determines the evaluation value according to the number of times the known information has been used in the past to generate the answer. An information processing system characterized by the following.

10. The steps include storing known information in the known information storage unit, A step of storing information that identifies the provider who provided the aforementioned known information, The steps include: searching the known information storage unit for the known information related to the question received from the questioner; The steps include presenting the provider who provided the searched known information to the questioner, An information processing method characterized by a computer executing the following.

11. The steps include storing known information in the known information storage unit, A step of storing information that identifies the provider who provided the aforementioned known information, The steps include: searching the known information storage unit for the known information related to the question received from the questioner; The steps include presenting the provider who provided the searched known information to the questioner, A program that causes a computer to execute something.

Images