Program and Information Processing Device

Abstract

The objective is to provide a program and information processing device that efficiently support learning specifically for qualification exams where past exam questions are highly valued. [Solution] The system includes a past question data acquisition unit that acquires past question data containing textual information about past questions asked in qualification exams, a necessary knowledge data generation unit that generates necessary knowledge data which is textual information about the knowledge needed to answer the questions in the acquired past question data, and an important knowledge data generation unit that generates important knowledge data which is textual information about common knowledge from a plurality of necessary knowledge data generated from each of a plurality of past question data, wherein at least one of the necessary knowledge data generation unit or the important knowledge data generation unit is configured to generate information based on a pre-trained model that has been trained in advance by inputting at least knowledge data which is textual information containing the knowledge necessary to pass the qualification exam.

Description

Technical Field

[0001] The present invention relates to a program and an information processing apparatus for supporting learning of qualification examinations.

Background Art

[0002] There is known an information processing apparatus including: a data acquisition unit that acquires data including character information regarding a problem; and a necessary knowledge data generation unit that generates necessary knowledge data which is character information regarding knowledge for answering the problem acquired by the data acquisition unit, based on a learned model that has been learned in advance by inputting a large number of knowledge data which is character information including various kinds of knowledge (for example, refer to Patent Document 1).

[0003] According to the information processing apparatus of the above document, knowledge data (in the above document, "explanation" regarding the answer to the problem) for answering individual problems is generated, so that it becomes possible to support efficient learning. On the other hand, since the information processing apparatus of the above document is learning support focused on knowledge in general such as learning of students, for example, when learning specialized in past questions such as qualification examinations is required, the support is insufficient.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] An object of the present invention is to provide a program and an information processing apparatus that efficiently support learning specialized in qualification examinations that place importance on past questions.

Means for Solving the Problems

[0006] To solve the above problems, the present invention provides a computer with the following features: a past question data acquisition function that acquires past question data containing text information relating to past questions asked in a target qualification examination; a necessary knowledge data generation function that generates necessary knowledge data, which is text information relating to the knowledge needed to answer the questions in the past question data acquired by the past question data acquisition function; and an important knowledge data generation function that generates important knowledge data, which is text information relating to common knowledge, from a plurality of necessary knowledge data generated by the necessary knowledge data generation function from each of the plurality of past question data; wherein at least one of the necessary knowledge data generation function or the important knowledge data generation function is configured to generate information based on a pre-trained model that has been trained in advance by inputting at least knowledge data, which is text information containing the knowledge necessary to pass the qualification examination.

[0007] On the other hand, the information processing device of the present invention comprises: a past question data acquisition unit that acquires past question data including textual information relating to past questions that have been asked in a target qualification examination; a necessary knowledge data generation unit that generates necessary knowledge data, which is textual information relating to the knowledge required to answer the questions in the past question data acquired by the past question data acquisition unit; and an important knowledge data generation unit that generates important knowledge data, which is textual information relating to common knowledge, from a plurality of necessary knowledge data generated by the necessary knowledge data generation unit from each of the plurality of past question data, wherein at least one of the necessary knowledge data generation unit or the important knowledge data generation unit is configured to generate information based on a pre-trained model that has been trained in advance by inputting at least knowledge data, which is textual information containing the knowledge necessary to pass the qualification examination.

[0008] The aforementioned important knowledge data generation unit may generate important knowledge data based on a pre-trained model that has been trained in advance by inputting the knowledge data and the past exam question data.

[0009] The aforementioned important knowledge data generation unit may generate the important knowledge data from a plurality of the necessary knowledge data and the past exam question data.

[0010] The aforementioned trained model may be an LLM trained using a neural network.

[0011] The system may also include at least one of the following: a necessary knowledge data providing unit that provides necessary knowledge data generated by the necessary knowledge data generation unit, and a key knowledge data providing unit that provides key knowledge data generated by the key knowledge data generation unit.

[0012] The aforementioned past exam data may include textual information about past questions that have appeared in the target qualification exam, as well as information about the correct answers to those questions.

[0013] The system may also include an important knowledge confirmation question generation unit that generates important knowledge confirmation questions, which are textual information related to the questions, based on important knowledge data generated by the important knowledge data generation unit based on a pre-trained model that has been trained in advance by inputting knowledge data, which are textual information containing the knowledge necessary to pass the aforementioned qualification examination, and the aforementioned past exam question data.

[0014] The aforementioned important knowledge confirmation question generation unit may generate the important knowledge confirmation questions based on the important knowledge data and the past question data.

[0015] The system may also include a key knowledge confirmation question providing unit that provides key knowledge confirmation questions generated by the aforementioned key knowledge confirmation question generation unit.

[0016] A storage device that stores the acquired / generated information, user information which is information of one or more users, and progress data which is information regarding the progress of each user's learning; a response data acquisition unit that acquires the response of the user; and a progress status confirmation unit that generates or updates the progress data of the user who made the response related to the response data based on the response data acquired by the response data acquisition unit and stores it in the storage device. The important knowledge confirmation question providing unit may be configured to provide the important knowledge confirmation question to the user.

[0017] The progress data prepared for each user may include the response status of that user sorted for each important knowledge confirmation question or for each related field.

[0018] It may be provided with a progress status providing unit that provides the progress data to the user related thereto.

Effect of the Invention

[0019] According to the configuration of the present invention, it becomes possible to perform efficient learning support specialized for qualification examinations that emphasize past questions.

Brief Description of the Drawings

[0020] [Figure 1] It is a schematic diagram of an information processing system to which the present invention is applied. <7000082>It is a block diagram showing the configuration of an information processing server. [Figure 3] It is an explanatory diagram showing the concept of a learned model for generating necessary knowledge data. [Figure 4] It is an explanatory diagram showing the concept of a learned model for generating important knowledge data. [Figure 5] It is an explanatory diagram showing the concept of a learned model for generating important knowledge confirmation questions. [Figure 6] It is a block diagram showing the configuration of a user terminal. [Figure 7]It is a table showing an example of the answer status of each user organized by field. [Figure 8] It is a flowchart showing a series of processing procedures related to the generation of necessary knowledge data and important knowledge data.

Mode for Carrying Out the Invention

[0021] FIG. 1 is a schematic diagram of an information processing system to which the present invention is applied. The illustrated information processing system includes an information processing server (information processing device) 10 and user terminals 20, 20 which are information terminals of a plurality of users. Note that the number of user terminals 20 is not limited to the illustrated example (specifically, two). Here, the user mainly refers to a person who studies in order to pass a predetermined qualification examination.

[0022] The information processing server 10 and the plurality of user terminals 20, 20 are configured to be able to communicate with each other via a global network 30 such as the Internet. The information processing server 10 is configured to be able to provide various data generated or extracted or the like by information processing described later to these user terminals 20, 20 via the network 30. Note that various information processed by this information processing system may be stored in the storage device 12 (see FIG. 2) of the information processing server 10 itself, but a part or all of it may be stored in the storage device (not shown) of an external server 40 which is another server configured to be able to communicate with the information processing server 10 via the network 30.

[0023] Also, it is not essential that the information processing server 10 is configured to be able to communicate with the user terminal 20. When it is configured independently from the user terminal 20, the information processing server 10 in this example is operated as the information processing device 10.

[0024] The information processing server 10 executes a past question data acquisition process which is a process of acquiring past question data including character information related to past questions presented in a target qualification examination.

[0025] The information processing server 10 executes a necessary knowledge data generation process, which generates necessary knowledge data, which is textual information relating to the knowledge needed to answer the questions in the past exam data acquired through the past exam data acquisition process.

[0026] The information processing server 10 repeatedly executes the past exam question data acquisition process and the necessary knowledge data generation process described above, and then executes the important knowledge data processing, which is the process of generating important knowledge data, which is textual information about common knowledge, from the multiple necessary knowledge data generated from each of the multiple past exam question data by the necessary knowledge data generation process.

[0027] The information processing server 10 is configured to perform at least one of the above-mentioned necessary knowledge data generation process or important knowledge data generation process based on a pre-trained model that has been trained by inputting at least knowledge data, which is textual information containing the knowledge necessary to pass the target qualification examination.

[0028] Next, the configuration of the information processing server and the pre-trained models used will be explained based on Figures 2 to 5.

[0029] Figure 2 is a block diagram showing the configuration of the information processing server, Figure 3 is an explanatory diagram illustrating the concept of a trained model for generating necessary knowledge data, Figure 4 is an explanatory diagram illustrating the concept of a trained model for extracting important knowledge data, and Figure 5 is an explanatory diagram illustrating the concept of a trained model for generating important knowledge confirmation questions.

[0030] The information processing server 10 is a computer comprising a control unit 11 composed of a CPU and RAM, etc., capable of implementing various functions (means) through program execution; a storage device 12 including RAM and storage such as an SSD or HDD; a wired or wireless communication interface 13 enabling communication via a network 30; an input interface 14 such as a keyboard, mouse, touch panel, or scanner for inputting information; and an output interface 15 such as a monitor, printer, or speaker for outputting information. The information processing server 10 may be a dedicated information processing device for this information processing system, but a general-purpose computer such as a personal computer may also be used as the information processing server 10.

[0031] The control unit 11 includes: a past question data acquisition unit 11a that executes the past question data acquisition process described above; a necessary knowledge data generation unit 11b that executes the necessary knowledge data generation process described above; a necessary knowledge data provision unit 11c that executes the necessary knowledge data provision process, which is the process of providing necessary knowledge data; a key knowledge data generation unit 11d that executes the key knowledge data generation process described above; a key knowledge data provision unit 11e that executes the key knowledge data provision process, which is the process of providing key knowledge data; a question generation unit 11f that executes the question generation process, which is the process of generating questions, which are textual information that tests understanding of predetermined knowledge data; a question provision unit 11g that executes the question provision process, which is the process of providing questions to the user; a user information acquisition unit 11h that executes the user information acquisition process, which is the information of one or more users; an answer data acquisition unit 11i that executes the answer data acquisition process, which is the process of acquiring the user's answer data to the questions; and the learning progress of the user who made the answer related to the answer data based on the acquired answer data. The system includes at least: a progress status confirmation unit 11j which performs a progress status confirmation process which generates or updates progress data, which is information about the status, and stores it in its own storage device 12 or the storage device of an external server 40; a progress status provision unit 11k which performs a progress status provision process which provides the progress data to the user concerned; a learning plan information generation unit 11l which performs a learning plan information generation process which generates learning plan information, which is information about the user's learning plan related to the progress data, based at least on the progress data; a learning plan information provision unit 11m which performs a learning plan information provision process which provides the learning plan information to the user; a question information acquisition unit 11n which performs a question information acquisition process which is information about questions from the user; a response information generation unit 11o which performs a response information generation process which generates response information, which is information about responses to the question information; and a response information provision unit 11p which performs a response information provision process which provides the response information to the user.

[0032] The past exam question data acquisition unit 11a is configured to acquire past exam question data by communication via the network 30 using the communication interface 13 during its past exam question data acquisition process. Specifically, the past exam question data acquisition process is performed by accessing a website or the like where the target past exam question data is published and acquiring the data, or by receiving the past exam question data from the user terminal 20. Furthermore, the manner of the past exam question data acquisition process is not limited to the above example, and may also be configured to acquire character information captured by information input via a keyboard or touch panel as past exam question data, or by capturing past exam questions as image data using a scanner or the like and extracting character information (specifically, the exam question text and answers, etc.) contained in the image data as past exam question data. In addition, the past exam question data acquisition unit 11a may perform the past exam question data acquisition process by acquiring past exam question data that has been previously stored in the storage device 12 of the information processing server 10 or the storage device of the external server 40.

[0033] The necessary knowledge data generation unit 11b generates the necessary knowledge data described above based on a large language model (LLM) that has been pre-trained using a neural network, by inputting knowledge data, which is textual information containing the knowledge necessary to pass the target qualification exam. Examples of large language models include ChatGPT-4o (registered trademark) by OpenAI (registered trademark), but the unit is not limited to this, and a pre-trained model that has been separately trained to be specific to the target qualification exam may also be used.

[0034] Specifically, the knowledge data includes textual information such as specialized terms, their meanings, and explanations found in textbooks or reference books for the relevant qualification exam. For example, in the case of a qualification exam related to medicine or radiation, the knowledge data would include textual information on specialized terms such as "sievert," "becquerel," "deterministic effects," and "stochastic effects," as well as their meanings.

[0035] As shown in Figure 3, the large-scale language model used by the necessary knowledge data generation unit 11b has the function of outputting necessary knowledge data using past exam data as at least an input parameter. Specifically, when a predetermined prompt (generation instruction) and input parameters are input, the large-scale language model weights the numerous knowledge data obtained through the learning described above to the extent necessary to answer the past exam data related to the input, and extracts and outputs the necessary knowledge data corresponding to the prompt. The neural network shown consists of a total of three layers: an input layer, a single hidden layer, and an output layer, but it is not limited to this example, and a configuration with multiple hidden layers is also possible. By increasing the number of hidden layers, more complex analysis can be performed on the input parameters, and it is expected that the necessary knowledge data can be output with higher accuracy. In addition, each layer has multiple nodes, but the number of these nodes is also not limited to the example shown.

[0036] Past exam data may include not only textual information about past questions that have appeared on the target qualification exam, but also information about the correct answers to those questions. This allows the trained model to input past exam data that includes the aforementioned correct answer information, which is expected to enable more accurate output of the necessary knowledge data.

[0037] Furthermore, the input parameters for the neural network's input layer must include at least past exam data, but other parameters (the +α parameters in Figure 3) may also be included. For example, data similar to past exam data can be included as input parameters, such as exam questions created by a test preparation school or other examination organization for the target qualification exam.

[0038] Furthermore, the necessary knowledge data generation unit 11b may also be configured to use RAG (Retrieval-Augmented Generation) in its generation process.

[0039] Here, "RAG" refers to a technique that improves the accuracy of the output by setting up a search database outside of a large-scale language model, searching for prompt information from the search database before inputting input parameters into the large-scale language model, and then inputting the search results data, prompts, and input parameters extracted through this search into the large-scale language model.

[0040] Specifically, when prompt and past exam question data are input, the necessary knowledge data generation unit 11b first performs a process to search for information about the prompt in the search database. Next, it performs a process to input the search result data, prompt and past exam question data obtained from the search process into the large-scale language model. As a result, even if the trends of the target qualification exam change or the information to be treated as knowledge data is updated, by adding the information related to the change or update to the search database, it is possible to obtain necessary knowledge data that reflects that information, thus eliminating the cost and effort of retraining the large-scale language model.

[0041] Furthermore, the RAG-based configuration described above can also be applied to the important knowledge data generation unit 11d, the problem generation unit 11f, and the response information generation unit 11o, which will be described later.

[0042] Furthermore, the necessary knowledge data generation unit 11b may be configured to perform necessary knowledge data generation preprocessing, which organizes past exam data as an ontology, prior to inputting input parameters to the large-scale language model, and then input that ontology.

[0043] Here, "ontology" refers to a data structure that represents a subject's knowledge (concept) in a hierarchical structure, along with related surrounding concepts, and describes their relationships in a tree-like manner. Specifically, for example, if the subject's concept is "polyp," then the hierarchical structure would describe "benign polyp" and "malignant polyp," which are sub-concepts of "polyp," below "polyp," and their relationship would be described as "benign polyps and malignant polyps are part of polyps (belong to polyps)." Furthermore, since "benign polyps" and "malignant polyps" have opposite meanings, their relationship would be described as "benign polyps and malignant polyps are antonyms."

[0044] In addition to ontologs that represent the meaning and relationships of concepts, functional ontologs may also be used. Here, a "functional ontology" is a data structure formed in a tree-like structure by breaking down the events necessary to achieve a given goal and organizing them according to causal relationships. When using a functional ontology, for example, the procedures for manufacturing a given product can be broken down and organized step by step. This makes it easy for large-scale language models to grasp how and in what situations the knowledge (concepts) in question are used.

[0045] In this way, by organizing the input data as an ontology before inputting it, the meaning of words and other elements included in the input data, as well as the relationships between those words, can be easily grasped by a large-scale language model, and thus the accuracy of the output information can be expected to improve. In other words, the necessary knowledge data generation unit 11b of this configuration inputs past exam data, which has been pre-organized as an ontology through necessary knowledge data generation preprocessing, into a large-scale language model, and therefore it can be expected to output the necessary knowledge data with high accuracy.

[0046] Furthermore, since the method of organizing information as an ontology is applicable to various types of input, it is not limited to the fields mentioned above. It is also possible to create an ontology for each field included in the target qualification exam and input it into a large-scale language model.

[0047] The trained model is stored in the memory of a server other than the information processing server 10, such as an external server 40 accessible via the network 30 through the communication interface 13. However, this configuration is not limited to this, and it may also be stored in the memory 12 of the information processing server 10. This storage method of the trained model is the same for the trained model used by the important knowledge data generation unit 11d and the important knowledge confirmation question generation unit 11f, which will be described later.

[0048] The necessary knowledge data provision unit 11c performs the necessary knowledge data provision process by either displaying or printing the necessary knowledge data, or by transmitting the necessary knowledge data to the user terminal 20 via the network 30 using the communication interface 13.

[0049] The important knowledge data generation unit 11d executes the important knowledge data generation process described above based on a large language model (LLM) that has been pre-trained using a neural network by inputting knowledge data, which is textual information containing the knowledge necessary to pass the target qualification exam, and past exam question data. The important knowledge data generation unit 11d is not limited to a configuration that generates important knowledge data from multiple necessary knowledge data; it can also generate important knowledge data from multiple necessary knowledge data and other data. For example, the important knowledge data generation unit 11d may generate important knowledge data from multiple necessary knowledge data and past exam question data.

[0050] As shown in Figure 4, the large-scale language model used by the important knowledge data generation unit 11d has the function of outputting important knowledge data using at least multiple necessary knowledge data as input parameters. Specifically, when a predetermined prompt (generation instruction) and input parameters are input, the large-scale language model weights the numerous knowledge data and past question data obtained through the learning process described above based on the degree of commonality of the multiple necessary knowledge data related to the input (for example, the degree to which they are commonly necessary when answering each of the numerous past question data learned), and extracts and outputs the important knowledge data corresponding to the prompt. In other words, the large-scale language model outputs knowledge data that is frequently used when answering a large number of questions as important knowledge data. The neural network shown in the figure consists of a total of three layers: an input layer, a single hidden layer, and an output layer, but it is not limited to this example, and a configuration with multiple hidden layers is also possible. By increasing the number of hidden layers, more complex analysis can be performed on the input parameters, and it is expected that important knowledge data can be output with higher accuracy. In addition, each layer has multiple nodes, but the number of these nodes is not limited to the example shown.

[0051] Furthermore, the input parameters for the input layer of the neural network must include at least several pieces of necessary knowledge data, but may also include parameters other than the multiple pieces of necessary knowledge data (the +α parameters in Figure 4). For example, the past exam data mentioned above, or question data created by examination preparation schools or other examination institutions for the target qualification exam, can be included as input parameters. When the large-scale language model used by the important knowledge data generation unit 11d takes multiple pieces of necessary knowledge data and past exam data as input parameters, it becomes possible to output, for example, important knowledge data that is highly related to the past exam data used as input.

[0052] Furthermore, the important knowledge data generation unit 11d is configured to perform important knowledge data generation processing by using the large-scale language model described above in combination with Word2vec and graph theory.

[0053] Here, "Word2vec" is a natural language processing technique that makes it possible to quantitatively understand the meaning of words by converting the words that make up the target (in this example, the textual information contained in past exam data, knowledge data, necessary knowledge data, and important knowledge data) into numerical vectors (semantic vectors) that represent the meaning of those words. For example, if the semantic vector obtained from one word is similar to the semantic vectors obtained from other words, it can be predicted that a strong relationship exists between these words.

[0054] Furthermore, "graph theory" is a method of analyzing data structures using graphs, which are models (data structures) that represent the relationships between words that make up a text in an understandable way. These graphs represent the relationships between words that make up the text by using vertices (nodes) for the objects to be analyzed (in this example, each of the multiple words that make up past exam data, knowledge data, necessary knowledge data, and important knowledge data) and the relationships between these objects to be analyzed as edges connecting these vertices. When using graph theory, it becomes possible to predict the meaning represented by the nodes and the relationships represented by the edges.

[0055] Specifically, the large-scale language model is trained in advance by a neural network using graph data as input. This graph data consists of vector data representing the semantic and similarity between words in the knowledge data, generated using Word2vec, and vector data representing the semantic and similarity between words in past exam questions, generated using Word2vec. Furthermore, multiple necessary knowledge data, which are input parameters, may also be converted into vector data using Word2vec and graphed before being input to the large-scale language model. This is expected to enable the output of important knowledge data with higher accuracy.

[0056] Furthermore, the important knowledge data generation unit 11d may perform important knowledge data generation preprocessing, which organizes at least one of the necessary knowledge data or past question data as the ontology described above, prior to inputting input parameters to the large-scale language model, and inputs the ontology along with the ontology. With this configuration, even if the character information related to the input may contain variations in notation, the information is organized into data suitable for input parameters to the large-scale language model and input, so it is expected that important knowledge data can be output with higher accuracy.

[0057] The important knowledge data provision unit 11e performs the important knowledge data provision process by either displaying or printing the important knowledge data, or by transmitting the important knowledge data to the user terminal 20 via the network 30 using the communication interface 13.

[0058] The problem generation unit 11f has the function of generating important knowledge confirmation questions, which are textual information about problems, based on important knowledge data, and the function of generating similar questions, which are textual information about problems similar to past questions, based on past question data. In other words, when the problem generation unit 11f generates important knowledge confirmation questions in its problem generation process, it can be treated as an important knowledge confirmation question generation unit 11f, while when it generates similar questions, it can be treated as a similar question generation unit 11f.

[0059] The Important Knowledge Confirmation Question Generation Unit 11f generates important knowledge confirmation questions based on a Large Language Model (LLM) that has been pre-trained using a neural network by inputting knowledge data, which is textual information containing the knowledge necessary to pass the target qualification exam, and past exam question data. The Important Knowledge Confirmation Question Generation Unit 11f is not limited to generating important knowledge confirmation questions from important knowledge data; it can also generate important knowledge confirmation questions from important knowledge data and other data. For example, the Important Knowledge Confirmation Question Generation Unit 11f may generate important knowledge confirmation questions from important knowledge data and past exam question data.

[0060] As shown in Figure 5, the large-scale language model used by the important knowledge confirmation question generation unit 11f has the function of outputting an important knowledge confirmation question using one or more important knowledge data as input parameters. Specifically, when a predetermined prompt (generation instruction) and input parameters are input, the large-scale language model weights one or more important knowledge data related to the input from the large number of knowledge data and past question data obtained through the above-mentioned learning, based on the degree of relevance to the question (for example, whether the information contained in the important knowledge data can be the information necessary to answer the question being generated), and generates and outputs an important knowledge confirmation question corresponding to the prompt. The neural network shown consists of a total of three layers: an input layer, a single hidden layer, and an output layer, but it is not limited to this example, and a configuration with multiple hidden layers is also possible. By increasing the number of hidden layers, more complex analysis can be performed on the input parameters, and it is expected that important knowledge confirmation questions can be output with higher accuracy. In addition, each layer has multiple nodes, but the number of these nodes is also not limited to the example shown.

[0061] Furthermore, the input parameters for the neural network's input layer must include at least important knowledge data, but may also include parameters other than important knowledge data (the +α parameters in Figure 5). For example, the past exam data mentioned above, question data created by examination institutions such as cram schools for the target qualification exam, and information regarding the difficulty level of the questions to be generated can be included as input parameters. When the large-scale language model used by the important knowledge confirmation question generation unit 11f takes important knowledge data and past exam data as input parameters, it becomes possible to output important knowledge confirmation questions that relate to the important knowledge data related to the input and whose format and the way the important knowledge is asked are similar to those in the past exam data.

[0062] Furthermore, if the target qualification exam covers multiple fields, the important knowledge confirmation questions may be comprehensive questions that test important knowledge related to some or all of these fields. On the other hand, the important knowledge confirmation questions may also be field-specific questions that test important knowledge for each of these fields. The specific content of such important knowledge confirmation questions can be easily changed by appropriately modifying the prompts input to the large-scale language model via the important knowledge confirmation question generation unit 11f.

[0063] The similar problem generation unit 11f executes a similar problem generation process, which is the process of generating similar problems, based on the large-scale language model (trained model) that has undergone the above-mentioned training. In this case, the large-scale language model has the function of outputting similar problems, with past question data as at least an input parameter. Specifically, when a predetermined prompt (generation instruction) and input parameters are input, the large-scale language model weights the past question data related to the input from the large amount of knowledge data and past question data obtained through the above-mentioned training, based on the degree of similarity with the problem (for example, whether the information required to answer the problem to be generated and the input problem are the same or similar, or how similar the past question data is to the input problem), and generates and outputs a similar problem corresponding to the prompt.

[0064] Furthermore, the input parameters for the neural network's input layer must include at least past exam data, but may also include parameters other than past exam data. For example, necessary knowledge data, important knowledge data, question data created by the aforementioned examination body, or information regarding the difficulty level of the questions can be included as input parameters. When the large-scale language model used by the similar problem generation unit 11f takes necessary knowledge data or important knowledge data in addition to past exam data as input parameters, it becomes possible, for example, to output similar problems that are similar to the questions in the past exam data used as input and that relate to that knowledge (specifically, require an understanding of that knowledge to answer).

[0065] As a result, the problem generation unit 11f is configured to generate problems related to knowledge data other than important knowledge data. For example, users can practice problems related to knowledge that is not frequently tested in the target qualification exam but is desirable to understand. Furthermore, similar problems may include the important knowledge confirmation problems mentioned above. For example, if a problem similar to a so-called frequently tested problem is generated, it is conceivable that the similar problem and the important knowledge confirmation problem will be identical or similar in content.

[0066] In the example described above, the problem generation unit 11f was treated as either a key knowledge confirmation problem generation unit 11f or a similar problem generation unit 11f depending on the type of problem generated by the single problem generation unit 11f. However, it is also possible to provide separate generation units for each type of problem, namely a key knowledge confirmation problem generation unit and a similar problem generation unit. Furthermore, the key knowledge confirmation problem generation unit 11f and the similar problem generation unit 11f may execute their respective problem generation processes using the same trained model, or they may each execute their respective problem generation processes using different trained models.

[0067] Furthermore, the problem generation unit 11f may be configured to perform a pre-problem generation process, which organizes the information related to the input as an ontology, prior to inputting the input parameters to the large-scale language model, and then input that ontology. With this configuration, even if the character information related to the input may contain variations in notation, the information is organized into data suitable for input parameters to the large-scale language model and input, so it is expected that important knowledge confirmation questions or similar questions can be output with higher accuracy, and is particularly useful when the similar question generation unit 11f performs the similar question generation process using past question data as input.

[0068] Incidentally, the pre-trained models used by the important knowledge data generation unit 11b and the question generation unit 11f are common in that they are pre-trained models that have been trained in advance by inputting knowledge data and past question data. Therefore, these generation units 11b and 11f may share and use the same pre-trained model. Such a configuration is useful, for example, when using a general-purpose pre-trained model such as ChatGPT-4o(registered trademark) as is.

[0069] Furthermore, the pre-trained models (large-scale language models) used by the necessary knowledge data generation unit 11b, the important knowledge data generation unit 11d, and the problem generation unit 11f, as described above, are all pre-trained models that have been pre-trained by inputting data including knowledge data as training data. Therefore, in the stage of executing the necessary knowledge data generation process, a pre-trained model that has been trained by inputting knowledge data is used, and in the stages of executing the important knowledge data generation process and the problem generation process, a pre-trained model obtained by additionally inputting past question data into the pre-trained model is used. Such a configuration is useful, for example, when performing fine-tuning on a general-purpose pre-trained model such as ChatGPT-4o(registered trademark) as the pre-trained model.

[0070] On the other hand, each generation unit 11b, 11d, and 11f may be configured to use a different pre-trained model.

[0071] The problem provision unit 11g executes problem provision processing in accordance with the functions of the problem generation unit 11f described above. That is, when the problem provision unit 11g provides important knowledge confirmation questions in its problem provision processing, it can be treated as an important knowledge confirmation question provision unit 11g, while when it provides similar questions, it can be treated as a similar question provision unit 11g.

[0072] The important knowledge confirmation question provision unit 11g performs the important knowledge confirmation question provision process by either displaying or printing the important knowledge confirmation questions, or by transmitting the important knowledge confirmation questions to the user terminal 20 via the network 30 using the communication interface 13.

[0073] Similarly, the similar problem provision unit 11g performs the similar problem provision process by providing similar problems by displaying or printing them, or by transmitting similar problems to the user terminal 20 via the network 30 using the communication interface 13.

[0074] In the example described above, the problem provision unit 11g was configured to be treated as either a key knowledge confirmation problem provision unit 11g or a similar problem provision unit 11g depending on the type of problem provided by the problem provision unit 11g. However, it is also possible to provide separate provision units for each type of problem provided, namely a key knowledge confirmation problem provision unit and a similar problem provision unit.

[0075] The user information acquisition unit 11h acquires user information of the user who owns the user terminal 20 by communicating with the user terminal 20 via the network 30 through the communication interface 13. User information is, for example, information that makes it possible to identify a single user from among multiple users, such as a user ID. If the user information acquisition process cannot be executed because the user terminal 20 communicating with the information processing server 10 does not have user information (for example, when communicating with a user terminal 20 of a new user who has not been assigned a user ID), the system may be configured to execute a process to assign new user information to that user instead of the user information acquisition process.

[0076] The answer data acquisition unit 11i acquires answer data, which is information about the user's answers to the important knowledge confirmation questions or similar questions provided to the user through the question provision process described above, by communicating with the user terminal 20 via the network 30 using the communication interface 13. Furthermore, the method of the answer data acquisition process is not limited to the example described above. It may also be configured to acquire information as answer data by inputting information via a keyboard or touch panel, or it may be configured to capture paper media on which answers are written as image data using a scanner, etc., and extract the character information contained in the image data as answer data.

[0077] The answer data acquisition unit 11i is configured to acquire answer data for each of the multiple important knowledge confirmation questions when multiple important knowledge confirmation questions are generated. Similarly, the answer data acquisition unit 11i is configured to acquire answer data for each of the multiple similar questions when multiple similar questions are generated.

[0078] The progress status confirmation unit 11j executes the progress status confirmation process described above based on the answer data acquired by the answer data acquisition unit 11i. The progress status confirmation unit 11j also stores the generated or updated progress data in the storage device 12 of the information processing server 10 or the storage device of the external server 40, in association with the user information of the user to whom the progress data pertains. Specifically, if the progress data for that user has already been generated, the progress status confirmation unit 11j executes a process to update the user's progress data with the newly generated progress data.

[0079] The progress data includes the user's answer status, organized by important knowledge confirmation question or related field. For example, when multiple important knowledge confirmation questions are generated and answer data is acquired for each of those questions, the progress status confirmation unit 11j organizes the progress data generated based on that answer data (specifically, the answer status, which in this example is the correct answer rate for that question) for each important knowledge confirmation question and stores it in the storage device 12 or the storage device of the external server 40. On the other hand, when multiple important knowledge confirmation questions are generated for each field and answer data is acquired for each of those questions, the progress status confirmation unit 11j organizes the progress data generated based on that answer data (specifically, the answer status, which in this example is the correct answer rate for that question) for each related field and stores it in the storage device 12 or the storage device of the external server 40.

[0080] Furthermore, the progress data may also include the user's response status (in this example, the correct answer rate for similar problems) organized by similar problems or related fields.

[0081] The progress status provision unit 11k performs progress data provision processing by communicating with the user terminal 20 via the network 30 through the communication interface 13 and transmitting the progress data to the user terminal 20 of the user identified by the user information related to the progress data. The progress status provision unit 11k can also perform progress data provision processing by displaying or printing the progress data.

[0082] The learning plan information generation unit 11l generates or updates learning plan information, which is information about the user's future learning plan related to the progress data, based on the progress data generated or updated by the progress status confirmation unit 11j or the progress data stored in the storage device 12 of the information processing server 10 or the storage device of the external server 40, and stores it in the storage device 12 of the information processing server 10 or the storage device of the external server 40. The learning plan information generation unit 11l may also be configured to store the generated learning plan information in association with the progress data from which it was generated or the user information of the user related to that progress data.

[0083] The learning plan information provision unit 11m performs the learning plan information provision process by either displaying or printing the learning plan information, or by transmitting the learning plan information to the user terminal 20 held by the user via communication with the user terminal 20 through the network 30 via the communication interface 13.

[0084] The question information acquisition unit 11n executes the above-described question information acquisition process by communicating with the user terminal 20 via the network 30 through the communication interface 13, or by receiving information through the input interface 14.

[0085] The question and answer information includes textual information relating to user questions about information handled (acquired, used, extracted, generated, or provided) by the information processing server 10, such as past exam data, knowledge data, necessary knowledge data, important knowledge data, important knowledge confirmation questions or similar questions, or answer data.

[0086] The response information generation unit 11o executes the above-described response information generation process based on a large language model (LLM) that has been previously trained using a neural network by inputting the above-described knowledge data. Incidentally, the data input for training the large language model is not limited to knowledge data; it may also be configured to train the model by inputting a wide range of data related to the target qualification exam. Examples of such wide range of data include data on the question trends of the target qualification exam and the average score of successful candidates in a given year.

[0087] The large-scale language model used by the response information generation unit 11o has the function of outputting response information using question information as at least an input parameter. Specifically, when a predetermined prompt (generation instruction) and input parameters are input, the large-scale language model weights a large amount of knowledge data obtained through the above-mentioned learning or a wide range of data related to the target qualification examination based on the degree of relevance to the question information related to the input (specifically, the degree to which it is necessary to respond to the user's question and resolve the doubt, etc.), and extracts or generates and outputs response information corresponding to the prompt.

[0088] The response information may simply be a list of extracted information, or it may be in the form of conversational text (text information) responding to questions from the user.

[0089] Furthermore, the input parameters for the neural network's input layer must include at least question information, but other parameters may also be included. For example, past question data, necessary knowledge data, important knowledge data, etc., related to the question information can be included as input parameters. This enables the response information generation unit 11o to output accurate response information in response to the question information.

[0090] Specifically, if the question is something like, "In what specific situations is the unit 'sievert' used?", then additional necessary knowledge data such as, "Sievert is a unit that indicates the magnitude of the biological effects that the human body receives from radiation exposure," can be input in relation to the question. In response, the response information generation unit 11o avoids including the aforementioned necessary knowledge data redundantly in the response information and generates textual information as response information such as, "Sievert is used to assess the risk of radiation exposure, taking into account the different effects depending on the type of radiation and the tissue of the human body."

[0091] Furthermore, the response information generation unit 11o may be configured to perform a pre-processing step for response information generation, which organizes the question information as an ontology, before inputting the question information to the large-scale language model, and then input that ontology. This is expected to enable the output of response information with higher accuracy, as it organizes the question information (text information) from users, which may contain diverse expressions, into uniform (uniquely understandable) data suitable as input parameters for the large-scale language model.

[0092] The response information provision unit 11p performs the response information provision process by either displaying or printing the response information, or by transmitting it to the user terminal 20, which is the source of the question information related to the response information, via communication with the user terminal 20 via the network 30 using the communication interface 13.

[0093] This allows users to obtain response information that resolves their questions by sending question information to the question information acquisition unit 11n if they have any doubts while studying for the certification exam. For example, if the question information includes textual information about the user's question regarding the answer data, such as "What score is this answer worth?", the response information provision unit 11p will provide the user with the scoring result of that answer data. In other words, the response information generation unit 11o can also be configured to have a scoring function for the answer data.

[0094] The storage device 12 and the storage device of the data server 40 are provided with at least the following: a past exam data storage area 12a for storing the past exam data; a knowledge data storage area 12b for storing the knowledge data; a necessary knowledge data storage area 12c for storing the required knowledge data; a key knowledge data storage area 12d for storing the important knowledge data; a question storage area 12e for storing the key knowledge confirmation questions and similar questions; a user information storage area 12f for storing the user information; a progress data storage area 12g for storing the progress data; a solution data storage area 12h for storing the solution data; a search data storage area 12i for storing the search database; a learning plan information storage area 12j for storing the learning plan information; and a question and answer information storage area 12k for storing the question and answer information.

[0095] The question and answer information storage area 12k stores question information and corresponding response information in an associated state. It is not mandatory for both question information and response information to be stored in the question and answer information storage area 12k; a separate question information storage area for storing question information and a response information storage area for storing response information may also be provided.

[0096] Next, the configuration of the user terminal 20 will be described based on Figure 6.

[0097] Figure 6 is a block diagram showing the configuration of a user terminal. The user terminal 20 comprises a control unit 21 composed of a CPU and RAM, etc., and capable of implementing various functions (means) through program execution; a storage device 22 including RAM and storage such as an SSD or HDD; a wired or wireless communication interface 23 enabling communication via the network 30; an input interface 24 such as a keyboard, mouse, touch panel, or scanner for inputting information; and an output interface 25 such as a monitor, printer, or speaker for outputting information. Note that the user terminal 20 may be an information terminal dedicated to this information processing system, but a general-purpose computer such as a personal computer, smartphone, or tablet terminal may also be used as the user terminal 20.

[0098] The control unit 21 includes at least an information acquisition unit 21a which performs an information acquisition process, which is the process of acquiring various information such as necessary knowledge data, important knowledge data, important knowledge confirmation questions, similar questions, progress data, learning plan information and response information from the information processing server 10 by communicating with the information processing server 10 via the network 30 through the communication interface 23, and an information provision unit 21b which performs an information provision process, which is the process of providing various information such as past question data, user information, answer data and question information to the information processing server 10 by communicating with the information processing server 10 via the network 30 through the communication interface 23.

[0099] Next, we will explain the progress data (answer status) and learning plan information provided to each user, based on Figure 7.

[0100] Figure 7 is a table showing an example of each user's answer status organized by subject area. In the example shown, the answer status of each user, identified by multiple user IDs, is organized by subject areas A to C, which are covered in the target certification exam. Note that the specific number of users and the content and number of subject areas are not limited to the example shown.

[0101] In this example, the answer status represents the user's correct answer rate for the important knowledge confirmation questions. Specifically, user ID 001 (hereinafter referred to as "User 001") has an 80% correct answer rate for the important knowledge confirmation questions related to field A, a 20% correct answer rate for the important knowledge confirmation questions related to field B, and a 70% correct answer rate for the important knowledge confirmation questions related to field C. Similarly, user ID 002 (hereinafter referred to as "User 002") has a 30% correct answer rate for the important knowledge confirmation questions related to field A, a 90% correct answer rate for the important knowledge confirmation questions related to field B, and a 50% correct answer rate for the important knowledge confirmation questions related to field C.

[0102] The organized response status is then provided to the user who submitted the response by the progress status provision unit 11k.

[0103] Furthermore, the learning plan information generation unit 11l generates learning plan information based at least on the progress data described above (specifically, the answer status of the important knowledge confirmation questions, and in this example, the correct answer rate). Specifically, the learning plan information generation unit 11l generates learning plan information that suggests learning in areas where the target user has a low correct answer rate (i.e., areas where they are weak).

[0104] The learning plan information generated in this manner is provided to the user by the learning plan information provision unit 11m.

[0105] In the illustrated example, because User 001's accuracy rate in field B is lower compared to fields A and C, information suggesting that User 001 should study field B is generated and provided to User 001 as learning plan information. On the other hand, information suggesting that User 002 should study field A is generated and provided to User 002 using a similar method.

[0106] Furthermore, the learning plan information generation unit 11l may be configured to set predetermined correct answer rate standards for each field and generate learning plan information that suggests learning in fields where the target user's correct answer rate falls below that standard. For example, if the correct answer rate standard is set to 60%, user 001 will be provided with information suggesting learning in field B, while user 002 will be provided with information suggesting learning in fields A and C.

[0107] Furthermore, the learning plan information may include important knowledge review questions related to areas with low correct answer rates, important knowledge data related to those important knowledge questions, or other important knowledge review questions that are highly relevant to that important knowledge data. This allows users to work on (review) important knowledge review questions related to their weak areas when they obtain the learning plan information, enabling them to learn more efficiently.

[0108] Incidentally, the learning plan information may also include information suggesting that the frequency of studying areas of strength should be reduced. In the illustrated example, the learning plan information provided to user 001 includes information suggesting that the frequency of studying areas A and C, which are areas where the correct answer rate for important knowledge confirmation questions is high, should be reduced, while the learning plan information provided to user 002 includes information suggesting that the frequency of studying area B should be reduced using a similar method.

[0109] Furthermore, if the progress confirmation unit 11j acquires answer data for multiple important knowledge confirmation questions without distinguishing by field, it may be configured to organize the correct answer rate for each of these important knowledge confirmation questions. In this case, the learning plan information generation unit 11l is configured to generate learning plan information that suggests the target user should study important knowledge confirmation questions with low correct answer rates. The learning plan information may also include important knowledge confirmation questions with low correct answer rates, important knowledge data related to those important knowledge confirmation questions, or other important knowledge confirmation questions that are highly related to that important knowledge data.

[0110] In the example above, the user's correct answer rate for the important knowledge confirmation questions was used as the answer status for information organization. However, this is not the only example; the correct answer rate for similar questions, or the correct answer rate for both the important knowledge confirmation questions and similar questions, may also be used as the answer status and organized by question or by subject area. Furthermore, if the correct answer rates for both the important knowledge confirmation questions and similar questions are used as the answer status, it is also possible to organize the answer status by the type of question.

[0111] Next, a series of processes related to the generation of information by a computer, as implemented by the program according to the present invention, will be described based on Figure 8.

[0112] Figure 8 is a flowchart showing the procedure for a series of processes related to the generation of necessary knowledge data and important knowledge data. The information processing server (information processing device) 10 starts the series of information processing shown in the figure at a predetermined timing.

[0113] In step S101, the past exam data acquisition unit 11a of the information processing server 10 executes a past exam data acquisition process to acquire past exam data, and then proceeds to step S102.

[0114] In step S102, the necessary knowledge data generation unit 11b of the information processing server 10 executes a necessary knowledge data generation process to generate necessary knowledge data, which is textual information relating to the knowledge needed to answer the questions in the past exam data acquired by the past exam data acquisition unit 11a, and then proceeds to step S103.

[0115] In step S103, the important knowledge data generation unit 11d of the information processing server 10 executes an important knowledge data generation process to generate important knowledge data, which is textual information about common knowledge, from multiple necessary knowledge data generated by the necessary knowledge data generation unit 11b from each of multiple past exam question data, and then completes the series of processes.

[0116] Step S102 is repeated until a sufficient amount of necessary knowledge data is generated to enable the important knowledge data generation process in step S103.

[0117] Furthermore, the information processing server 10 (necessary knowledge data generation unit 11b) may perform necessary knowledge data generation preprocessing after step S101 and before step S102, which is the process of organizing the past exam data acquired in step S101 as the ontology described above.

[0118] Furthermore, the information processing server 10 (important knowledge data generation unit 11d) may perform important knowledge data generation preprocessing after step S102 and before step S103, which is the process of organizing the necessary knowledge data generated in step S102 as the ontology described above.

[0119] The program and information processing device configured as described above make it possible to provide efficient learning support specifically for qualification exams where past questions are considered important.

[0120] In the example described above, both the necessary knowledge data generation unit 11b and the important knowledge data generation unit 11d are configured to generate information based on a trained model. However, the system is not limited to this configuration, and it may be configured such that at least one of the necessary knowledge data generation unit 11b or the important knowledge data generation unit 11d generates information based on a trained model.

[0121] If only the necessary knowledge data generation unit 11b generates information (necessary knowledge data) based on a trained model, the important knowledge data generation unit 11d generates important knowledge data from multiple necessary knowledge data, or from multiple necessary knowledge data and past exam data, based on a predetermined algorithm that does not use a trained model. On the other hand, if only the important knowledge data generation unit 11d generates information (important knowledge data) based on a trained model, the necessary knowledge data generation unit 11b generates necessary knowledge data from past exam data based on a predetermined algorithm that does not use a trained model.

[0122] Furthermore, in the example described above, the important knowledge confirmation question generation unit 11f is configured to generate important knowledge confirmation questions from important knowledge data based on a pre-trained model that has been trained in advance by inputting the aforementioned knowledge data and past question data. However, the configuration is not limited to this, and important knowledge confirmation questions may be generated based on a predetermined algorithm that does not use a pre-trained model. Similarly, it is not mandatory for the similar question generation unit 11f to use a pre-trained model.

[0123] Furthermore, while the pre-trained model in the above example was a large-scale language model, the configuration is not limited to this; a configuration using a pre-trained model obtained through supervised learning is also acceptable.

[0124] In this case, the necessary knowledge data generation unit 11b uses a trained model obtained by training a neural network using multiple sets of training data, with past exam data and the necessary knowledge data (training data) corresponding to those past exam data as a single dataset, evaluating the difference between the necessary knowledge data output by the neural network and the training data using a loss function, and updating (correcting) the weights and biases of the neural network based on backpropagation or the like so that the loss by the loss function is small, until the desired data can be output based on multiple sets of training data.

[0125] When using a pre-trained model obtained through supervised learning as the pre-trained model for the important knowledge data generation unit 11d, a pre-trained model obtained by the same method as described above is used, with multiple necessary knowledge data and important knowledge data (training data) corresponding to the multiple necessary knowledge data as a single dataset.

[0126] Furthermore, when using a pre-trained model obtained through supervised learning as the pre-trained model for the important knowledge confirmation question generation unit 11f, the important knowledge data and the important knowledge confirmation questions (training data) corresponding to the important knowledge data are used as a single dataset, and a pre-trained model obtained by the same method as described above is used.

[0127] Furthermore, when using a pre-trained model obtained through supervised learning as the pre-trained model for the similar problem generation unit 11f, the past problem data and similar problems (training data) corresponding to the past problem data are used as a single dataset, and the pre-trained model obtained by the same method as described above is used. [Explanation of Symbols]

[0128] 10. Information Processing Server (Information Processing Device) 11a Past Exam Data Acquisition Unit 11b Required Knowledge Data Generation Unit 11c Required Knowledge Data Provision Department 11d Important Knowledge Data Generation Unit 11e Important Knowledge Data Provision Department 11f Question generation section (Important knowledge confirmation question generation section) 11g Question Providing Department (Important Knowledge Confirmation Question Providing Department) 11i Answer Data Acquisition Unit 11j Progress Status Confirmation Department 11k Progress Reporting Department 12 Storage device

Claims

1. On the computer, A past exam data acquisition function that retrieves past exam data, including textual information about past questions that have appeared on the target qualification exam, The necessary knowledge data generation function generates necessary knowledge data, which is textual information relating to the knowledge required to answer the questions in the past exam data acquired by the aforementioned past exam data acquisition function, A key knowledge data generation function is implemented that generates key knowledge data, which is textual information about common knowledge, from multiple sets of key knowledge data generated by the key knowledge data generation function from each of the multiple sets of past exam question data. At least one of the aforementioned necessary knowledge data generation function or the aforementioned important knowledge data generation function is configured to generate information based on a pre-trained model that has been trained by inputting at least knowledge data, which is textual information containing the knowledge necessary to pass the qualification examination. A program characterized by the following features.

2. A past exam data acquisition unit acquires past exam data, including textual information about past questions that have appeared on the target qualification exam. A necessary knowledge data generation unit generates necessary knowledge data, which is textual information relating to the knowledge needed to answer the questions in the past exam data acquired by the aforementioned past exam data acquisition unit, The system comprises: an important knowledge data generation unit that generates important knowledge data, which is textual information relating to common knowledge, from a plurality of necessary knowledge data generated by the necessary knowledge data generation unit from each of the plurality of past exam question data; At least one of the necessary knowledge data generation unit or the important knowledge data generation unit is configured to generate information based on a pre-trained model that has been trained by inputting at least knowledge data, which is textual information containing the knowledge necessary to pass the qualification examination. An information processing device characterized by the following:

3. The aforementioned important knowledge data generation unit generates important knowledge data based on a pre-trained model that has been trained in advance by inputting the knowledge data and the past exam question data. The information processing apparatus according to claim 2.

4. The aforementioned important knowledge data generation unit generates the important knowledge data from a plurality of the necessary knowledge data and the past exam question data. The information processing apparatus according to claim 2.

5. The aforementioned trained model is an LLM trained using a neural network. The information processing apparatus according to claim 2.

6. The system comprises at least one of the following: a necessary knowledge data providing unit that provides necessary knowledge data generated by the necessary knowledge data generation unit, and a key knowledge data providing unit that provides key knowledge data generated by the key knowledge data generation unit. The information processing apparatus according to claim 2.

7. The aforementioned past exam data includes textual information about past questions that have appeared on the target qualification exam, as well as information about the correct answers to those questions. The information processing apparatus according to claim 2.

8. The system includes a knowledge data generation unit that generates important knowledge confirmation questions, which are textual information related to the questions, based on the important knowledge data generated by the important knowledge data generation unit based on a pre-trained model that has been trained in advance by inputting knowledge data, which are textual information containing the knowledge necessary to pass the aforementioned qualification examination, and the aforementioned past exam question data. The information processing apparatus according to claim 2.

9. The aforementioned important knowledge confirmation question generation unit generates the important knowledge confirmation questions based on the important knowledge data and the past question data. The information processing apparatus according to claim 8.

10. The system includes a unit that provides important knowledge confirmation questions generated by the aforementioned important knowledge confirmation question generation unit. The information processing apparatus according to claim 8.

11. A storage device that stores acquired and generated information, user information which is information of one or more users, and progress data which is information regarding the learning progress of each user, The answer data acquisition unit acquires the user's answer, The system includes a progress status confirmation unit that generates or updates the progress data of the user who provided the answer based on the answer data acquired by the answer data acquisition unit, and stores it in the storage device, The aforementioned important knowledge confirmation question provision unit is configured to provide the user with the aforementioned important knowledge confirmation questions. The information processing apparatus according to claim 10.

12. The progress data provided to each user includes the user's answer status, organized by each of the important knowledge confirmation questions or related fields. The information processing apparatus according to claim 11.

13. The aforementioned progress data is provided by a progress status provision unit that provides the relevant user information. The information processing apparatus according to claim 11.

Images