Problem generation device, problem generation method, and problem generation program

Abstract

This invention provides a problem generation device, a problem generation method, and a problem generation program that easily generate problems including diagrams and charts. [Solution] In a problem generation system that generates diagrams and charts included in a problem, the server 10, in the extraction instruction step S102, gives an extraction instruction to the artificial intelligence 20 to extract diagram and chart target information cited in the problem statement from the problem statement input by the user terminal 30, in the acquisition instruction step S103, gives an acquisition instruction to the artificial intelligence to acquire diagram and chart elements constituting the diagram and chart target information from a predetermined database, and in the output step S104, outputs source code including diagram and chart generation code that outputs the diagram and chart elements as diagrams or tables. The artificial intelligence has a first trained model that takes the problem statement as input and outputs the diagram and chart target information, and a second trained model that takes the diagram and chart target information as input and outputs the diagram and chart elements constituting the diagram and chart target information.

Description

Technical Field

[0001] The present invention relates to a problem generation device, a problem generation method, and a problem generation program.

Background Art

[0002] There is a known technique in which a computer generates a problem for a respondent to answer.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Patent Document 1 discloses an information analysis device that generates numerical data for creating a graph from search results based on input processing conditions and creates a graph.

[0005] Generally, as a problem for evaluating academic ability or understanding in a predetermined field, there is a so-called chart reading problem in which a chart is presented to a respondent and the respondent is asked to explain the information that can be read from the chart. Therefore, there is a need for a technique for easily generating so-called chart reading problems.

[0006] An object of the present invention is to provide a problem generation device that easily generates a problem including a chart.

Means for Solving the Problems

[0007] To achieve the above objective, a problem generation device according to one aspect of the present invention is a problem generation device that generates diagrams and tables included in a problem, comprising a processor, wherein in an extraction instruction step, the processor gives an extraction instruction to the artificial intelligence to extract diagram-to-table information cited in the problem statement from the problem statement; in an acquisition instruction step, the processor gives an acquisition instruction to the artificial intelligence to acquire diagram elements constituting the diagram-to-table information from a predetermined database; and in an output step, the processor outputs source code including diagram generation code that outputs the diagram elements as a diagram or a table, wherein the artificial intelligence has a first trained model that takes the problem statement as input and outputs the diagram-to-table information, and a second trained model that takes the diagram-to-table information as input and outputs the diagram elements constituting the diagram-to-table information.

[0008] The system further performs a selection instruction step in which it refers to a storage unit that stores a plurality of diagram generation codes that output predetermined diagram elements as diagrams or tables in different forms, and gives a selection instruction to the artificial intelligence to select the diagram generation code to apply to the diagram elements. The artificial intelligence has a third trained model that takes information obtained by analyzing the problem statement using natural language processing as input and outputs the diagram generation code to apply to the diagram elements, and the output step may output source code that includes an instruction to output the diagram elements using the selected diagram generation code.

[0009] The input step may accept input of tags indicating the position where figures or tables are to be inserted in the problem statement, and the output step may output the source code in which the figure or table generation code is placed at the position where the tags are entered.

[0010] To achieve the above objective, a problem generation method according to another aspect of the present invention is a problem generation method for generating diagrams and tables included in a problem, the method comprising: an extraction instruction step in which a computer gives an extraction instruction to artificial intelligence to extract diagrammatic information cited in the problem statement from the problem statement; an acquisition instruction step in which a computer gives an acquisition instruction to artificial intelligence to acquire diagrammatic elements constituting the diagrammatic information from a predetermined database; and an output step in which an output of source code including diagram generation code that outputs the diagrammatic elements as diagrams or tables, wherein the artificial intelligence has a first trained model that takes the problem statement as input and outputs the diagrammatic information, and a second trained model that takes the diagrammatic information as input and outputs the diagrammatic elements constituting the diagrammatic information.

[0011] To achieve the above objective, a problem generation program according to yet another aspect of the present invention is a problem generation program that generates charts and graphs included in a problem, and causes a computer to execute an extraction instruction step in which it gives an extraction instruction to artificial intelligence to extract chart and graph target information cited in the problem statement from the problem statement; an acquisition instruction step in which it gives an acquisition instruction to artificial intelligence to acquire chart and graph elements constituting the chart and graph target information from a predetermined database; and an output step in which it outputs source code including chart and graph generation code that outputs the chart and graph elements as a chart or graph, wherein the artificial intelligence has a first trained model that takes the problem statement as input and outputs the chart and graph target information, and a second trained model that takes the chart and graph target information as input and outputs chart and graph elements constituting the chart and graph target information. Computer programs can be provided by storing them on various data-readable storage media, or by making them available for download via networks such as the Internet. [Effects of the Invention]

[0012] According to the present invention, problems including diagrams and charts can be easily generated. [Brief explanation of the drawing]

[0013] [Figure 1]This figure shows the overall configuration and functional configuration of a problem generation device according to an embodiment of the present invention. [Figure 2] This is a sequence diagram showing the processing flow executed by the problem generation device described above. [Figure 3] This figure shows an example of a problem generation screen displayed on a user terminal by the problem generation device described above. [Modes for carrying out the invention]

[0014] Hereinafter, embodiments of the problem generation device according to the present invention will be described with reference to the drawings.

[0015] ●Problem Generation System (1)● ● Overview First, I will explain the general flow of the problem generation system 1. The problem generation system 1 comprises at least a server device 10. The server device 10 is an example of a problem generation device that, based on the problem statement, creates diagrams and charts of information cited in the problem statement and outputs them together with the problem statement. The user is, for example, a service administrator who manages a service that provides generated problems. The problems may be, for example, so-called quizzes for entertainment, or problems that serve as learning materials to measure or improve academic ability. The problems output by the problem generation system 1 include diagrams and charts, and consist of the problem statement and the diagrams and charts presented side by side. In other words, the problems output by the problem generation system 1 are so-called diagram reading problems in which the answerer refers to the diagrams and charts to answer the question. Furthermore, the technical scope of the present invention is not limited to systems that generate so-called problems requiring appropriate answers, but also extends to systems that generate diagrams and charts included in appropriate texts.

[0016] The server device 10 transmits an instruction regarding the generation of the chart cited in the problem statement to the artificial intelligence unit 20. The problem statement for creating the chart may, for example, accept an input from the user or may be a problem statement generated by the server device 10 or an external device. Also, the artificial intelligence unit 20 transmits the output chart reading problem to the server device 10 as a response. The server device 10 selects the generated problem as needed and registers it in the storage device as a problem management database. When the service administrator requests a question via the user terminal 30, the chart reading problem is displayed on the user terminal 30.

[0017] Also, the user of the problem generation system 1 may be the respondent who answers the generated problem. In this case, when the user as the respondent calls the problem via the user terminal 30, the server device 10 transmits the problem statement to the artificial intelligence unit 20. The artificial intelligence unit 20 refers to the chart element database DB1 and the chart generation database DB2 and generates a chart reading problem. Also, the artificial intelligence unit 20 transmits the output chart reading problem to the user terminal 30, and the chart reading problem is displayed on the user terminal 30.

[0018] ● Functional block Hereinafter, the functional blocks of each device related to the problem generation system 1 will be described in detail. ● Server device 10 As shown in FIG. 1, the server device 10 is configured to be communicable with the user terminal 30 used by the user via the network NW. The server device 10 may be composed of a hardware device, or some or all of its functions may be realized by a cloud computer. Also, each configuration of the server device 10 may be realized by an API (Application Programming Interface). The mutual communication between the server device 10 and the user terminal 30 is wireless in this embodiment, but some or all of the connections may be wired. Furthermore, the server device 10 may be composed of a plurality of hardware configurations. In this case, the plurality of hardware configurations may be connected by wire or wirelessly, and information may be transmitted and received between them.

[0019] Also, the server device 10 is configured to be communicable with the artificial intelligence unit 20. The artificial intelligence unit 20 has the function of AI (Artificial Intelligence). Further, the artificial intelligence unit 20 can search the chart element database DB1, the chart generation database DB2, and the statistical analysis database DB3, which will be described in detail later, and acquire information. The server device 10 transmits appropriate instructions to the artificial intelligence unit 20 and acquires the information output from the artificial intelligence unit 20.

[0020] ● User terminal 30 The user terminal 30 is a terminal operated by the user, such as a smartphone, a tablet terminal, or a personal computer. The user terminal 30 mainly constitutes a functional block including a display unit 31, an operation reception unit 32, and a communication processing unit 33 by a CPU (Central Processing Unit), a computer program executed by the CPU, a RAM (Random Access Memory) and a ROM (Read Only Memory) that store the computer program and predetermined data.

[0021] The display unit 31 is realized by a display or the like for outputting data. The problem received from the server device 10 is displayed on the display unit 31.

[0022] The operation reception unit 32 is realized by a touch panel, a keyboard, a mouse, a microphone, or the like for inputting data. The operation reception unit 32 can input a problem statement regarding the problem to be generated.

[0023] The communication processing unit 33 is a processing unit that can execute data transmission and reception processing according to a predetermined protocol with the server device 10 via a network NW such as the Internet, and is realized by an application or a web browser or the like.

[0024] ●Server device 10 Here, the configuration of the server device 10 will be explained using Figure 1. The server device 10 is an information processing device that performs problem generation processing while exchanging data with the user terminal 30 via the network NW. The server device 10 sends instructions to the artificial intelligence unit 20 and obtains output from the artificial intelligence unit 20.

[0025] As shown in Figure 1, the server device 10 mainly consists of a display control unit 11, a storage control unit 12, an AI control unit 13, an output control unit 14, and a communication processing unit 19, comprising a CPU (Central Processing Unit, an example of a processor in the claims), a computer program executed by the CPU, and RAM (Random Access Memory) and ROM (Read Only Memory) for storing the computer program and predetermined data.

[0026] ●Display control unit 11 The display control unit 11 executes processing to display a system screen related to the problem generation system on the user terminal 30. For example, the display control unit 11 performs processing such as generating and sending an HTML (Hyper Text Markup Language) file to display a web page showing the system screen on the user terminal 30. The display control unit 11 may also perform processing such as generating and sending display data for the application that uses the problem generation system 1.

[0027] The display control unit 11, for example, causes the user terminal 30 to display a generation screen G100 (see Figure 3) that displays the problem output by the artificial intelligence unit 20, i.e., the problem statement and a diagram or table. Furthermore, the display control unit 11 may display a generation screen G100 or other appropriate screen on the user terminal 30 that accepts input of the problem statement to be used to generate the figures and tables. In addition, the display control unit 11 accepts input of a tag indicating the figure and table insertion position along with the problem statement. This tag may be a natural language phrase such as [Figure / Table Insertion Point], or it may be a string of characters in a predetermined programming language or machine language.

[0028] ● Memory control unit 12 The memory control unit 12 is a functional unit that controls the appropriate storage device provided by the server device 10 and performs data writing and reading. The memory control unit 12 stores, for example, the problem statement to be used to generate a diagram or chart, and the diagram generation code acquired by the AI ​​control unit 13 (described later) in the storage device. The memory control unit 12 may also store the history of problems presented to the user and the correct answers to those problems in the storage device.

[0029] ● AI control unit 13 The AI ​​control unit 13 is a functional unit that controls the input to the artificial intelligence unit 20. The various instructions generated by the AI ​​control unit 13 are, for example, prompts, but are not limited to prompts as long as they are instructions that the artificial intelligence unit 20 can interpret.

[0030] The AI ​​control unit 13, for example, sends an extraction instruction to the artificial intelligence unit 20 via the communication processing unit 19 to extract information to be turned into diagrams or tables that is cited in the problem statement.

[0031] The AI ​​control unit 13 may, in place of or in addition to the problem statement, transmit information obtained by analyzing the input problem statement using natural language processing as input information to the artificial intelligence unit 20. In this case, the first trained model is a trained model that takes information obtained by analyzing the problem statement using natural language processing as input.

[0032] Natural language processing (NLP) analysis is performed using appropriate processes such as morphological analysis, syntactic analysis, or semantic analysis. Through NLP analysis, information such as "sales trends," "time progression," and "comparison targets" is extracted from the problem statement. For example, if the problem statement is "The graph below shows the sales trends of product A from 2019 to 2023. Explain the trends that can be observed from this graph," then NLP will extract terms such as "2019," "2023," "product A," "sales," "trend," and "graph" from the problem statement. The information to be graphed is a term that identifies the information to be graphed; for example, in the above example, it is "the sales trends of product A from 2019 to 2023." The information to be graphed can be included as a prompt for acquisition instructions to retrieve graph elements, as described later.

[0033] Natural language processing analysis may be performed as a preprocessing step using a rule-based approach by an appropriate functional unit provided by the server device 10. Alternatively, the analysis using natural language processing may be performed by the artificial intelligence unit 20. In this case, for example, the AI ​​control unit 13 may send an instruction to the artificial intelligence unit 20 to analyze the problem statement using natural language processing and then extract information to be plotted using the analysis results.

[0034] Furthermore, the AI ​​control unit 13 transmits an acquisition instruction to the artificial intelligence unit 20 via the communication processing unit 19 to acquire the chart elements constituting the information to be charted from the chart element database DB1. The chart elements are actual data representing the information to be charted, so-called raw data sets, such as numerical tables, graph data, or metadata. The chart elements may also be time-series data including changes over time, or multiple data sets having appropriate conceptual relationships. The chart elements are, for example, data that is appropriately published by public institutions such as statistics bureaus or on academic data portal sites, and may include economic indicators, social statistics, or academic data.

[0035] Here, the chart element database DB1 is pre-specified by an appropriate administrator, including the user. The chart element database DB1 may be, for example, a data set associated with a specific domain on the internet. In this case, the appropriate administrator can identify the chart element database DB1 by specifying the domain. Alternatively, the chart element database DB1 may be a data set made available through an appropriate API or an economic indicator provision service. With this configuration, chart elements can be extracted based on the information contained in the specified data set.

[0036] Furthermore, the AI ​​control unit 13 may instruct the artificial intelligence unit 20 to obtain the desired chart element if it cannot obtain the chart element to be acquired from the chart element database DB1. This instruction may include a directive to select and use a processing function from a pre-prepared database, such as the statistical analysis database DB3, for the calculation of the extracted chart element.

[0037] Furthermore, the AI ​​control unit 13 refers to the chart generation database DB2 and sends a selection instruction to the artificial intelligence to select a chart generation code to apply to the chart elements.

[0038] Here, the chart generation database DB2 is a database that stores multiple chart generation codes that output chart elements as charts or tables of different forms. The chart generation code is code written in an appropriate programming language such as Python. The chart generation database DB2 is an example of a storage unit in the claims and is specified in advance by an appropriate administrator, including a user. The chart generation database DB2 stores multiple chart generation codes that output chart elements as charts or tables. The multiple chart generation codes stored in the chart generation database DB2 each output chart elements in different forms, and include chart generation codes that output charts and chart generation codes that output tables. In addition, the chart generation database DB2 stores chart generation codes that output various forms of charts, such as line graphs, bar graphs, pie charts, box plots, scatter plots, and bubble charts, and by referencing chart elements in the chart generation code, it is possible to obtain code that draws charts of appropriate forms.

[0039] Furthermore, the chart generation database DB2 may be a so-called closed-system data set that is accessible only to designated users. The chart generation database DB2 may be constructed with an appropriate structure. In the example in Figure 1, the chart generation database DB2 is described as being connected to the server device 10 via a network NW, but it may be directly connected to the server device 10 without going through a network NW, or it may be connected via a communication network different from the network NW.

[0040] As described above, by configuring the artificial intelligence unit 20 to select a chart generation code from the data contained in the chart generation database DB2, unintended behavior of the artificial intelligence unit 20 can be suppressed and more accurate charts can be generated compared to a configuration in which the artificial intelligence unit 20 directly outputs charts.

[0041] The selection instruction may also be an instruction to select a chart generation code based on information obtained by analyzing the problem statement using natural language processing. In this case, the artificial intelligence unit 20 has a third trained model that takes information obtained by analyzing the problem statement using natural language processing as input and outputs a chart generation code to be applied to the chart elements. With such a configuration, it is possible to select a chart that is suitable for reading the information to be read from the chart in the problem. Alternatively, the selection instruction may also be an instruction to select a chart generation code based on the chart elements, for example.

[0042] The AI ​​control unit 13 may determine the setting information for the generated chart, such as axis elements, units for each axis, axis labels, legends, or scales, and send instructions to the artificial intelligence unit 20 to include this information in the chart generation code. In this case, the artificial intelligence unit 20 has a suitable trained model that takes the problem statement or information obtained by analyzing the problem statement using natural language processing as input and outputs the setting information for the chart. With such a configuration, the chart can be displayed in a manner more suitable for reading.

[0043] Furthermore, the server device 10 may execute a correct answer generation process to generate the correct answers to the chart reading problems. The correct answer generation process will be described later.

[0044] The output control unit 14 outputs source code for outputting a chart reading problem. The source code may be written in an appropriate programming language, assembly language, or appropriate intermediate code. The output control unit 14 may also output information in machine language that enables the output of a chart reading problem.

[0045] The source code output by the output control unit 14 includes a problem statement generation code that outputs the problem statement, and a figure / table generation code that outputs figure / table elements as figures or tables. For example, the output control unit 14 generates a problem statement generation code with appropriate markdown formatting applied to the problem statement. The output control unit 14 also places the figure / table generation code at the location where a tag indicating the figure / table insertion position is entered. Furthermore, if the server device 10 holds the correct answer to a chart reading problem, the output control unit 14 may output a correct answer generation code that outputs the correct answer along with the chart reading problem.

[0046] The display control unit 11 executes the source code generated by the output control unit 14 to display a chart reading problem, including the problem statement and charts, with adjusted formatting, on the user terminal 30.

[0047] ● Artificial Intelligence Department 20 The Artificial Intelligence Unit 20 is an artificial intelligence (AI) equipped with language models such as transformers including BART (Bidirectional and Auto-regressive Transformer), BERT (Bidirectional Encoder Representations from Transformers), or GPT (Generative Pretrained Transformer, including GPT-1, GPT-2, GPT-3, GPT-4, etc.), and particularly learning models such as large language models (LLM). A learning model (also called a machine learning model) refers to a learning model based on a machine learning algorithm. Specific machine learning algorithms include nearest neighbors, naive Bayes, decision trees, and support vector machines. Deep learning, which uses neural networks to generate features and connection weighting coefficients for learning, is another example. The Artificial Intelligence Unit 20 can appropriately apply the above algorithms.

[0048] The artificial intelligence unit 20 has a pre-trained model that has been appropriately machine-learned. The training data may be provided by an administrator or the like, or it may include information collected from the internet or elsewhere. For example, the artificial intelligence unit 20 has a first pre-trained model that takes a problem statement as input and outputs information to be graphed / tableted, a second pre-trained model that takes information to be graphed / tableted as input and outputs graph elements that constitute the information to be graphed / tableted, and a third pre-trained model that takes a problem statement or information obtained by analyzing the problem statement using natural language processing as input and outputs graph generation code to be applied to the graph elements. A pre-trained model has already obtained appropriate training data, and additional training can be performed as needed.

[0049] The artificial intelligence unit 20, in response to extraction instructions from the AI ​​control unit 13, uses a first trained model to take the problem statement as input and output information to be graphed or charted.

[0050] Furthermore, in response to acquisition instructions from the AI ​​control unit 13, the artificial intelligence unit 20 uses a second trained model to take the information to be graphed as input and output graph elements that constitute the information to be graphed. The artificial intelligence unit 20 refers to the graph element database DB1 and acquires graph elements. The artificial intelligence unit 20 performs API calls and data fetching to acquire the raw dataset as graph elements.

[0051] Furthermore, the artificial intelligence unit 20 selects a chart generation code to be applied to the chart element using the third trained model, in response to a selection instruction from the AI ​​control unit 13. The artificial intelligence unit 20 selects the chart generation code by referring to the chart generation database DB2. The artificial intelligence unit 20 also outputs the chart setting information included in the chart generation code, in response to an instruction from the AI ​​control unit 13.

[0052] As described above, the configuration that generates chart reading problems by having the artificial intelligence unit 20 select and output a chart generation code allows for the accurate generation of chart reading problems that include charts. Furthermore, even if the problem statement is changed, it is easy to create charts corresponding to the new problem statement. In addition, it is possible to guarantee the quality of charts and generate charts with a consistent design. Moreover, the configuration that retrieves chart elements from the chart element database DB1 and draws them when outputting the problem allows for the generation of problems using the latest data stored in the chart element database DB1.

[0053] The communication processing unit 19 is a processing unit that enables the transmission and reception of data with the user terminal 30 via a network NW such as the Internet, in accordance with a predetermined protocol. For example, the communication processing unit 19 receives a problem statement from the user terminal 30. The communication processing unit 19 also transmits source code generated by the artificial intelligence unit 20, which displays a chart reading problem, to the user terminal 30.

[0054] ● Correct answer generation process Here, we will describe an example of the correct answer generation process performed by the problem generation system 1. For example, the AI ​​control unit 13 sends a correct answer output instruction to the artificial intelligence unit 20 to output the correct answer to a chart reading problem. The artificial intelligence unit 20 searches, for example, the internet or an appropriate database and outputs the correct answer to the problem statement. In this case, the artificial intelligence unit 20 has an appropriate trained model that takes a chart reading problem as input and outputs the correct answer.

[0055] Furthermore, the AI ​​control unit 13 may instruct the artificial intelligence unit 20 to refer to the statistical analysis database DB3 and calculate appropriate statistics using the source code stored in the statistical analysis database DB3 if the problem involves describing the characteristics of a figure or chart. The AI ​​control unit 13 may also instruct the artificial intelligence unit 20 to refer to the statistical analysis database DB3 if it is necessary to calculate appropriate statistics when answering a problem.

[0056] The statistical analysis database DB3 is a memory unit that stores multiple source code files (hereinafter also referred to as "analysis codes") that describe processing functions for calculating statistics for chart elements. For example, the statistical analysis database DB3 stores analysis codes that calculate statistics such as the mean, median, or variance for chart elements. The statistical analysis database DB3 also stores analysis codes that perform trend analysis on chart elements of time series data and extract trend information such as increasing or decreasing trends. Furthermore, the statistical analysis database DB3 stores analysis codes that calculate skewness, kurtosis, etc., when the chart elements are data showing a distribution. In addition, the statistical analysis database DB3 may also store analysis codes that perform comparative analysis. In analysis codes that perform comparative analysis, for example, when there is data from different perspectives obtained along the same axis, the difference or proportion between perspectives is calculated as a statistical measure. For example, if the chart elements are data obtained for each country, the comparative analysis calculates the results of comparison with other countries. Also, if the chart elements are data obtained for each period of each year, the comparative analysis calculates the year-on-year comparison.

[0057] The statistical analysis database DB3 may store analysis codes that calculate the derivative of a figure or table element when the element is time-series data, and calculate the degree of increase or decrease, or the presence, number, or location of extreme values. Furthermore, the statistical analysis database DB3 may store analysis codes that calculate the presence or strength of correlations between figure or table elements when the element contains data from multiple perspectives, such as analysis codes that calculate correlation functions.

[0058] The AI ​​control unit 13 may instruct the artificial intelligence unit 20 to determine the type of information to be graphed and to select an analysis code to execute according to the type of information to be graphed. In this case, the artificial intelligence unit 20 has a trained model that takes the problem statement, the information to be graphed, or graph elements as input and outputs the type of information to be graphed, and a trained model that takes the type of information to be graphed as input and outputs an analysis code to execute.

[0059] The instruction to select the analysis code may, for example, be an instruction to select the type of information to be graphed based on the information obtained by natural language processing analysis of the problem statement. In this case, the artificial intelligence unit 20 has a trained model that takes the information obtained by natural language processing analysis of the problem statement as input and outputs the type of information to be graphed.

[0060] Furthermore, the process of selecting an analysis code to be executed for a graph element whose type has been identified may, instead of being configured to be selected by a trained model, refer to information that has been pre-linked with the type of information to be graphed and the analysis code to be executed. That is, for example, the AI ​​control unit 13 may provide the artificial intelligence unit 20 with information that links the type of information to be graphed and the analysis code to be executed, and instruct the artificial intelligence unit 20 to apply the analysis code that has been linked to the obtained type of information to be graphed.

[0061] The instruction to select an analysis code may also be an instruction to select multiple analysis codes. The AI ​​control unit 13 may instruct the artificial intelligence unit 20 to extract a statistical quantity that represents the characteristics of the figure element from multiple statistical quantities obtained from each of the multiple analysis codes. For example, a threshold may be set in advance for each statistical quantity, and the unit may be instructed to determine that a feature is represented if the difference between the statistical quantity and the threshold is greater than or equal to a predetermined value, or the determination may be output by an appropriate pre-trained model. In this case, the artificial intelligence unit 20 has, for example, a pre-trained model that takes a figure element and a statistical quantity as input and outputs a statistical quantity that represents the characteristics of the figure element.

[0062] The AI ​​control unit 13 may send an instruction to the artificial intelligence unit 20 to generate a text describing the features of the chart elements based on statistical information representing the features. The AI ​​control unit 13 obtains the text describing the features of the chart elements from the artificial intelligence unit 20. The text describing the features of the chart elements is an example of a correct answer to a chart reading problem.

[0063] Furthermore, the memory control unit 12 stores the correct answers obtained through the acquisition process in association with the chart reading questions. The correct answers to the chart reading questions may be displayed on the user terminal 30 together with the chart reading questions. For example, the display of answers may be changed based on whether the user is a service administrator or a respondent to the chart reading questions. If the user is a service administrator, the correct answers to the chart reading questions are displayed, and if the user is a respondent to the chart reading questions, only the chart reading questions are displayed. Alternatively, on a display screen showing only the chart reading questions, input from the respondent may be accepted, and the correct answers may be displayed after input is received.

[0064] With this configuration, statistical quantities can be accurately calculated using analysis code that describes processing functions, while the artificial intelligence unit 20 generates natural-sounding responses in text form.

[0065] ● Processing flow for generating chart and graph reading problems As shown in Figure 4, first, the user terminal 30 displays an input screen that accepts input of a problem statement related to the diagram reading problem to be generated, and accepts the input of the problem statement (step S101), and transmits it to the server device 10. Note that the problem statement may be data that the server device 10 has in its possession beforehand.

[0066] The server device 10 sends an extraction instruction to the artificial intelligence unit 20 to extract information to be turned into a diagram from the problem statement (step S102).

[0067] Next, the server device 10 issues an acquisition instruction to the artificial intelligence unit 20 to obtain the chart elements that constitute the information to be charted from the chart element database DB1 (step S103).

[0068] Next, the server device 10 issues a selection instruction to the artificial intelligence unit 20 to select a chart generation code to apply to a chart element from the chart generation database DB2 (step S104). Then, the server device 10 refers to the chart generation database DB2 in accordance with the selection instruction and generates the source code for the problem, which includes the problem statement and the chart generation code (step S105).

[0069] Next, the server device 10 issues a correct answer generation instruction to the artificial intelligence unit 20 to create the correct answer to the generated chart reading problem (step S106). At this time, the generation instruction may include an instruction to refer to the statistical analysis database DB3, and the artificial intelligence unit 20 refers to the statistical analysis database DB3 as appropriate. The server device 10 retrieves the created correct answer and stores it together with the source code of the chart reading problem (step S107). Then the process moves to step S108, and the generated problem is sent to the user terminal 30. Note that the above process is just one example, and step S106, which generates the correct answer to the chart reading problem, may be omitted.

[0070] ●Example of display Figure 3 shows an example of the generation screen G100 for a chart reading problem displayed on the user terminal 30. As shown in Figure 3, the generation screen G100 includes, for example, a problem statement input area G110, a problem preview area G120, and a chart adjustment area G130. The problem statement input area G110 is an area that accepts the user's input of the problem statement. The problem statement input area G110 also accepts the input of tags that indicate the position where charts are inserted into the problem statement. In the example shown in the figure, the text "[Chart Insertion Point]" is an example of the above tag.

[0071] The problem preview area G120 is the area where the problem statement and the resulting figure / table reading problem are displayed after executing the source code that generates the figure / table. In the problem preview area G120, for example, the problem statement is displayed in the same way as in the problem statement input area G110, but a certain rectangular area is displayed where "[Figure / Table Insertion Point]" was entered in the problem statement input area G110, and the message "Automatically generated figures / tables will be displayed here" is shown. Note that the automatically generated figures / tables may also be displayed in this rectangular area.

[0072] The chart adjustment area G130 is the area where user specifications regarding chart generation are accepted. For example, the chart type selection area G131, the data source type selection area G132, and the design template selection area G133 are displayed in the chart adjustment area G130.

[0073] The chart type selection area G131 is an area where the user selects the method of representing the chart elements as charts, and for example, a pull-down box is displayed. In this figure, a line graph is selected. The chart generation code applied to the chart elements is changed according to the method selected in the chart type selection area G131. With this configuration, even if the chart type selected by the artificial intelligence unit 20 does not meet the user's expectations, the user can manually change it to the desired chart, making it highly convenient.

[0074] The data source type selection area G132 is an area that accepts the selection of the type of chart element to be used as the data source. In this figure, the type of chart element is set to automatically generated data. The type of chart element includes specifying the source from which to acquire the chart element, and whether or not to perform calculation processing on the extracted chart element. For example, if automatically generated data is selected, the chart element is acquired from the predetermined chart element database DB1, and appropriate calculation processing is performed as needed to output the chart reading problem.

[0075] The design template selection area G133 is an area where users can select a template to be reflected in the design of a chart or graph. A design template is a set of information that includes various display characteristics such as color scheme, font, font size, or line type, and multiple design templates are pre-stored in the server device 10. Each design template is given a name, and by selecting the name of a design template in the design template selection area G133, it can be applied to the chart or graph.

[0076] When the selection mode is changed in the chart adjustment area G130, the chart displayed in the problem preview area G120 is immediately rendered.

[0077] As described above, the problem generation system according to the present invention makes it possible to easily generate problems that include diagrams and charts.

[0078] ●Problem Generation System (2)● Here, we will describe a different aspect of the problem generation system 1 according to an embodiment of the present invention from the embodiment described earlier. In particular, here we will describe an aspect in which the problem is processed by an appropriate functional unit of the server device 10, instead of the aspect in the previously described embodiment in which instructions were sent to the artificial intelligence unit 20 and the results were obtained.

[0079] The process of extracting information to be graphed or charted may be performed by an appropriate extraction processing unit in the server device 10, instead of the configuration instructed to the artificial intelligence unit 20. This extraction processing unit, for example, performs natural language processing (NLP) analysis on the problem statement and extracts information to be graphed or charted using an appropriate method.

[0080] Furthermore, the process of obtaining the chart elements that constitute the information to be charted from the chart element database DB1 may be performed by an appropriate acquisition processing unit on the server device 10. This acquisition processing unit may, for example, search the chart element database DB1 using a rule-based search function and obtain the target chart elements.

[0081] Although the present invention has been described above using embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of its gist. [Explanation of Symbols]

[0082] 1. Problem generation system 10. Server device (problem generation device) 11 Display Control Unit 12 Memory Control Unit 13 AI control section 14 Output Control Unit 19 Communication Processing Unit 20 Artificial Intelligence Department DB1 Chart Element Database DB2 Chart Generation Database DB3 Statistical Analysis Database 30 User terminals

Claims

1. A problem generation device that generates diagrams and charts included in the problem, A processor is provided, and the processor is In the extraction instruction step, an extraction instruction is given to the artificial intelligence to extract the information to be turned into a diagram or chart that is cited in the problem statement. In the acquisition instruction step, an acquisition instruction is given to the artificial intelligence to acquire the chart elements constituting the charted information from a predetermined database. In the selection instruction step, a storage unit that stores multiple chart generation codes that output predetermined chart elements as charts or tables in different forms is referenced, and a selection instruction is given to the artificial intelligence to select the chart generation code to apply to the chart elements. In the output step, source code including the diagram generation code is output. The artificial intelligence comprises a first trained model that takes the problem statement as input and outputs the information to be graphed; a second trained model that takes the information to be graphed as input and outputs the graph elements that constitute the information to be graphed; and a third trained model that takes the problem statement or information obtained by analyzing the problem statement as input and outputs the graph generation code to be applied to the graph elements. Problem generator.

2. Further, a setting information addition step is performed in which the setting information of the generated chart is determined and an instruction is given to the artificial intelligence to include in the chart generation code, The artificial intelligence further comprises a fourth trained model that takes the problem statement or information obtained by analyzing the problem statement using natural language processing as input and outputs the setting information. The problem generation device according to claim 1.

3. The setting information of the figure includes at least axis elements, units for each axis, axis labels, legend or scale, The problem generation apparatus according to claim 2.

4. The correct answer generation step is further performed by giving the artificial intelligence a correct answer generation instruction to create the correct answer to the problem, In the output step, the correct answer to the problem is further output. The artificial intelligence further comprises a fifth trained model that takes the problem as input and outputs the correct answer. The problem generation device according to claim 1.

5. The correct answer is a sentence that describes the characteristics of the diagram element. The problem generation apparatus according to claim 4.

6. In the correct answer generation step, A statistical calculation step involves referring to a second storage unit that stores multiple analysis codes, each containing a processing function for calculating statistical quantities for the aforementioned chart elements, and issuing a statistical calculation instruction to the artificial intelligence to calculate the statistical quantities using the analysis codes. An analysis code selection step involves determining the type of information to be graphed and issuing an analysis code selection instruction to the artificial intelligence, instructing it to select the analysis code to be executed according to the type of information to be graphed; A statistical extraction step in which the artificial intelligence is instructed to extract statistical quantities that represent the characteristics of the chart elements from the aforementioned statistical quantities, A text generation step involves giving text generation instructions to the artificial intelligence to generate text that explains the characteristics of the chart elements based on the information of the aforementioned statistics, conduct, The problem generation apparatus according to claim 4.

7. In the input step, the input of tags indicating the insertion position of figures and tables in the problem statement is accepted. In the output step, the source code in which the chart generation code is placed at the position where the tag is entered is output. The problem generation device according to claim 1.

8. A problem generation method for generating diagrams and charts included in the problem, By computer, An extraction instruction step involves giving an extraction instruction to the artificial intelligence to extract the information to be turned into a diagram or chart that is cited in the problem statement, A step of issuing an acquisition instruction to artificial intelligence to acquire chart elements constituting the information to be charted from a predetermined database, A selection instruction step involves referring to a storage unit that stores multiple diagram generation codes that output predetermined diagram elements as diagrams or tables in different forms, and giving a selection instruction to artificial intelligence to select the diagram generation code to apply to the diagram elements. An output step which outputs source code including the aforementioned diagram generation code, Execute, The artificial intelligence comprises a first trained model that takes the problem statement as input and outputs the information to be graphed; a second trained model that takes the information to be graphed as input and outputs the graph elements that constitute the information to be graphed; and a third trained model that takes the problem statement or information obtained by analyzing the problem statement as input and outputs the graph generation code to be applied to the graph elements. Problem generation method.

9. A problem generation program that generates diagrams and charts included in the problem, On the computer, An extraction instruction step involves giving an extraction instruction to the artificial intelligence to extract the information to be turned into a diagram or chart that is cited in the problem statement, A step of issuing an acquisition instruction to artificial intelligence to acquire chart elements constituting the information to be charted from a predetermined database, A selection instruction step involves referring to a storage unit that stores multiple diagram generation codes that output predetermined diagram elements as diagrams or tables in different forms, and giving a selection instruction to artificial intelligence to select the diagram generation code to apply to the diagram elements. An output step which outputs source code including the aforementioned diagram generation code, Make it run, The artificial intelligence comprises a first trained model that takes the problem statement as input and outputs the information to be graphed; a second trained model that takes the information to be graphed as input and outputs the graph elements that constitute the information to be graphed; and a third trained model that takes the problem statement or information obtained by analyzing the problem statement as input and outputs the graph generation code to be applied to the graph elements. Problem generation program.

Images