System

Abstract

To provide a system according to an embodiment that efficiently automatically corrects an answer in long sentences and an essay.SOLUTION: A system according to an embodiment comprises a reading unit, a summarization unit, a similarity calculation unit, and a providing unit. The reading unit reads an answer written by a student. The summarization unit summarizes the answer read by the reading unit. The similarity calculation unit calculates the similarity between a summary generated by the summarization unit and a model answer. The providing unit quantifies the similarity calculated by the similarity calculation unit in terms of points and provides a result of scoring.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including a description and related instruction sentence regarding the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-180282 A Summary of the Invention [Problem to be solved by the invention]

[0004] With conventional technology, it was difficult to automatically correct long answer sheets or essays, and there was a problem that efficient grading was not possible.

[0005] The system according to the embodiment aims to efficiently and automatically correct long answers and essays. [Means for solving the problem]

[0006] The system according to the embodiment includes a reading unit, a summarizing unit, a similarity calculation unit, and a providing unit. The reading unit reads answers written by students. The summarizing unit summarizes the answers read by the reading unit. The similarity calculation unit calculates the similarity between the summary generated by the summarizing unit and the model answer. The providing unit converts the similarity calculated by the similarity calculation unit into a score and provides the scoring result. Effect of the Invention

[0007] The system according to the embodiment can automatically and efficiently correct long answers and essays. [Brief description of the drawings]

[0008] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Diagram 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. FIG. [Diagram 3] FIG. 11 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Diagram 5] FIG. 13 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 13 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 13 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0010] First, the terms used in the following description will be explained.

[0011] In the following embodiments, a signed processor (hereinafter simply referred to as a "processor") may be one arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be one type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), or a TPU (Tensor Processing Unit).

[0012] In the following embodiments, a signed RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by the processor.

[0013] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0014] In the following embodiments, a communication I / F (Interface) with a code is an interface including a communication processor and an antenna. The communication I / F controls communication between multiple computers. An example of a communication standard applied to the communication I / F is a wireless communication standard including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0015] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. In addition, in this specification, the same idea as "A and / or B" is also applied when three or more things are expressed by connecting them with "and / or."

[0016] [First embodiment] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0017] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0018] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0019] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0020] The reception device 38 includes a touch panel 38A and a microphone 38B, and receives user input. The touch panel 38A detects contact with a pointer (e.g., a pen or a finger) to receive user input by the touch of the pointer. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 (see FIG. 2) acquires the data indicating the user input.

[0021] Output device 40 includes a display 40A and a speaker 40B, and presents data to a user by outputting the data in a form of expression that the user can perceive (e.g., voice and / or text). Display 40A displays visible information such as text and images according to instructions from processor 46. Speaker 40B outputs voice according to instructions from processor 46. Camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a Complementary Metal-Oxide-Semiconductor (CMOS) image sensor or a Charge Coupled Device (CCD) image sensor.

[0022] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for transmitting and receiving various types of information between the processor 46 and the processor 28 via the network 54.

[0023] FIG. 2 shows an example of main functions of the data processing device 12 and the smart device 14.

[0024] As shown in Fig. 2, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32. The specific process program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific process program 56 from the storage 32, and executes the read specific process program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific process program 56 executed on the RAM 30.

[0025] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the emotion of the user using the emotion identification model 59 and perform identification processing using the user's emotion. In the emotion estimation function (emotion identification function) using the emotion identification model 59, various estimations and predictions regarding the user's emotion, including estimation and prediction of the user's emotion, are performed, but are not limited to such examples. In addition, the estimation and prediction of the emotion also include, for example, analysis of the emotion.

[0026] In the smart device 14, the specific processing is performed by the processor 46. The storage 50 stores a specific processing program 60. The specific processing program 60 is used together with the specific processing program 56 by the data processing system 10. The processor 46 reads the specific processing program 60 from the storage 50 and executes the specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as the control unit 46A in accordance with the specific processing program 60 executed on the RAM 48. Note that the smart device 14 has a data generation model and an emotion identification model similar to the data generation model 58 and the emotion identification model 59, and can also perform processing similar to that of the specific processing unit 290 using these models.

[0027] Note that a device other than the data processing device 12 may have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains a processing result (prediction result, etc.) using the data generation model 58 by communicating with the server device having the data generation model 58. Also, the data processing device 12 may be a server device, or may be a terminal device (e.g., a mobile phone, a robot, a home appliance, etc.) owned by a user. Next, an example of processing by the data processing system 10 according to the first embodiment will be described.

[0028] (Example 1) The automatic correction system according to the embodiment of the present invention is a system that automatically reads an answer written by a student, has a generation AI summarize the answer, calculates the similarity to the model answer, and gives a score. This automatic correction system reads an answer written by a student, and a generation AI summarizes the answer. The generation AI analyzes the content of the answer, extracts important points, and creates a summary. For example, a long answer written by a student is summarized in a short form to clarify the main points. This summary is used to calculate the similarity to the model answer. Next, the generation AI calculates the similarity to the model answer. The generation AI compares the summarized answer with the model answer and calculates the similarity. For example, the generation AI evaluates the word agreement and sentence structure similarity between the summary of the answer and the model answer. This similarity is scored and provided as a scoring result. Furthermore, a similar process is applied to essay tests. An essay answer written by a student is read, and a generation AI creates a summary. Next, the generation AI calculates the similarity to the model answer and gives a score. This enables automatic correction of essays. This system reduces the burden on teachers and allows students' answers to be evaluated quickly and accurately. For example, teachers no longer need to manually correct a large number of answers, saving time and effort. In addition, students can receive quick feedback and quickly identify and improve their weak points. This allows the automatic correction system to automatically correct students' answers and provide scoring results.

[0029] The automatic correction system according to the embodiment includes a reading unit, a summarizing unit, a similarity calculation unit, and a providing unit. The reading unit reads an answer written by a student. The answer written by a student includes, for example, a handwritten answer and an answer in digital format, but is not limited to such examples. For example, the reading unit digitizes and reads the handwritten answer using scanning technology. The reading unit can also directly read an answer submitted in digital format. Furthermore, the reading unit can also read a printed answer using OCR technology. For example, the reading unit scans a handwritten answer with a high-resolution scanner and converts it into character information using OCR technology. For digital format answers, those submitted in a specific file format can be directly read. OCR technology recognizes printed characters with high accuracy and converts them into digital text. The summarizing unit summarizes the answer read by the reading unit using a generation AI. The summarization is performed based on, for example, the length of the sentence and the importance of the information to be summarized, but is not limited to such examples. For example, the generation AI uses a text generation AI (e.g., LLM) to concisely summarize the answer. The summarization unit can also use a multimodal generation AI to summarize the contents of the answer. The summarization unit can also use the generation AI to extract and summarize important parts of a sentence. For example, the text generation AI has learned a large amount of text data and has advanced natural language processing capabilities. The multimodal generation AI can handle multiple modalities such as images and audio as well as text. The generation AI uses keyword extraction technology to pick up particularly important information in the answer and summarizes it based on that. The similarity calculation unit calculates the similarity between the summary generated by the summarization unit and the model answer. The similarity is calculated by a method such as cosine similarity or Jaccard coefficient, but is not limited to such examples. The similarity calculation unit calculates the similarity between the answer and the model answer using, for example, cosine similarity. The similarity calculation unit can also calculate the similarity using the Jaccard coefficient. The similarity calculation unit can also calculate the similarity using TF-IDF.For example, the similarity calculation unit calculates the angle between two vectors in a vector space model as the cosine similarity, and determines that the smaller the angle, the higher the similarity. The Jaccard coefficient is a value obtained by dividing the size of the common part of two sets by the total size, and indicates the degree of overlap of the sets. The TF-IDF is an index for evaluating the importance of words in a document, and is used to calculate the similarity between the answer sheet and the model answer. A part or all of the above-mentioned processing in the similarity calculation unit may be performed using, for example, AI, or may be performed without using AI. For example, the similarity calculation unit can calculate the similarity using an AI model that inputs the summary generated by the summarization unit and the model answer and outputs the similarity. The providing unit calculates a score based on the similarity calculated by the similarity calculation unit. The score is calculated, for example, according to a scoring method or evaluation criteria based on the similarity. The providing unit assigns a score in the range of 0 to 100 points based on the similarity. The providing unit can also assign an evaluation from A to F based on the similarity. The providing unit can also calculate percentiles based on the similarity. For example, the providing unit performs scaling so that the higher the similarity, the higher the score. The providing unit performs evaluation, for example, with a similarity of 90% or more as A and a similarity of 80% or more as B. The percentile is an index indicating a relative position in the whole, and for example, the providing unit gives a high score to an answer that is in the top 10%. In this way, the automatic correction system according to the embodiment can automatically correct the student's answer and provide the grading result. A part or all of the above-mentioned processing in the providing unit may be performed using, for example, AI, or may be performed without using AI. For example, the providing unit can calculate the score using an AI model that inputs the similarity calculated by the similarity calculation unit and outputs the score. For example, the output unit displays the grading result to the student or teacher through a web application or a mobile application. If feedback on paper is desired, the result is printed using a printer. Sending by email provides quick feedback by sending the result directly to the student or guardian. Some or all of the above-described processing in the output section may be performed using AI, for example, or may be performed without using AI.

[0030] The reading unit reads the answer sheet written by the student. The answer sheet written by the student includes, for example, a handwritten answer sheet, an answer sheet in digital format, and the like, but is not limited to such examples. For example, the reading unit digitizes and reads the handwritten answer sheet using scanning technology. The reading unit can also directly read an answer sheet submitted in digital format. Furthermore, the reading unit can also read a printed answer sheet using OCR technology. For example, the reading unit scans a handwritten answer sheet with a high-resolution scanner and converts it into character information using OCR technology. Digital answers submitted in a specific file format can be directly read. OCR technology recognizes printed characters with high accuracy and converts them into digital text. By combining these technologies, the reading unit can efficiently digitize answers in various formats and import them into the system. For example, a handwritten answer sheet is scanned by a scanner and then converted into character information using OCR technology. At this time, the OCR technology learns a large amount of handwritten data in advance to improve the recognition accuracy of handwritten characters. Digital answer sheets are often submitted in formats such as PDF or Word files, which are directly imported into the system. In addition, the reader also has the function of performing error checks when reading answer sheets, detecting and correcting incomplete data and reading errors. For example, it corrects noise and distortions that occur during scanning to obtain accurate character information. In addition, OCR technology is compatible with different fonts and character sizes, and can recognize character information on printed answer sheets with high accuracy. This allows the reader to quickly and accurately digitize answer sheets in various formats and provide the data required for subsequent processing.

[0031] The summarizing unit uses the generation AI to summarize the answer read by the reading unit. The summarization is performed, for example, based on the length of the sentence and the importance of the information to be summarized, but is not limited to such examples. For example, the generation AI uses a text generation AI (for example, LLM) to briefly summarize the answer. The summarizing unit can also summarize the contents of the answer using a multimodal generation AI. The summarizing unit can also extract and summarize important parts of the sentence using the generation AI. For example, the text generation AI has learned a large amount of text data and has advanced natural language processing capabilities. The multimodal generation AI can handle multiple modals such as images and audio as well as text. The generation AI uses keyword extraction technology to pick up particularly important information in the answer and summarize it based on that. The summarizing unit has a large amount of answer data learned in advance to improve the accuracy of the summarization performed by the generation AI. For example, a model for generating accurate summaries is trained using past answer data. The summarizing unit also has a function of evaluating the summary generated by the generation AI and making corrections as necessary. For example, if the content of the summary is inappropriate or important information is missing, the summarization unit automatically corrects it. The summarization unit also continuously evaluates the accuracy of the summary and updates the generation AI model to improve the quality of the summary. This allows the summarization unit to quickly and accurately summarize the answer sheet read by the reading unit and provide the information necessary for subsequent processing.

[0032] The similarity calculation unit calculates the similarity between the summary generated by the summarization unit and the model answer. The similarity is calculated by, for example, a method such as cosine similarity or Jaccard coefficient, but is not limited to such examples. The similarity calculation unit calculates the similarity between the answer and the model answer using, for example, cosine similarity. The similarity calculation unit can also calculate the similarity using the Jaccard coefficient. The similarity calculation unit can also calculate the similarity using TF-IDF. For example, the similarity calculation unit calculates the angle between two vectors in a vector space model as cosine similarity, and determines that the smaller the angle, the higher the similarity. The Jaccard coefficient is a value obtained by dividing the size of the common part of two sets by the total size, and indicates the degree of overlap of the sets. The TF-IDF is an index that evaluates the importance of words in a document, and is used to calculate the similarity between the answer and the model answer. A part or all of the above-mentioned processing in the similarity calculation unit may be performed using, for example, AI, or may be performed without using AI. For example, the similarity calculation unit can calculate the similarity using an AI model that inputs the summary generated by the summarization unit and the model answer and outputs the similarity. The similarity calculation unit can perform a more accurate similarity evaluation by combining these calculation methods. For example, by using cosine similarity and Jaccard coefficient together, the degree of agreement and the similarity of structure between the content of the answer and the model answer are evaluated simultaneously. In addition, by using TF-IDF, the degree of agreement of important keywords is evaluated, and the extent to which the content of the answer matches the model answer is analyzed in detail. As a result, the similarity calculation unit can calculate the similarity between the summary generated by the summarization unit and the model answer with high accuracy and provide information necessary for subsequent processing.

[0033] The providing unit calculates a score based on the similarity calculated by the similarity calculation unit. The score is calculated, for example, according to a scoring method or evaluation criteria based on the similarity. The providing unit, for example, gives a score in the range of 0 to 100 points based on the similarity. The providing unit can also give an evaluation from A to F based on the similarity. The providing unit can also calculate a percentile based on the similarity. For example, the providing unit performs scaling so that the higher the similarity, the higher the score. The providing unit performs evaluation, for example, A for a similarity of 90% or more, B for 80% or more, and so on. The percentile is an index indicating a relative position in the whole, and for example, the providing unit gives a high evaluation to an answer that is in the top 10%. In this way, the automatic correction system according to the embodiment can automatically correct the student's answer and provide the grading result. A part or all of the above-mentioned processing in the providing unit may be performed, for example, using AI, or may be performed without using AI. For example, the providing unit can calculate the score using an AI model that inputs the similarity calculated by the similarity calculation unit and outputs the score. For example, the output unit displays the grading results to students and teachers through a web application or a mobile application. If feedback on paper is desired, the results are printed using a printer. Sending by email provides quick feedback by sending the results directly to students and parents. Some or all of the above-mentioned processing in the output unit may be performed using AI, for example, or may be performed without using AI. The providing unit can generate not only the calculation of the score but also the content of the feedback. For example, it provides feedback that specifically indicates which parts of the answer match the model answer and which parts have room for improvement. This allows students to understand the strengths and weaknesses of their answers and use them to create their next answer. Furthermore, the providing unit can also customize the content of the feedback individually. For example, it proposes individual advice and study plans based on the student's past grades and learning history. This allows the providing unit to provide optimal feedback to each student and maximize the learning effect.

[0034] The summarizing unit can analyze the contents of the answer sheet, extract key points, and create a summary. The summarizing unit can, for example, use a generating AI to analyze the contents of the answer sheet, extract key points, and create a summary. For example, the generating AI can pick up particularly important information in the answer sheet and create a summary based on that. The summarizing unit can also use a generating AI to extract and summarize important parts of a sentence. For example, the generating AI can use a keyword extraction technique to pick up particularly important information in the answer sheet and create a summary based on that. This improves the accuracy of the summary by extracting and summarizing important points of the answer sheet. Some or all of the above-mentioned processing in the summarizing unit can be performed, for example, using a generating AI, or can be performed without using a generating AI. For example, the summarizing unit can analyze the contents of the answer sheet, extract key points, and create a summary using an algorithm.

[0035] The similarity calculation unit can compare the answer and the model answer and calculate the similarity. The similarity calculation unit can, for example, use a generation AI to compare the answer and the model answer and calculate the similarity. For example, the generation AI evaluates the word agreement and the similarity of the sentence structure between the summary of the answer and the model answer. The similarity calculation unit can also use the generation AI to calculate the similarity using a method such as cosine similarity or Jaccard coefficient. For example, the generation AI calculates the similarity between the answer and the model answer using cosine similarity. As a result, the accuracy of the similarity calculation is improved by comparing the summarized answer and the model answer. A part or all of the above-mentioned processing in the similarity calculation unit may be performed using, for example, a generation AI, or may be performed without using a generation AI. For example, the similarity calculation unit can compare the answer and the model answer and calculate the similarity using an algorithm for calculating the similarity.

[0036] The providing unit can score the similarity and provide the scoring result. The providing unit can score the similarity using, for example, a generating AI and provide the scoring result. For example, the generating AI calculates the score according to a scoring method or evaluation criteria based on the similarity. The providing unit can also use the generating AI to give a score in the range of 0 to 100 points based on the similarity. This improves the efficiency of scoring by scoring the similarity and providing the scoring result. A part or all of the above-mentioned processing in the providing unit can be performed using, for example, a generating AI, or can be performed without using a generating AI. For example, the providing unit can calculate the score using an algorithm for scoring the similarity and providing the scoring result.

[0037] The automatic correction system according to the embodiment further includes a training data collection unit. The training data collection unit collects data for improving the accuracy of the summarization section. For example, the training data collection unit collects past answer data and model answer data, and uses them as training data for the summarization section. The training data collection unit can also use a generation AI to collect data for improving the accuracy of the summarization section. For example, the generation AI analyzes past answer data and selects data to be used as training data for the summarization section. As a result, the accuracy of the summarization section is improved by collecting training data. Some or all of the above-mentioned processing in the training data collection unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the training data collection unit collects past answer data, inputs it to the generation AI, and allows the generation AI to select data to be used as training data for the summarization section.

[0038] The automatic correction system according to the embodiment further includes a feedback unit. The feedback unit provides feedback to the student based on the grading result. For example, the feedback unit uses a generation AI to analyze the grading result and provide the student with specific improvements and advice. The feedback unit can also use the generation AI to identify weak points in the student's answer and provide feedback based on the weak points. For example, the generation AI analyzes the content of the answer, extracts points that need improvement, and generates feedback based on the weak points. As a result, the student's learning effect is improved by providing feedback based on the grading result. A part or all of the above-mentioned processing in the feedback unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the feedback unit can generate feedback using an algorithm for analyzing the grading result and providing the student with specific improvements and advice.

[0039] The reading unit can analyze the student's past answer submission history and select the optimal reading method. The reading unit can, for example, use the generation AI to analyze the student's past answer submission history. For example, the generation AI can analyze the format and content of the answer submitted by the student in the past and select the optimal reading method. The reading unit can also adjust the reading speed based on the student's past answer submission history using the generation AI. For example, the generation AI can adjust the reading speed based on the amount of answers submitted by the student in the past. In this way, the optimal reading method can be selected by analyzing the past answer submission history. Part or all of the above-mentioned processing in the reading unit can be performed using, for example, the generation AI, or can be performed without using the generation AI. For example, the reading unit can input the student's past answer submission history into the generation AI and cause the generation AI to select the optimal reading method.

[0040] The reading unit can filter the answer sheets based on the student's current learning situation and field of interest when reading the answer sheets. The reading unit, for example, uses the generation AI to analyze the student's current learning situation and field of interest and performs filtering. For example, the generation AI preferentially reads answer sheets related to the unit the student is currently studying. The reading unit can also use the generation AI to filter and read answer sheets related to the student's field of interest. For example, the generation AI extracts keywords related to the student's field of interest and filters the answer sheets based on the keywords. This makes it possible to read more relevant answer sheets by filtering based on the student's learning situation and field of interest. Part or all of the above-mentioned processing in the reading unit may be performed using, for example, the generation AI, or may be performed without using the generation AI. For example, the reading unit can input the student's current learning situation and field of interest into the generation AI and have the generation AI execute filtering.

[0041] When reading the answer sheet, the reading unit can preferentially read highly relevant answer sheets by taking into consideration the geographical location information of the student. The reading unit, for example, uses the generation AI to analyze the geographical location information of the student and preferentially reads highly relevant answer sheets. For example, if the student lives in a specific area, the generation AI preferentially reads answer sheets related to that area. In addition, the reading unit can also use the generation AI to preferentially read answer sheets related to the curriculum of a specific school if the student attends that school. For example, if the student has a specific cultural background, the generation AI preferentially reads answer sheets related to that culture. In this way, by taking into consideration the geographical location information of the student, highly relevant answer sheets can be preferentially read. Part or all of the above-mentioned processing in the reading unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the reading unit can input the geographical location information of the student to the generation AI and cause the generation AI to select highly relevant answer sheets.

[0042] The reading unit can analyze the student's social media activity and read the related answer when reading the answer. The reading unit can, for example, use the generation AI to analyze the student's social media activity. For example, the generation AI reads the answer related to a topic in which the student is interested on social media. The reading unit can also use the generation AI to read the answer related to an account that the student follows on social media. For example, the generation AI reads the answer related to the content shared by the student on social media. In this way, the related answer can be read by analyzing the student's social media activity. Some or all of the above-mentioned processing in the reading unit can be performed, for example, using the generation AI or can be performed without using the generation AI. For example, the reading unit can input the student's social media activity into the generation AI and have the generation AI select the related answer.

[0043] The summarization unit can adjust the level of detail of the summary based on the importance of the answer when generating a summary. The summarization unit, for example, uses a generation AI to evaluate the importance of the answer and adjust the level of detail of the summary based on the evaluation. For example, the generation AI picks up particularly important information in the answer and generates a detailed summary based on the information. The summarization unit can also adjust the length and content of the summary based on the importance of the answer using the generation AI. For example, the generation AI uses a keyword extraction technique to pick up particularly important information in the answer and adjusts the level of detail of the summary based on the information. This allows a more appropriate summary to be generated by adjusting the level of detail of the summary based on the importance of the answer. Some or all of the above-mentioned processing in the summarization unit may be performed using, for example, a generation AI, or may be performed without using a generation AI. For example, the summarization unit can generate a summary using an algorithm for evaluating the importance of the answer and adjusting the level of detail of the summary based on the evaluation.

[0044] The summarization unit can apply different summarization algorithms depending on the category of the answer when generating a summary. The summarization unit, for example, uses a generation AI to classify the category of the answer and apply different summarization algorithms accordingly. For example, the generation AI applies a logical summarization algorithm in the case of an essay, and applies an emotional summarization algorithm in the case of a narrative text. The summarization unit can also use the generation AI to adjust the content and format of the summary depending on the category of the answer. For example, the generation AI applies an information-oriented summarization algorithm in the case of an explanatory text. This makes it possible to generate a more appropriate summary by applying different summarization algorithms depending on the category of the answer. Some or all of the above-mentioned processing in the summarization unit may be performed, for example, using a generation AI, or may be performed without using a generation AI. For example, the summarization unit can generate a summary using an algorithm for classifying the category of the answer and applying different summarization algorithms accordingly.

[0045] The summarization unit can determine the priority of summaries based on the time of submission of answers when generating summaries. The summarization unit, for example, uses a generation AI to evaluate the time of submission of answers and determines the priority of summaries based on the evaluation. For example, the generation AI prioritizes summarization of answers submitted just before the deadline. The summarization unit can also use the generation AI to summarize answers submitted early next. For example, the generation AI determines the priority of summaries based on importance regardless of the time of submission. In this way, a more appropriate summary can be generated by determining the priority of summaries based on the time of submission of answers. Some or all of the above-mentioned processing in the summarization unit may be performed, for example, using a generation AI or may be performed without using a generation AI. For example, the summarization unit can generate summaries using an algorithm for evaluating the time of submission of answers and determining the priority of summaries based on the evaluation.

[0046] The summarization unit can adjust the order of summaries based on the relevance of the answers when generating summaries. The summarization unit, for example, uses a generation AI to evaluate the relevance of the answers and adjust the order of summaries based on the evaluation. For example, the generation AI prioritizes summarization of answers related to important topics. The summarization unit can also use the generation AI to summarize answers related to normal topics next. For example, the generation AI summarizes answers related to less relevant topics last. This allows a more appropriate summary to be generated by adjusting the order of summaries based on the relevance of the answers. Some or all of the above-mentioned processing in the summarization unit may be performed, for example, using a generation AI, or may be performed without using a generation AI. For example, the summarization unit can generate a summary using an algorithm for evaluating the relevance of answers and adjusting the order of summaries based on the evaluation.

[0047] The similarity calculation unit can improve the accuracy of the similarity calculation by considering the interrelationships of the answers when calculating the similarity. The similarity calculation unit, for example, uses a generation AI to evaluate the interrelationships of the answers and improves the accuracy of the similarity calculation based on the evaluation. For example, the generation AI calculates the similarity by considering the interrelationships of multiple answers. The similarity calculation unit can also use the generation AI to evaluate the relevance of the contents of the answers and improve the accuracy of the similarity calculation based on the evaluation. For example, the generation AI calculates the similarity by considering the context of the answers. As a result, the accuracy of the similarity calculation is improved by considering the interrelationships of the answers. A part or all of the above-mentioned processing in the similarity calculation unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the similarity calculation unit can calculate the similarity using an algorithm for evaluating the interrelationships of the answers and calculating the similarity based on the evaluation.

[0048] The similarity calculation unit can perform the similarity calculation while taking into account the attribute information of the person who submitted the answer sheet when calculating the similarity. The similarity calculation unit, for example, uses a generation AI to evaluate the attribute information of the person who submitted the answer sheet and performs the similarity calculation based on the attribute information. For example, the generation AI adjusts the criterion for the similarity calculation based on the grade of the person who submitted the answer sheet. The similarity calculation unit can also adjust the criterion for the similarity calculation based on the learning history of the person who submitted the answer sheet using the generation AI. For example, the generation AI adjusts the criterion for the similarity calculation based on the learning style of the person who submitted the answer sheet. This improves the accuracy of the similarity calculation by taking into account the attribute information of the person who submitted the answer sheet. A part or all of the above-mentioned processing in the similarity calculation unit may be performed using, for example, a generation AI, or may be performed without using a generation AI. For example, the similarity calculation unit can calculate the similarity using an algorithm for evaluating the attribute information of the person who submitted the answer sheet and performing the similarity calculation based on the evaluation.

[0049] The similarity calculation unit can perform the similarity calculation taking into account the geographical distribution of the answers when calculating the similarity. The similarity calculation unit, for example, uses the generation AI to evaluate the geographical distribution of the answers and performs the similarity calculation based on the evaluation. For example, the generation AI preferentially calculates the similarity of answers in geographically close areas. The similarity calculation unit can also use the generation AI to calculate the similarity of answers in geographically distant areas next. For example, the generation AI performs the similarity calculation based on the importance regardless of the geographical distribution. As a result, the accuracy of the similarity calculation is improved by taking the geographical distribution into consideration. A part or all of the above-mentioned processing in the similarity calculation unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the similarity calculation unit can calculate the similarity using an algorithm for evaluating the geographical distribution of the answers and performing the similarity calculation based on the evaluation.

[0050] The similarity calculation unit can improve the accuracy of the similarity calculation by referring to the related literature of the answer when calculating the similarity. The similarity calculation unit, for example, uses the generation AI to evaluate the related literature of the answer and improves the accuracy of the similarity calculation based on the evaluation. For example, the generation AI evaluates the contents of the answer by referring to the related literature and improves the accuracy of the similarity calculation. The similarity calculation unit can also improve the accuracy of the similarity calculation by using the generation AI and taking into account the citation of the related literature. For example, the generation AI extracts the main points of the related literature and improves the accuracy of the similarity calculation based on the extracted main points. As a result, the accuracy of the similarity calculation is improved by referring to the related literature. A part or all of the above-mentioned processing in the similarity calculation unit may be performed, for example, using the generation AI or may be performed without using the generation AI. For example, the similarity calculation unit can calculate the similarity by using an algorithm for evaluating the related literature of the answer and performing the similarity calculation based on the evaluation.

[0051] When providing the grading results, the providing unit can select the optimal display method by referring to the student's past grade history. The providing unit, for example, uses the generating AI to evaluate the student's past grade history and selects the optimal display method based on the evaluation. For example, the generating AI selects the optimal display method based on the student's past grade history. The providing unit can also use the generating AI to analyze the student's grade trends and provide an appropriate display method based on the analysis. For example, the generating AI adjusts the display method taking into account fluctuations in the student's grades. In this way, the optimal display method can be selected by referring to the past grade history. Some or all of the above-mentioned processing in the providing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the providing unit can input the student's past grade history into the generating AI and have the generating AI select the optimal display method.

[0052] The providing unit can customize the feedback based on the student's current learning situation when providing the grading result. The providing unit, for example, uses the generating AI to evaluate the student's current learning situation and customizes the feedback based on the evaluation. For example, the generating AI provides appropriate feedback based on the student's current learning progress. The providing unit can also use the generating AI to provide customized feedback based on the student's learning goal. For example, the generating AI provides optimal feedback based on the student's learning style. This makes it possible to provide more appropriate feedback by customizing the feedback based on the current learning situation. Some or all of the above-mentioned processing in the providing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the providing unit can input the student's current learning situation to the generating AI and have the generating AI customize the feedback.

[0053] The providing unit can select the optimal display method by considering the geographical location information of the student when providing the grading result. The providing unit, for example, uses the generating AI to evaluate the geographical location information of the student and selects the optimal display method based on the evaluation. For example, if the student lives in a specific area, the generating AI provides a display method including information related to the area. The providing unit can also use the generating AI to provide a display method including information related to the curriculum of a specific school if the student attends the specific school. For example, if the student has a specific cultural background, the generating AI provides a display method including information related to the culture. This makes it possible to select the optimal display method by considering the geographical location information. A part or all of the above-mentioned processing in the providing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the providing unit can input the geographical location information of the student to the generating AI and cause the generating AI to select the optimal display method.

[0054] The providing unit can provide feedback by analyzing the social media activity of the student when providing the grading result. The providing unit can, for example, use the generating AI to analyze the social media activity of the student. For example, the generating AI can provide feedback related to topics in which the student shows interest on social media. The providing unit can also use the generating AI to provide feedback related to accounts that the student follows on social media. For example, the generating AI can provide feedback related to content shared by the student on social media. This makes it possible to provide more relevant feedback by analyzing social media activity. Some or all of the above-mentioned processing in the providing unit can be performed, for example, using the generating AI, or can be performed without using the generating AI. For example, the providing unit can input the social media activity of the student to the generating AI and have the generating AI select the feedback.

[0055] The training data collection unit can optimize the collection algorithm by referring to past data when collecting training data. The training data collection unit, for example, uses the generation AI to evaluate past data and optimize the collection algorithm based on the evaluation. For example, the generation AI analyzes past data and selects an optimal collection algorithm. The training data collection unit can also adjust the collection algorithm by using the generation AI to take into account trends in past data. For example, the generation AI evaluates the quality of past data and optimizes the collection algorithm based on the evaluation. In this way, the collection algorithm can be optimized by referring to past data, and more appropriate data can be collected. Some or all of the above-mentioned processing in the training data collection unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the training data collection unit can input past data to the generation AI and cause the generation AI to optimize the collection algorithm.

[0056] The training data collection unit can weight data based on the time of submission of the answer when collecting training data. The training data collection unit, for example, uses the generation AI to evaluate the time of submission of the answer and weights the data based on the evaluation. For example, the generation AI assigns a high weight to an answer submitted just before the deadline. The training data collection unit can also use the generation AI to assign a low weight to an answer submitted early. For example, the generation AI weights data based on importance regardless of the time of submission. This makes it possible to collect more appropriate data by weighting data based on the time of submission of the answer. Part or all of the above-mentioned processing in the training data collection unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the training data collection unit can input the time of submission of the answer to the generation AI and have the generation AI execute the weighting of the data.

[0057] When providing feedback, the feedback unit can provide optimal feedback by referring to the student's past feedback history. The feedback unit, for example, uses the generation AI to evaluate the student's past feedback history and provides optimal feedback based on the evaluation. For example, the generation AI provides appropriate feedback based on the student's past feedback history. The feedback unit can also use the generation AI to analyze the student's feedback tendency and provide optimal feedback based on the analysis. For example, the generation AI adjusts the feedback content in consideration of fluctuations in the student's feedback. In this way, optimal feedback can be provided by referring to the past feedback history. Some or all of the above-mentioned processing in the feedback unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the feedback unit can input the student's past feedback history into the generation AI and have the generation AI select the optimal feedback.

[0058] When providing feedback, the feedback unit can provide optimal feedback by taking into account the device information of the student. The feedback unit, for example, uses a generation AI to evaluate the device information of the student and provides optimal feedback based on the evaluation. For example, when the student is using a smartphone, the generation AI provides feedback that matches the screen size. In addition, when the student is using a tablet, the feedback unit can also use the generation AI to provide feedback optimized for a large screen. For example, when the student is using a smartwatch, the generation AI provides concise and highly visible feedback. This makes it possible to provide optimal feedback by taking into account the device information. Part or all of the above-mentioned processing in the feedback unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the feedback unit inputs the device information of the student into the generation AI and allows the generation AI to select optimal feedback.

[0059] The system according to the embodiment is not limited to the above-described example, and various modifications are possible, for example, as follows.

[0060] The automatic correction system may further include a learning style analysis unit that analyzes the learning style of the student. The learning style analysis unit analyzes the student's past learning data and behavioral patterns to identify what kind of learning style the student has. For example, the student may be classified into categories such as visual learners, auditory learners, or experiential learners. The learning style analysis unit may provide visual learners with feedback that makes extensive use of diagrams and graphs, and audio feedback to auditory learners. It may also provide practical advice to experiential learners. This allows the student to receive optimal feedback according to their learning style, improving their learning effectiveness.

[0061] The summarization unit can further adjust the content of the summary taking into account the student's learning progress. For example, the summarization unit can create a summary that omits content that the student already understands and focuses on new information or areas of insufficient understanding. This allows the student to focus on the information that is most important to them, enabling efficient learning. The summarization unit can also adjust the level of detail of the summary depending on the student's learning progress. For example, a detailed summary can be provided for beginners, and a concise summary can be provided for advanced learners. This allows the optimal summary to be provided according to each student's learning level.

[0062] The similarity calculation unit can further calculate the similarity taking into account the style and expression of the answer. For example, the similarity calculation unit can evaluate whether the style of the answer matches the model answer, and adjust the similarity based on the degree of match of the style. The similarity calculation unit can also evaluate to what extent the expression method of the answer matches the model answer, and adjust the similarity based on the degree of match of the expression method. This enables a more accurate similarity calculation that takes into account differences in style and expression. Furthermore, the similarity calculation unit can evaluate the logical structure of the answer and the way the argument is developed, and adjust the similarity based on that. This enables a similarity calculation that takes into account logical consistency and clarity of the argument.

[0063] The providing unit can further customize the feedback based on the student's learning goals. For example, the providing unit can evaluate the achievement level and progress based on the learning goals set by the student and provide feedback accordingly. Specifically, if the student has a goal of improving a specific skill, the providing unit can provide feedback related to that skill with a focus. The providing unit can also suggest the next assignment or learning content to be tackled based on the student's learning goals. This allows the student to efficiently progress in learning toward his or her learning goals. Furthermore, the providing unit can provide encouragement and advice to increase motivation based on the student's learning goals.

[0064] The training data collection unit can further select the training data taking into consideration the learning history of the student. For example, the training data collection unit can analyze what kind of answers the student has submitted in the past and what kind of evaluations he or she has received, and select the training data based on the results. This makes it possible to collect more appropriate training data based on the learning history of the student. The training data collection unit can also collect training data that focuses on a specific field or skill based on the learning history of the student. This makes it possible to provide training data according to the learning needs of the student. Furthermore, the training data collection unit can adjust the frequency of collecting the training data based on the learning history of the student.

[0065] The process flow of the first embodiment will be briefly described below.

[0066] Step 1: The reader reads the answer sheet written by the student. The answer sheet written by the student may be handwritten or digital. The reader digitizes the handwritten answer sheet using scanning technology and reads it. The reader can also directly read an answer sheet submitted in digital format. In addition, the reader can read a printed answer sheet using OCR technology. Step 2: The summarization unit uses a generation AI to summarize the answer sheet read by the reading unit. The summarization is based on the length of the sentence and the importance of the information being summarized. For example, the generation AI uses a text generation AI (e.g., LLM) to concisely summarize the answer sheet. The summarization unit can also use a multimodal generation AI to summarize the contents of the answer sheet. The generation AI uses keyword extraction technology to pick out particularly important information in the answer sheet and summarizes it based on that. Step 3: The similarity calculation unit calculates the similarity between the summary generated by the summarization unit and the model answer. The similarity is calculated using a method such as cosine similarity or Jaccard coefficient. For example, the similarity between the answer and the model answer is calculated using cosine similarity. The similarity can also be calculated using Jaccard coefficient or TF-IDF. The similarity calculation unit can calculate the similarity using an AI model that inputs the summary generated by the summarization unit and the model answer and outputs the similarity. Step 4: The provider calculates a score based on the similarity calculated by the similarity calculation unit. The score is calculated according to a scoring method and evaluation criteria based on the similarity. For example, a score ranging from 0 to 100 points is given based on the similarity. Also, a grade from A to F can be given based on the similarity. The provider scales the score so that the higher the similarity, the higher the score. For example, an A is given for a similarity of 90% or more, and a B for 80% or more. A percentile is an index showing a relative position in the whole, and answers that fall in the top 10% are given a high grade.

[0067] (Example 2) The automatic correction system according to the embodiment of the present invention is a system that automatically reads an answer written by a student, has a generation AI summarize the answer, calculates the similarity to the model answer, and gives a score. This automatic correction system reads an answer written by a student, and a generation AI summarizes the answer. The generation AI analyzes the content of the answer, extracts important points, and creates a summary. For example, a long answer written by a student is summarized in a short form to clarify the main points. This summary is used to calculate the similarity to the model answer. Next, the generation AI calculates the similarity to the model answer. The generation AI compares the summarized answer with the model answer and calculates the similarity. For example, the generation AI evaluates the word agreement and sentence structure similarity between the summary of the answer and the model answer. This similarity is scored and provided as a scoring result. Furthermore, a similar process is applied to essay tests. An essay answer written by a student is read, and a generation AI creates a summary. Next, the generation AI calculates the similarity to the model answer and gives a score. This enables automatic correction of essays. This system reduces the burden on teachers and allows students' answers to be evaluated quickly and accurately. For example, teachers no longer need to manually correct a large number of answers, saving time and effort. In addition, students can receive feedback quickly and quickly identify and improve their weak points. This allows the automatic correction system to automatically correct students' answers and provide scoring results.

[0068] The automatic correction system according to the embodiment includes a reading unit, a summarizing unit, a similarity calculation unit, and a providing unit. The reading unit reads an answer written by a student. The answer written by a student includes, for example, a handwritten answer and an answer in digital format, but is not limited to such examples. For example, the reading unit digitizes and reads the handwritten answer using scanning technology. The reading unit can also directly read an answer submitted in digital format. Furthermore, the reading unit can also read a printed answer using OCR technology. For example, the reading unit scans a handwritten answer with a high-resolution scanner and converts it into character information using OCR technology. For digital format answers, those submitted in a specific file format can be directly read. OCR technology recognizes printed characters with high accuracy and converts them into digital text. The summarizing unit summarizes the answer read by the reading unit using a generation AI. The summarization is performed based on, for example, the length of the sentence and the importance of the information to be summarized, but is not limited to such examples. For example, the generation AI uses a text generation AI (e.g., LLM) to concisely summarize the answer. The summarization unit can also use a multimodal generation AI to summarize the contents of the answer. The summarization unit can also use the generation AI to extract and summarize important parts of a sentence. For example, the text generation AI has learned a large amount of text data and has advanced natural language processing capabilities. The multimodal generation AI can handle multiple modalities such as images and audio as well as text. The generation AI uses keyword extraction technology to pick up particularly important information in the answer and summarizes it based on that. The similarity calculation unit calculates the similarity between the summary generated by the summarization unit and the model answer. The similarity is calculated by a method such as cosine similarity or Jaccard coefficient, but is not limited to such examples. The similarity calculation unit calculates the similarity between the answer and the model answer using, for example, cosine similarity. The similarity calculation unit can also calculate the similarity using the Jaccard coefficient. The similarity calculation unit can also calculate the similarity using TF-IDF.For example, the similarity calculation unit calculates the angle between two vectors in a vector space model as the cosine similarity, and determines that the smaller the angle, the higher the similarity. The Jaccard coefficient is a value obtained by dividing the size of the common part of two sets by the total size, and indicates the degree of overlap of the sets. The TF-IDF is an index for evaluating the importance of words in a document, and is used to calculate the similarity between the answer sheet and the model answer. A part or all of the above-mentioned processing in the similarity calculation unit may be performed using, for example, AI, or may be performed without using AI. For example, the similarity calculation unit can calculate the similarity using an AI model that inputs the summary generated by the summarization unit and the model answer and outputs the similarity. The providing unit calculates a score based on the similarity calculated by the similarity calculation unit. The score is calculated, for example, according to a scoring method or evaluation criteria based on the similarity. The providing unit assigns a score in the range of 0 to 100 points based on the similarity. The providing unit can also assign an evaluation from A to F based on the similarity. The providing unit can also calculate percentiles based on the similarity. For example, the providing unit performs scaling so that the higher the similarity, the higher the score. The providing unit performs evaluation, for example, with a similarity of 90% or more as A and a similarity of 80% or more as B. The percentile is an index indicating a relative position in the whole, and for example, the providing unit gives a high score to an answer that is in the top 10%. In this way, the automatic correction system according to the embodiment can automatically correct the student's answer and provide the grading result. A part or all of the above-mentioned processing in the providing unit may be performed using, for example, AI, or may be performed without using AI. For example, the providing unit can calculate the score using an AI model that inputs the similarity calculated by the similarity calculation unit and outputs the score. For example, the output unit displays the grading result to the student or teacher through a web application or a mobile application. If feedback on paper is desired, the result is printed using a printer. Sending by email provides quick feedback by sending the result directly to the student or guardian. Some or all of the above-described processing in the output section may be performed using AI, for example, or may be performed without using AI.

[0069] The reading unit reads the answer sheet written by the student. The answer sheet written by the student includes, for example, a handwritten answer sheet, an answer sheet in digital format, and the like, but is not limited to such examples. For example, the reading unit digitizes and reads the handwritten answer sheet using scanning technology. The reading unit can also directly read an answer sheet submitted in digital format. Furthermore, the reading unit can also read a printed answer sheet using OCR technology. For example, the reading unit scans a handwritten answer sheet with a high-resolution scanner and converts it into character information using OCR technology. Digital answers submitted in a specific file format can be directly read. OCR technology recognizes printed characters with high accuracy and converts them into digital text. By combining these technologies, the reading unit can efficiently digitize answers in various formats and import them into the system. For example, a handwritten answer sheet is scanned by a scanner and then converted into character information using OCR technology. At this time, the OCR technology learns a large amount of handwritten data in advance to improve the recognition accuracy of handwritten characters. Digital answer sheets are often submitted in formats such as PDF or Word files, which are directly imported into the system. In addition, the reader also has the function of performing error checks when reading answer sheets, detecting and correcting incomplete data and reading errors. For example, it corrects noise and distortions that occur during scanning to obtain accurate character information. In addition, OCR technology is compatible with different fonts and character sizes, and can recognize character information on printed answer sheets with high accuracy. This allows the reader to quickly and accurately digitize answer sheets in various formats and provide the data required for subsequent processing.

[0070] The summarizing unit uses the generation AI to summarize the answer read by the reading unit. The summarization is performed, for example, based on the length of the sentence and the importance of the information to be summarized, but is not limited to such examples. For example, the generation AI uses a text generation AI (for example, LLM) to briefly summarize the answer. The summarizing unit can also summarize the contents of the answer using a multimodal generation AI. The summarizing unit can also extract and summarize important parts of the sentence using the generation AI. For example, the text generation AI has learned a large amount of text data and has advanced natural language processing capabilities. The multimodal generation AI can handle multiple modals such as images and audio as well as text. The generation AI uses keyword extraction technology to pick up particularly important information in the answer and summarize it based on that. The summarizing unit has a large amount of answer data learned in advance to improve the accuracy of the summarization performed by the generation AI. For example, a model for generating accurate summaries is trained using past answer data. The summarizing unit also has a function of evaluating the summary generated by the generation AI and making corrections as necessary. For example, if the content of the summary is inappropriate or important information is missing, the summarization unit automatically corrects it. The summarization unit also continuously evaluates the accuracy of the summary and updates the generation AI model to improve the quality of the summary. This allows the summarization unit to quickly and accurately summarize the answer sheet read by the reading unit and provide the information necessary for subsequent processing.

[0071] The similarity calculation unit calculates the similarity between the summary generated by the summarization unit and the model answer. The similarity is calculated by, for example, a method such as cosine similarity or Jaccard coefficient, but is not limited to such examples. The similarity calculation unit calculates the similarity between the answer and the model answer using, for example, cosine similarity. The similarity calculation unit can also calculate the similarity using the Jaccard coefficient. The similarity calculation unit can also calculate the similarity using TF-IDF. For example, the similarity calculation unit calculates the angle between two vectors in a vector space model as cosine similarity, and determines that the smaller the angle, the higher the similarity. The Jaccard coefficient is a value obtained by dividing the size of the common part of two sets by the total size, and indicates the degree of overlap of the sets. The TF-IDF is an index that evaluates the importance of words in a document, and is used to calculate the similarity between the answer and the model answer. A part or all of the above-mentioned processing in the similarity calculation unit may be performed using, for example, AI, or may be performed without using AI. For example, the similarity calculation unit can calculate the similarity using an AI model that inputs the summary generated by the summarization unit and the model answer and outputs the similarity. The similarity calculation unit can perform a more accurate similarity evaluation by combining these calculation methods. For example, by using cosine similarity and Jaccard coefficient together, the degree of agreement and the similarity of structure between the content of the answer and the model answer are evaluated simultaneously. In addition, by using TF-IDF, the degree of agreement of important keywords is evaluated, and the extent to which the content of the answer matches the model answer is analyzed in detail. As a result, the similarity calculation unit can calculate the similarity between the summary generated by the summarization unit and the model answer with high accuracy and provide information necessary for subsequent processing.

[0072] The providing unit calculates a score based on the similarity calculated by the similarity calculation unit. The score is calculated, for example, according to a scoring method or evaluation criteria based on the similarity. The providing unit, for example, gives a score in the range of 0 to 100 points based on the similarity. The providing unit can also give an evaluation from A to F based on the similarity. The providing unit can also calculate a percentile based on the similarity. For example, the providing unit performs scaling so that the higher the similarity, the higher the score. The providing unit performs evaluation, for example, A for a similarity of 90% or more, B for 80% or more, and so on. The percentile is an index indicating a relative position in the whole, and for example, the providing unit gives a high evaluation to an answer that is in the top 10%. In this way, the automatic correction system according to the embodiment can automatically correct the student's answer and provide the grading result. A part or all of the above-mentioned processing in the providing unit may be performed, for example, using AI, or may be performed without using AI. For example, the providing unit can calculate the score using an AI model that inputs the similarity calculated by the similarity calculation unit and outputs the score. For example, the output unit displays the grading results to students and teachers through a web application or a mobile application. If feedback on paper is desired, the results are printed using a printer. Sending by email provides quick feedback by sending the results directly to students and parents. Some or all of the above-mentioned processing in the output unit may be performed using AI, for example, or may be performed without using AI. The providing unit can generate not only the calculation of the score but also the content of the feedback. For example, it provides feedback that specifically indicates which parts of the answer match the model answer and which parts have room for improvement. This allows students to understand the strengths and weaknesses of their answers and use them to create their next answer. Furthermore, the providing unit can also customize the content of the feedback individually. For example, it proposes individual advice and study plans based on the student's past grades and learning history. This allows the providing unit to provide optimal feedback to each student and maximize the learning effect.

[0073] The summarizing unit can analyze the contents of the answer sheet, extract key points, and create a summary. The summarizing unit can, for example, use a generating AI to analyze the contents of the answer sheet, extract key points, and create a summary. For example, the generating AI can pick up particularly important information in the answer sheet and create a summary based on that. The summarizing unit can also use a generating AI to extract and summarize important parts of a sentence. For example, the generating AI can use a keyword extraction technique to pick up particularly important information in the answer sheet and create a summary based on that. This improves the accuracy of the summary by extracting and summarizing important points of the answer sheet. Some or all of the above-mentioned processing in the summarizing unit can be performed, for example, using a generating AI, or can be performed without using a generating AI. For example, the summarizing unit can analyze the contents of the answer sheet, extract key points, and create a summary using an algorithm.

[0074] The similarity calculation unit can compare the answer and the model answer and calculate the similarity. The similarity calculation unit can, for example, use a generation AI to compare the answer and the model answer and calculate the similarity. For example, the generation AI evaluates the word agreement and the similarity of the sentence structure between the summary of the answer and the model answer. The similarity calculation unit can also use the generation AI to calculate the similarity using a method such as cosine similarity or Jaccard coefficient. For example, the generation AI calculates the similarity between the answer and the model answer using cosine similarity. As a result, the accuracy of the similarity calculation is improved by comparing the summarized answer and the model answer. A part or all of the above-mentioned processing in the similarity calculation unit may be performed using, for example, a generation AI, or may be performed without using a generation AI. For example, the similarity calculation unit can compare the answer and the model answer and calculate the similarity using an algorithm for calculating the similarity.

[0075] The providing unit can score the similarity and provide the scoring result. The providing unit can score the similarity using, for example, a generating AI and provide the scoring result. For example, the generating AI calculates the score according to a scoring method or evaluation criteria based on the similarity. The providing unit can also use the generating AI to give a score in the range of 0 to 100 points based on the similarity. This improves the efficiency of scoring by scoring the similarity and providing the scoring result. A part or all of the above-mentioned processing in the providing unit can be performed using, for example, a generating AI, or can be performed without using a generating AI. For example, the providing unit can calculate the score using an algorithm for scoring the similarity and providing the scoring result.

[0076] The automatic correction system according to the embodiment further includes a training data collection unit. The training data collection unit collects data for improving the accuracy of the summarization section. For example, the training data collection unit collects past answer data and model answer data, and uses them as training data for the summarization section. The training data collection unit can also use a generation AI to collect data for improving the accuracy of the summarization section. For example, the generation AI analyzes past answer data and selects data to be used as training data for the summarization section. As a result, the accuracy of the summarization section is improved by collecting training data. Some or all of the above-mentioned processing in the training data collection unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the training data collection unit collects past answer data, inputs it to the generation AI, and allows the generation AI to select data to be used as training data for the summarization section.

[0077] The automatic correction system according to the embodiment further includes a feedback unit. The feedback unit provides feedback to the student based on the grading result. For example, the feedback unit uses a generation AI to analyze the grading result and provide the student with specific improvements and advice. The feedback unit can also use the generation AI to identify weak points in the student's answer and provide feedback based on the weak points. For example, the generation AI analyzes the content of the answer, extracts points that need improvement, and generates feedback based on the weak points. As a result, the student's learning effect is improved by providing feedback based on the grading result. A part or all of the above-mentioned processing in the feedback unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the feedback unit can generate feedback using an algorithm for analyzing the grading result and providing the student with specific improvements and advice.

[0078] The reading unit can estimate the emotion of the student and adjust the timing of reading the answer sheet based on the estimated emotion of the student. The reading unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the reading unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The reading unit can also record the student's voice and estimate the emotion using a voice analysis technique. For example, the reading unit analyzes the tone and speed of the voice and calculates an emotion score. The reading unit can also collect the student's biometric data (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the reading unit calculates an emotion score based on the fluctuation of the heart rate. This allows the answer sheet to be read at a more appropriate timing by adjusting the timing of reading the answer sheet according to the student's emotion. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI is a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the reading unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the reading unit may input image data of a student taken by a camera to the generation AI, and cause the generation AI to estimate the student's emotion.

[0079] The reading unit can analyze the student's past answer submission history and select the optimal reading method. The reading unit can, for example, use the generation AI to analyze the student's past answer submission history. For example, the generation AI can analyze the format and content of the answer submitted by the student in the past and select the optimal reading method. The reading unit can also adjust the reading speed based on the student's past answer submission history using the generation AI. For example, the generation AI can adjust the reading speed based on the amount of answers submitted by the student in the past. In this way, the optimal reading method can be selected by analyzing the past answer submission history. Part or all of the above-mentioned processing in the reading unit can be performed using, for example, the generation AI, or can be performed without using the generation AI. For example, the reading unit can input the student's past answer submission history into the generation AI and cause the generation AI to select the optimal reading method.

[0080] The reading unit can filter the answer sheets based on the student's current learning situation and field of interest when reading the answer sheets. The reading unit, for example, uses the generation AI to analyze the student's current learning situation and field of interest and performs filtering. For example, the generation AI preferentially reads answer sheets related to the unit the student is currently studying. The reading unit can also use the generation AI to filter and read answer sheets related to the student's field of interest. For example, the generation AI extracts keywords related to the student's field of interest and filters the answer sheets based on the keywords. This makes it possible to read more relevant answer sheets by filtering based on the student's learning situation and field of interest. Part or all of the above-mentioned processing in the reading unit may be performed using, for example, the generation AI, or may be performed without using the generation AI. For example, the reading unit can input the student's current learning situation and field of interest into the generation AI and have the generation AI execute filtering.

[0081] The reading unit can estimate the emotion of the student and determine the priority of the answer sheets to be read based on the estimated emotion of the student. The reading unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the reading unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The reading unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the reading unit analyzes the tone and speed of the voice and calculates an emotion score. The reading unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the reading unit calculates an emotion score based on the fluctuation of the heart rate. As a result, the priority of the answer sheets can be determined according to the emotion of the student, so that more appropriate answer sheets can be preferentially read. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI can be a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the reading unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the reading unit may input image data of a student taken by a camera to the generation AI, and cause the generation AI to estimate the student's emotion.

[0082] When reading the answer sheet, the reading unit can preferentially read highly relevant answer sheets by taking into consideration the geographical location information of the student. The reading unit, for example, uses the generation AI to analyze the geographical location information of the student and preferentially reads highly relevant answer sheets. For example, if the student lives in a specific area, the generation AI preferentially reads answer sheets related to that area. In addition, the reading unit can also use the generation AI to preferentially read answer sheets related to the curriculum of a specific school if the student attends that school. For example, if the student has a specific cultural background, the generation AI preferentially reads answer sheets related to that culture. In this way, by taking into consideration the geographical location information of the student, highly relevant answer sheets can be preferentially read. Part or all of the above-mentioned processing in the reading unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the reading unit can input the geographical location information of the student to the generation AI and cause the generation AI to select highly relevant answer sheets.

[0083] The reading unit can analyze the student's social media activity and read the related answer when reading the answer. The reading unit can, for example, use the generation AI to analyze the student's social media activity. For example, the generation AI reads the answer related to a topic in which the student is interested on social media. The reading unit can also use the generation AI to read the answer related to an account that the student follows on social media. For example, the generation AI reads the answer related to the content shared by the student on social media. In this way, the related answer can be read by analyzing the student's social media activity. Some or all of the above-mentioned processing in the reading unit can be performed, for example, using the generation AI or can be performed without using the generation AI. For example, the reading unit can input the student's social media activity into the generation AI and have the generation AI select the related answer.

[0084] The summarizing unit can estimate the emotion of the student and adjust the expression method of the summary based on the estimated emotion of the student. The summarizing unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the summarizing unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The summarizing unit can also record the voice of the student and estimate the emotion using a voice analysis technology. For example, the summarizing unit analyzes the tone and speed of the voice and calculates an emotion score. The summarizing unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the summarizing unit calculates an emotion score based on the fluctuation of the heart rate. This allows a more appropriate summary to be generated by adjusting the expression method of the summary according to the emotion of the student. The emotion estimation is realized using, for example, an emotion engine or a generation AI using an emotion estimation function. The generation AI is, for example, a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the summarizing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the summarizing unit may input image data of the students taken by a camera to the generating AI, and cause the generating AI to estimate the students' emotions.

[0085] The summarization unit can adjust the level of detail of the summary based on the importance of the answer when generating a summary. The summarization unit, for example, uses a generation AI to evaluate the importance of the answer and adjust the level of detail of the summary based on the evaluation. For example, the generation AI picks up particularly important information in the answer and generates a detailed summary based on the information. The summarization unit can also adjust the length and content of the summary based on the importance of the answer using the generation AI. For example, the generation AI uses a keyword extraction technique to pick up particularly important information in the answer and adjusts the level of detail of the summary based on the information. This allows a more appropriate summary to be generated by adjusting the level of detail of the summary based on the importance of the answer. Some or all of the above-mentioned processing in the summarization unit may be performed using, for example, a generation AI, or may be performed without using a generation AI. For example, the summarization unit can generate a summary using an algorithm for evaluating the importance of the answer and adjusting the level of detail of the summary based on the evaluation.

[0086] The summarization unit can apply different summarization algorithms depending on the category of the answer when generating a summary. The summarization unit, for example, uses a generation AI to classify the category of the answer and apply different summarization algorithms accordingly. For example, the generation AI applies a logical summarization algorithm in the case of an essay, and applies an emotional summarization algorithm in the case of a narrative text. The summarization unit can also use the generation AI to adjust the content and format of the summary depending on the category of the answer. For example, the generation AI applies an information-oriented summarization algorithm in the case of an explanatory text. This makes it possible to generate a more appropriate summary by applying different summarization algorithms depending on the category of the answer. Some or all of the above-mentioned processing in the summarization unit may be performed, for example, using a generation AI, or may be performed without using a generation AI. For example, the summarization unit can generate a summary using an algorithm for classifying the category of the answer and applying different summarization algorithms accordingly.

[0087] The summarizing unit can estimate the emotion of the student and adjust the length of the summary based on the estimated emotion of the student. The summarizing unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the summarizing unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The summarizing unit can also record the student's voice and estimate the emotion using a voice analysis technique. For example, the summarizing unit analyzes the tone and speed of the voice and calculates an emotion score. The summarizing unit can also collect the student's biometric data (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the summarizing unit calculates an emotion score based on the fluctuation of the heart rate. This allows the length of the summary to be adjusted according to the student's emotion, thereby generating a more appropriate summary. The emotion estimation is realized using, for example, an emotion engine or a generation AI using an emotion estimation function. The generation AI can be a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the summarizing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the summarizing unit may input image data of the students taken by a camera to the generating AI, and cause the generating AI to estimate the students' emotions.

[0088] The summarization unit can determine the priority of summaries based on the time of submission of answers when generating summaries. The summarization unit, for example, uses a generation AI to evaluate the time of submission of answers and determines the priority of summaries based on the evaluation. For example, the generation AI prioritizes summarization of answers submitted just before the deadline. The summarization unit can also use the generation AI to summarize answers submitted early next. For example, the generation AI determines the priority of summaries based on importance regardless of the time of submission. In this way, a more appropriate summary can be generated by determining the priority of summaries based on the time of submission of answers. Some or all of the above-mentioned processing in the summarization unit may be performed, for example, using a generation AI or may be performed without using a generation AI. For example, the summarization unit can generate summaries using an algorithm for evaluating the time of submission of answers and determining the priority of summaries based on the evaluation.

[0089] The summarization unit can adjust the order of summaries based on the relevance of the answers when generating summaries. The summarization unit, for example, uses a generation AI to evaluate the relevance of the answers and adjust the order of summaries based on the evaluation. For example, the generation AI prioritizes summarization of answers related to important topics. The summarization unit can also use the generation AI to summarize answers related to normal topics next. For example, the generation AI summarizes answers related to less relevant topics last. This allows a more appropriate summary to be generated by adjusting the order of summaries based on the relevance of the answers. Some or all of the above-mentioned processing in the summarization unit may be performed, for example, using a generation AI, or may be performed without using a generation AI. For example, the summarization unit can generate a summary using an algorithm for evaluating the relevance of answers and adjusting the order of summaries based on the evaluation.

[0090] The similarity calculation unit can estimate the emotion of the student and adjust the criterion of the similarity calculation based on the estimated emotion of the student. The similarity calculation unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the similarity calculation unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The similarity calculation unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the similarity calculation unit analyzes the tone and speed of the voice and calculates an emotion score. The similarity calculation unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the similarity calculation unit calculates an emotion score based on the fluctuation of the heart rate. This allows the criterion of the similarity calculation to be adjusted according to the emotion of the student, thereby making it possible to perform a more appropriate similarity calculation. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI is, for example, a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the similarity calculation unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the similarity calculation unit may input image data of a student taken by a camera to the generation AI, and cause the generation AI to estimate the student's emotion.

[0091] The similarity calculation unit can improve the accuracy of the similarity calculation by considering the interrelationships of the answers when calculating the similarity. The similarity calculation unit, for example, uses a generation AI to evaluate the interrelationships of the answers and improves the accuracy of the similarity calculation based on the evaluation. For example, the generation AI calculates the similarity by considering the interrelationships of multiple answers. The similarity calculation unit can also use the generation AI to evaluate the relevance of the contents of the answers and improve the accuracy of the similarity calculation based on the evaluation. For example, the generation AI calculates the similarity by considering the context of the answers. As a result, the accuracy of the similarity calculation is improved by considering the interrelationships of the answers. A part or all of the above-mentioned processing in the similarity calculation unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the similarity calculation unit can calculate the similarity using an algorithm for evaluating the interrelationships of the answers and calculating the similarity based on the evaluation.

[0092] The similarity calculation unit can perform the similarity calculation while taking into account the attribute information of the person who submitted the answer sheet when calculating the similarity. The similarity calculation unit, for example, uses a generation AI to evaluate the attribute information of the person who submitted the answer sheet and performs the similarity calculation based on the attribute information. For example, the generation AI adjusts the criterion for the similarity calculation based on the grade of the person who submitted the answer sheet. The similarity calculation unit can also adjust the criterion for the similarity calculation based on the learning history of the person who submitted the answer sheet using the generation AI. For example, the generation AI adjusts the criterion for the similarity calculation based on the learning style of the person who submitted the answer sheet. This improves the accuracy of the similarity calculation by taking into account the attribute information of the person who submitted the answer sheet. A part or all of the above-mentioned processing in the similarity calculation unit may be performed using, for example, a generation AI, or may be performed without using a generation AI. For example, the similarity calculation unit can calculate the similarity using an algorithm for evaluating the attribute information of the person who submitted the answer sheet and performing the similarity calculation based on the evaluation.

[0093] The similarity calculation unit can estimate the emotion of the student and adjust the order of displaying the results of the similarity calculation based on the estimated emotion of the student. The similarity calculation unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the similarity calculation unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The similarity calculation unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the similarity calculation unit analyzes the tone and speed of the voice and calculates an emotion score. The similarity calculation unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the similarity calculation unit calculates an emotion score based on the fluctuation of the heart rate. As a result, by adjusting the order of displaying the results of the similarity calculation according to the emotion of the student, it is possible to provide a more appropriate result. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI is, for example, a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the similarity calculation unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the similarity calculation unit may input image data of a student taken by a camera to the generation AI, and cause the generation AI to estimate the student's emotion.

[0094] The similarity calculation unit can perform the similarity calculation taking into account the geographical distribution of the answers when calculating the similarity. The similarity calculation unit, for example, uses the generation AI to evaluate the geographical distribution of the answers and performs the similarity calculation based on the evaluation. For example, the generation AI preferentially calculates the similarity of answers in geographically close areas. The similarity calculation unit can also use the generation AI to calculate the similarity of answers in geographically distant areas next. For example, the generation AI performs the similarity calculation based on the importance regardless of the geographical distribution. As a result, the accuracy of the similarity calculation is improved by taking the geographical distribution into consideration. A part or all of the above-mentioned processing in the similarity calculation unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the similarity calculation unit can calculate the similarity using an algorithm for evaluating the geographical distribution of the answers and performing the similarity calculation based on the evaluation.

[0095] The similarity calculation unit can improve the accuracy of the similarity calculation by referring to the related literature of the answer when calculating the similarity. The similarity calculation unit, for example, uses the generation AI to evaluate the related literature of the answer and improves the accuracy of the similarity calculation based on the evaluation. For example, the generation AI evaluates the contents of the answer by referring to the related literature and improves the accuracy of the similarity calculation. The similarity calculation unit can also improve the accuracy of the similarity calculation by using the generation AI and taking into account the citation of the related literature. For example, the generation AI extracts the main points of the related literature and improves the accuracy of the similarity calculation based on the extracted main points. As a result, the accuracy of the similarity calculation is improved by referring to the related literature. A part or all of the above-mentioned processing in the similarity calculation unit may be performed, for example, using the generation AI or may be performed without using the generation AI. For example, the similarity calculation unit can calculate the similarity by using an algorithm for evaluating the related literature of the answer and performing the similarity calculation based on the evaluation.

[0096] The providing unit can estimate the emotion of the student and adjust the display method of the grading result based on the estimated emotion of the student. The providing unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the providing unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The providing unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the providing unit analyzes the tone and speed of the voice and calculates an emotion score. The providing unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the providing unit calculates an emotion score based on the fluctuation of the heart rate. This makes it possible to provide more appropriate feedback by adjusting the display method of the grading result according to the emotion of the student. The emotion estimation is realized using, for example, an emotion engine or a generation AI using an emotion estimation function. The generation AI can be, but is not limited to, a text generation AI (for example, LLM) or a multimodal generation AI. A part or all of the above-mentioned processing in the providing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the providing unit may input image data of the student taken by a camera to the generating AI, and cause the generating AI to estimate the student's emotion.

[0097] When providing the grading results, the providing unit can select the optimal display method by referring to the student's past grade history. The providing unit, for example, uses the generating AI to evaluate the student's past grade history and selects the optimal display method based on the evaluation. For example, the generating AI selects the optimal display method based on the student's past grade history. The providing unit can also use the generating AI to analyze the student's grade trends and provide an appropriate display method based on the analysis. For example, the generating AI adjusts the display method taking into account fluctuations in the student's grades. In this way, the optimal display method can be selected by referring to the past grade history. Some or all of the above-mentioned processing in the providing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the providing unit can input the student's past grade history into the generating AI and have the generating AI select the optimal display method.

[0098] The providing unit can customize the feedback based on the student's current learning situation when providing the grading result. The providing unit, for example, uses the generating AI to evaluate the student's current learning situation and customizes the feedback based on the evaluation. For example, the generating AI provides appropriate feedback based on the student's current learning progress. The providing unit can also use the generating AI to provide customized feedback based on the student's learning goal. For example, the generating AI provides optimal feedback based on the student's learning style. This makes it possible to provide more appropriate feedback by customizing the feedback based on the current learning situation. Some or all of the above-mentioned processing in the providing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the providing unit can input the student's current learning situation to the generating AI and have the generating AI customize the feedback.

[0099] The providing unit can estimate the emotion of the student and determine the priority of the grading results based on the estimated emotion of the student. The providing unit, for example, estimates the emotion of the student using an emotion estimation algorithm. For example, the providing unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The providing unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the providing unit analyzes the tone and speed of the voice and calculates an emotion score. The providing unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the providing unit calculates an emotion score based on the fluctuation of the heart rate. This makes it possible to provide more appropriate feedback by determining the priority of the grading results according to the emotion of the student. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI is, for example, a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the providing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the providing unit may input image data of the student taken by a camera to the generating AI, and cause the generating AI to estimate the student's emotion.

[0100] The providing unit can select the optimal display method by considering the geographical location information of the student when providing the grading result. The providing unit, for example, uses the generating AI to evaluate the geographical location information of the student and selects the optimal display method based on the evaluation. For example, if the student lives in a specific area, the generating AI provides a display method including information related to the area. The providing unit can also use the generating AI to provide a display method including information related to the curriculum of a specific school if the student attends the specific school. For example, if the student has a specific cultural background, the generating AI provides a display method including information related to the culture. This makes it possible to select the optimal display method by considering the geographical location information. A part or all of the above-mentioned processing in the providing unit may be performed, for example, using the generating AI, or may be performed without using the generating AI. For example, the providing unit can input the geographical location information of the student to the generating AI and cause the generating AI to select the optimal display method.

[0101] The providing unit can provide feedback by analyzing the social media activity of the student when providing the grading result. The providing unit can, for example, use the generating AI to analyze the social media activity of the student. For example, the generating AI can provide feedback related to topics in which the student shows interest on social media. The providing unit can also use the generating AI to provide feedback related to accounts that the student follows on social media. For example, the generating AI can provide feedback related to content shared by the student on social media. This makes it possible to provide more relevant feedback by analyzing social media activity. Some or all of the above-mentioned processing in the providing unit can be performed, for example, using the generating AI, or can be performed without using the generating AI. For example, the providing unit can input the social media activity of the student to the generating AI and have the generating AI select the feedback.

[0102] The training data collection unit can estimate the emotion of the student and select the training data based on the estimated emotion of the student. The training data collection unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the training data collection unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The training data collection unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the training data collection unit analyzes the tone and speed of the voice and calculates an emotion score. The training data collection unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the training data collection unit calculates an emotion score based on the fluctuation of the heart rate. This allows more appropriate data to be collected by selecting the training data according to the emotion of the student. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI is, for example, a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the training data collection unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the training data collection unit may input image data of students taken by a camera to the generation AI, and cause the generation AI to estimate the emotions of the students.

[0103] The training data collection unit can optimize the collection algorithm by referring to past data when collecting training data. The training data collection unit, for example, uses the generation AI to evaluate past data and optimize the collection algorithm based on the evaluation. For example, the generation AI analyzes past data and selects an optimal collection algorithm. The training data collection unit can also adjust the collection algorithm by using the generation AI to take into account trends in past data. For example, the generation AI evaluates the quality of past data and optimizes the collection algorithm based on the evaluation. In this way, the collection algorithm can be optimized by referring to past data, and more appropriate data can be collected. Some or all of the above-mentioned processing in the training data collection unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the training data collection unit can input past data to the generation AI and cause the generation AI to optimize the collection algorithm.

[0104] The training data collection unit can estimate the emotion of the student and adjust the collection frequency of the training data based on the estimated emotion of the student. The training data collection unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the training data collection unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The training data collection unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the training data collection unit analyzes the tone and speed of the voice and calculates an emotion score. The training data collection unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the training data collection unit calculates an emotion score based on the fluctuation of the heart rate. As a result, by adjusting the collection frequency according to the emotion of the student, it is possible to collect data at a more appropriate time. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI is, for example, a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the training data collection unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the training data collection unit may input image data of students taken by a camera to the generation AI, and cause the generation AI to estimate the emotions of the students.

[0105] The training data collection unit can weight data based on the time of submission of the answer when collecting training data. The training data collection unit, for example, uses the generation AI to evaluate the time of submission of the answer and weights the data based on the evaluation. For example, the generation AI assigns a high weight to an answer submitted just before the deadline. The training data collection unit can also use the generation AI to assign a low weight to an answer submitted early. For example, the generation AI weights data based on importance regardless of the time of submission. This makes it possible to collect more appropriate data by weighting data based on the time of submission of the answer. Part or all of the above-mentioned processing in the training data collection unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the training data collection unit can input the time of submission of the answer to the generation AI and have the generation AI execute the weighting of the data.

[0106] The feedback unit can estimate the emotion of the student and adjust the content of the feedback based on the estimated emotion of the student. The feedback unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the feedback unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The feedback unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the feedback unit analyzes the tone and speed of the voice and calculates an emotion score. The feedback unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the feedback unit calculates an emotion score based on the fluctuation of the heart rate. This makes it possible to provide more appropriate feedback by adjusting the content of the feedback according to the emotion of the student. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI can be, but is not limited to, a text generation AI (for example, LLM) or a multimodal generation AI. A part or all of the above-mentioned processing in the feedback unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the feedback unit may input image data of the student taken by a camera to the generation AI and cause the generation AI to estimate the student's emotion.

[0107] When providing feedback, the feedback unit can provide optimal feedback by referring to the student's past feedback history. The feedback unit, for example, uses the generation AI to evaluate the student's past feedback history and provides optimal feedback based on the evaluation. For example, the generation AI provides appropriate feedback based on the student's past feedback history. The feedback unit can also use the generation AI to analyze the student's feedback tendency and provide optimal feedback based on the analysis. For example, the generation AI adjusts the feedback content in consideration of fluctuations in the student's feedback. In this way, optimal feedback can be provided by referring to the past feedback history. Some or all of the above-mentioned processing in the feedback unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the feedback unit can input the student's past feedback history into the generation AI and have the generation AI select the optimal feedback.

[0108] The feedback unit can estimate the emotion of the student and determine the priority of the feedback based on the estimated emotion of the student. The feedback unit estimates the emotion of the student using, for example, an emotion estimation algorithm. For example, the feedback unit captures the facial expression of the student with a camera and estimates the emotion using an emotion estimation algorithm. The feedback unit can also record the voice of the student and estimate the emotion using a voice analysis technique. For example, the feedback unit analyzes the tone and speed of the voice and calculates an emotion score. The feedback unit can also collect the biometric data of the student (heart rate and electrodermal activity) with a sensor and estimate the emotion using an emotion estimation algorithm. For example, the feedback unit calculates an emotion score based on the fluctuation of the heart rate. This allows the feedback priority to be determined according to the emotion of the student, thereby providing more appropriate feedback. The emotion estimation is realized using an emotion estimation function using, for example, an emotion engine or a generation AI. The generation AI can be a text generation AI (for example, LLM) or a multimodal generation AI, but is not limited to such examples. A part or all of the above-mentioned processing in the feedback unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the feedback unit may input image data of the student taken by a camera to the generation AI and cause the generation AI to estimate the student's emotion.

[0109] When providing feedback, the feedback unit can provide optimal feedback by taking into account the device information of the student. The feedback unit, for example, uses a generation AI to evaluate the device information of the student and provides optimal feedback based on the evaluation. For example, when the student is using a smartphone, the generation AI provides feedback that matches the screen size. In addition, when the student is using a tablet, the feedback unit can also use the generation AI to provide feedback optimized for a large screen. For example, when the student is using a smartwatch, the generation AI provides concise and highly visible feedback. This makes it possible to provide optimal feedback by taking into account the device information. Part or all of the above-mentioned processing in the feedback unit may be performed, for example, using the generation AI, or may be performed without using the generation AI. For example, the feedback unit inputs the device information of the student into the generation AI and allows the generation AI to select optimal feedback.

[0110] The system according to the embodiment is not limited to the above-described example, and various modifications are possible, for example, as follows.

[0111] The automatic correction system may further include a learning style analysis unit that analyzes the learning style of the student. The learning style analysis unit analyzes the student's past learning data and behavioral patterns to identify what kind of learning style the student has. For example, the student may be classified into categories such as visual learners, auditory learners, or experiential learners. The learning style analysis unit may provide visual learners with feedback that makes extensive use of diagrams and graphs, and audio feedback to auditory learners. It may also provide practical advice to experiential learners. This allows the student to receive optimal feedback according to their learning style, improving their learning effectiveness.

[0112] The summarization unit can further adjust the content of the summary taking into account the student's learning progress. For example, the summarization unit can create a summary that omits content that the student already understands and focuses on new information or areas of insufficient understanding. This allows the student to focus on the information that is most important to them, enabling efficient learning. The summarization unit can also adjust the level of detail of the summary depending on the student's learning progress. For example, a detailed summary can be provided for beginners, and a concise summary can be provided for advanced learners. This allows the optimal summary to be provided according to each student's learning level.

[0113] The similarity calculation unit can further calculate the similarity taking into account the style and expression of the answer. For example, the similarity calculation unit can evaluate whether the style of the answer matches the model answer, and adjust the similarity based on the degree of match of the style. The similarity calculation unit can also evaluate to what extent the expression method of the answer matches the model answer, and adjust the similarity based on the degree of match of the expression method. This enables a more accurate similarity calculation that takes into account differences in style and expression. Furthermore, the similarity calculation unit can evaluate the logical structure of the answer and the way the argument is developed, and adjust the similarity based on that. This enables a similarity calculation that takes into account logical consistency and clarity of the argument.

[0114] The providing unit can further customize the feedback based on the student's learning goals. For example, the providing unit can evaluate the achievement level and progress based on the learning goals set by the student and provide feedback accordingly. Specifically, if the student has a goal of improving a specific skill, the providing unit can provide feedback related to that skill with a focus. The providing unit can also suggest the next assignment or learning content to be tackled based on the student's learning goals. This allows the student to efficiently progress in learning toward his or her learning goals. Furthermore, the providing unit can provide encouragement and advice to increase motivation based on the student's learning goals.

[0115] The training data collection unit can further select the training data taking into consideration the learning history of the student. For example, the training data collection unit can analyze what kind of answers the student has submitted in the past and what kind of evaluations he or she has received, and select the training data based on the results. This makes it possible to collect more appropriate training data based on the learning history of the student. The training data collection unit can also collect training data that focuses on a specific field or skill based on the learning history of the student. This makes it possible to provide training data according to the learning needs of the student. Furthermore, the training data collection unit can adjust the frequency of collecting the training data based on the learning history of the student.

[0116] The feedback unit can estimate the student's emotions and adjust the tone of the feedback based on the estimated student's emotions. For example, if the student is feeling down, the feedback unit can provide feedback in an encouraging or comforting tone. Also, if the student is feeling confident, the feedback unit can provide feedback in a tone that encourages further challenges. This makes it possible to provide appropriate feedback according to the student's emotions and increase motivation to learn. Furthermore, the feedback unit can adjust the content of the feedback based on the student's emotions. For example, if the student is feeling anxious, the feedback unit can provide feedback that clearly indicates specific improvements and next steps.

[0117] The reading unit can estimate the student's emotions and adjust the reading order of the answer sheets based on the estimated student's emotions. For example, when a student is feeling stressed, the reading unit can reduce the student's burden by reading the easier questions first. Also, when a student is relaxed, the reading unit can maintain concentration by reading the more difficult questions first. This makes it possible to provide an appropriate reading order according to the student's emotions and improve the learning effect. Furthermore, the reading unit can adjust the reading speed based on the student's emotions. For example, when a student is impatient, the student can deepen his / her understanding by reading slowly.

[0118] The summarizing unit can estimate the student's emotion and adjust the way the summary is expressed based on the estimated student's emotion. For example, if the student is feeling anxious, the summarizing unit can provide the summary in a gentle tone. Also, if the student is feeling confident, the summarizing unit can provide the summary in a challenging tone. This makes it possible to provide an appropriate summary according to the student's emotion and to increase the student's motivation to learn. Furthermore, the summarizing unit can adjust the length of the summary based on the student's emotion. For example, if the student is tired, a short and concise summary can be provided.

[0119] The similarity calculation unit can estimate the student's emotions and adjust the criteria for similarity calculation based on the estimated student's emotions. For example, the similarity calculation unit can calculate the similarity using strict criteria when the student is confident. Also, the similarity calculation unit can calculate the similarity using loose criteria when the student is anxious. This allows an appropriate similarity calculation according to the student's emotions, and improves the ease of accepting feedback. Furthermore, the similarity calculation unit can adjust the order in which the results of the similarity calculation are displayed based on the student's emotions. For example, when the student is impatient, the most important results can be displayed first.

[0120] The providing unit can estimate the student's emotions and adjust the display method of the grading results based on the estimated student's emotions. For example, if the student is feeling down, the providing unit can emphasize and display positive feedback. Also, if the student is feeling confident, the providing unit can display the points for improvement in detail. This makes it possible to provide appropriate feedback according to the student's emotions and increase motivation to learn. Furthermore, the providing unit can adjust the content of the feedback based on the student's emotions. For example, if the student is feeling anxious, the providing unit can provide feedback that clearly indicates specific points for improvement and next steps.

[0121] The process flow of the second embodiment will be briefly described below.

[0122] Step 1: The reader reads the answer sheet written by the student. The answer sheet written by the student may be handwritten or digital. The reader digitizes the handwritten answer sheet using scanning technology and reads it. The reader can also directly read an answer sheet submitted in digital format. In addition, the reader can read a printed answer sheet using OCR technology. Step 2: The summarization unit uses a generation AI to summarize the answer sheet read by the reading unit. The summarization is based on the length of the sentence and the importance of the information being summarized. For example, the generation AI uses a text generation AI (e.g., LLM) to concisely summarize the answer sheet. The summarization unit can also use a multimodal generation AI to summarize the contents of the answer sheet. The generation AI uses keyword extraction technology to pick out particularly important information in the answer sheet and summarizes it based on that. Step 3: The similarity calculation unit calculates the similarity between the summary generated by the summarization unit and the model answer. The similarity is calculated using a method such as cosine similarity or Jaccard coefficient. For example, the similarity between the answer and the model answer is calculated using cosine similarity. The similarity can also be calculated using Jaccard coefficient or TF-IDF. The similarity calculation unit can calculate the similarity using an AI model that inputs the summary generated by the summarization unit and the model answer and outputs the similarity. Step 4: The provider calculates a score based on the similarity calculated by the similarity calculation unit. The score is calculated according to a scoring method and evaluation criteria based on the similarity. For example, a score ranging from 0 to 100 points is given based on the similarity. Also, a grade from A to F can be given based on the similarity. The provider scales the score so that the higher the similarity, the higher the score. For example, an A is given for a similarity of 90% or more, and a B for 80% or more. A percentile is an index showing a relative position in the whole, and answers that fall in the top 10% are given a high grade.

[0123] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires a voice indicating a user input for the result of the specific processing. The control unit 46A transmits the voice data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0124] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image (e.g., still image data or video data) are input. The data generation model 58 infers the input inference data according to the instruction indicated by the prompt, and outputs the inference result in one or more data formats of voice data, text data, and image data. The data generation model 58 includes, for example, a text generation AI, an image generation AI, and a multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specification processing unit 290 performs the above-mentioned specification processing while using the data generation model 58. The data generation model 58 may be a fine-tuned model so as to output an inference result from a prompt that does not include an instruction, and in this case, the data generation model 58 can output an inference result from a prompt that does not include an instruction. In the data processing device 12 and the like, the data generation model 58 includes a plurality of types, and the data generation model 58 includes an AI other than the generation AI. The AI ​​other than the generation AI is, for example, a linear regression, a logistic regression, a decision tree, a random forest, a support vector machine (SVM), a k-means clustering, a convolutional neural network (CNN), a recurrent neural network (RNN), a generative adversarial network (GAN), or a naive Bayes, and can perform various processes, but is not limited to such examples. The AI ​​may also be an AI agent. In addition, when the processes of each of the above-mentioned parts are performed by an AI, the processes are performed in part or in whole by the AI, but are not limited to such examples.In addition, processing performed by AI including the generation AI may be replaced with rule-based processing, and rule-based processing may be replaced with processing performed by AI including the generation AI.

[0125] Furthermore, the above-described processing by the data processing system 10 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart device 14, but may also be executed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart device 14. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart device 14 or an external device, and the smart device 14 acquires or collects information required for processing from the data processing device 12 or an external device.

[0126] For example, the reading unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart device 14. The summarizing unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart device 14. The similarity calculation unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart device 14. The providing unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart device 14. The training data collection unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart device 14. The feedback unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart device 14. The correspondence between each part and the device or control part is not limited to the above-mentioned example, and various modifications are possible.

[0127] [Second embodiment] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0128] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0129] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN and / or a LAN.

[0130] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0131] The microphone 238 receives instructions and the like from the user by receiving voice uttered by the user. The microphone 238 captures the voice uttered by the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs the voice according to instructions from the processor 46.

[0132] Camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the user's surroundings (e.g., an imaging range defined by an angle of view equivalent to the width of the field of vision of an average healthy person).

[0133] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is performed in a secure state.

[0134] Fig. 4 shows an example of main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0135] The processor 28 reads the specific processing program 56 from the storage 32, and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0136] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the emotion of the user using the emotion identification model 59 and perform identification processing using the user's emotion. In the emotion estimation function (emotion identification function) using the emotion identification model 59, various estimations and predictions regarding the user's emotion, including estimation and prediction of the user's emotion, are performed, but are not limited to such examples. In addition, the estimation and prediction of the emotion also include, for example, analysis of the emotion.

[0137] In the smart glasses 214, the specific processing is performed by the processor 46. The storage 50 stores a specific processing program 60. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as the control unit 46A in accordance with the specific processing program 60 executed on the RAM 48. Note that the smart glasses 214 have a data generation model and an emotion identification model similar to the data generation model 58 and the emotion identification model 59, and can also perform processing similar to that of the specific processing unit 290 using these models.

[0138] Note that a device other than the data processing device 12 may have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains a processing result (such as a prediction result) using the data generation model 58 by communicating with the server device having the data generation model 58. In addition, the data processing device 12 may be a server device, or may be a terminal device (such as a mobile phone, a robot, or a home appliance) owned by a user.

[0139] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires a voice indicating a user input for the result of the specific processing. The control unit 46A transmits the voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0140] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image (e.g., still image data or video data) is input. The data generation model 58 infers the input inference data according to the instruction indicated by the prompt, and outputs the inference result in one or more data formats of voice data, text data, and image data. The data generation model 58 includes, for example, a text generation AI, an image generation AI, and a multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specification processing unit 290 performs the above-mentioned specification processing while using the data generation model 58. The data generation model 58 may be a fine-tuned model so as to output an inference result from a prompt that does not include an instruction, and in this case, the data generation model 58 can output an inference result from a prompt that does not include an instruction. In the data processing device 12 and the like, the data generation model 58 includes a plurality of types, and the data generation model 58 includes an AI other than the generation AI. The AI ​​other than the generation AI is, for example, a linear regression, a logistic regression, a decision tree, a random forest, a support vector machine (SVM), a k-means clustering, a convolutional neural network (CNN), a recurrent neural network (RNN), a generative adversarial network (GAN), or a naive Bayes, and can perform various processes, but is not limited to such examples. The AI ​​may be an AI agent. In addition, when the processes of each part described above are performed by an AI, the processes are performed in part or in whole by the AI, but are not limited to such examples. In addition, the processes performed by the AI ​​including the generation AI may be replaced with rule-based processes, and the rule-based processes may be replaced with processes performed by the AI ​​including the generation AI.

[0141] The data processing system 210 according to the second embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 210 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart glasses 214, but may be executed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart glasses 214 or an external device, and the smart glasses 214 acquires or collects information required for processing from the data processing device 12 or an external device.

[0142] For example, the reading unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart glasses 214. The summarizing unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart glasses 214. The similarity calculation unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart glasses 214. The providing unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart glasses 214. The training data collection unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart glasses 214. The feedback unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the smart glasses 214. The correspondence between each part and the device or control part is not limited to the above-mentioned example, and various modifications are possible.

[0143] [Third embodiment] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0144] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0145] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN and / or a LAN.

[0146] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0147] The microphone 238 receives instructions and the like from the user by receiving voice uttered by the user. The microphone 238 captures the voice uttered by the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs the voice according to instructions from the processor 46.

[0148] Camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the user's surroundings (e.g., an imaging range defined by an angle of view equivalent to the width of the field of vision of an average healthy person).

[0149] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is performed in a secure state.

[0150] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0151] The processor 28 reads the specific processing program 56 from the storage 32, and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0152] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the emotion of the user using the emotion identification model 59 and perform identification processing using the user's emotion. In the emotion estimation function (emotion identification function) using the emotion identification model 59, various estimations and predictions regarding the user's emotion, including estimation and prediction of the user's emotion, are performed, but are not limited to such examples. In addition, the estimation and prediction of the emotion also include, for example, analysis of the emotion.

[0153] In the headset type terminal 314, the specific processing is performed by the processor 46. The storage 50 stores a specific program 60. The processor 46 reads the specific program 60 from the storage 50, and executes the read specific program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as the control unit 46A in accordance with the specific program 60 executed on the RAM 48. Note that the headset type terminal 314 has a data generation model and an emotion identification model similar to the data generation model 58 and the emotion identification model 59, and can also perform processing similar to that of the identification processing unit 290 using these models.

[0154] Note that a device other than the data processing device 12 may have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains a processing result (such as a prediction result) using the data generation model 58 by communicating with the server device having the data generation model 58. In addition, the data processing device 12 may be a server device, or may be a terminal device (such as a mobile phone, a robot, or a home appliance) owned by a user.

[0155] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires a voice indicating a user input for the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0156] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image (e.g., still image data or video data) is input. The data generation model 58 infers the input inference data according to the instruction indicated by the prompt, and outputs the inference result in one or more data formats of voice data, text data, and image data. The data generation model 58 includes, for example, a text generation AI, an image generation AI, and a multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specification processing unit 290 performs the above-mentioned specification processing while using the data generation model 58. The data generation model 58 may be a fine-tuned model so as to output an inference result from a prompt that does not include an instruction, and in this case, the data generation model 58 can output an inference result from a prompt that does not include an instruction. In the data processing device 12 and the like, the data generation model 58 includes a plurality of types, and the data generation model 58 includes an AI other than the generation AI. The AI ​​other than the generation AI is, for example, a linear regression, a logistic regression, a decision tree, a random forest, a support vector machine (SVM), a k-means clustering, a convolutional neural network (CNN), a recurrent neural network (RNN), a generative adversarial network (GAN), or a naive Bayes, and can perform various processes, but is not limited to such examples. The AI ​​may be an AI agent. In addition, when the processes of each part described above are performed by an AI, the processes are performed in part or in whole by the AI, but are not limited to such examples. In addition, the processes performed by the AI ​​including the generation AI may be replaced with rule-based processes, and the rule-based processes may be replaced with processes performed by the AI ​​including the generation AI.

[0157] The data processing system 310 according to the third embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 310 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the headset type terminal 314, but may be executed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the headset type terminal 314. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the headset type terminal 314 or an external device, and the headset type terminal 314 acquires or collects information required for processing from the data processing device 12 or an external device.

[0158] For example, the reading unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the headset type terminal 314. The summarizing unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the headset type terminal 314. The similarity calculation unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the headset type terminal 314. The providing unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the headset type terminal 314. The training data collection unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the headset type terminal 314. The feedback unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the headset type terminal 314. The correspondence between each part and the device or control part is not limited to the above-mentioned example, and various modifications are possible.

[0159] [Fourth embodiment] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[0160] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0161] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN and / or a LAN.

[0162] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[0163] The microphone 238 receives instructions and the like from the user by receiving voice uttered by the user. The microphone 238 captures the voice uttered by the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs the voice according to instructions from the processor 46.

[0164] Camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS image sensor or a CCD image sensor, and captures images of the user's surroundings (e.g., an imaging range defined by an angle of view equivalent to the width of the field of vision of a typical healthy person).

[0165] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is performed in a secure state.

[0166] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and legs. The posture and behavior of the robot 414 are controlled by controlling the motors of the arms, hands, and legs. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[0167] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0168] The processor 28 reads the specific processing program 56 from the storage 32, and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0169] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the emotion of the user using the emotion identification model 59 and perform identification processing using the user's emotion. In the emotion estimation function (emotion identification function) using the emotion identification model 59, various estimations and predictions regarding the user's emotion, including estimation and prediction of the user's emotion, are performed, but are not limited to such examples. In addition, the estimation and prediction of the emotion also include, for example, analysis of the emotion.

[0170] In the robot 414, the specific processing is performed by the processor 46. The storage 50 stores a specific program 60. The processor 46 reads out the specific program 60 from the storage 50, and executes the specific program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as the control unit 46A in accordance with the specific program 60 executed on the RAM 48. The robot 414 has a data generation model and an emotion identification model similar to the data generation model 58 and the emotion identification model 59, and can also perform processing similar to that of the specific processing unit 290 using these models.

[0171] Note that a device other than the data processing device 12 may have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains a processing result (such as a prediction result) using the data generation model 58 by communicating with the server device having the data generation model 58. In addition, the data processing device 12 may be a server device, or may be a terminal device (such as a mobile phone, a robot, or a home appliance) owned by a user.

[0172] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires a voice indicating a user input for the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0173] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image (e.g., still image data or video data) is input. The data generation model 58 infers the input inference data according to the instruction indicated by the prompt, and outputs the inference result in one or more data formats of voice data, text data, and image data. The data generation model 58 includes, for example, a text generation AI, an image generation AI, and a multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specification processing unit 290 performs the above-mentioned specification processing while using the data generation model 58. The data generation model 58 may be a fine-tuned model so as to output an inference result from a prompt that does not include an instruction, and in this case, the data generation model 58 can output an inference result from a prompt that does not include an instruction. In the data processing device 12 and the like, the data generation model 58 includes a plurality of types, and the data generation model 58 includes an AI other than the generation AI. The AI ​​other than the generation AI is, for example, a linear regression, a logistic regression, a decision tree, a random forest, a support vector machine (SVM), a k-means clustering, a convolutional neural network (CNN), a recurrent neural network (RNN), a generative adversarial network (GAN), or a naive Bayes, and can perform various processes, but is not limited to such examples. The AI ​​may be an AI agent. In addition, when the processes of each part described above are performed by an AI, the processes are performed in part or in whole by the AI, but are not limited to such examples. In addition, the processes performed by the AI ​​including the generation AI may be replaced with rule-based processes, and the rule-based processes may be replaced with processes performed by the AI ​​including the generation AI.

[0174] The data processing system 410 according to the fourth embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 410 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the robot 414, but may be executed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the robot 414 or an external device, etc., and the robot 414 acquires or collects information required for processing from the data processing device 12 or an external device, etc.

[0175] For example, the reading unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the robot 414. The summarizing unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the robot 414. The similarity calculation unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the robot 414. The providing unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the robot 414. The training data collection unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the robot 414. The feedback unit is realized by the specific processing unit 290 of the data processing device 12. For example, it is also realized by the control unit 46A of the robot 414. The correspondence between each part and the device or control part is not limited to the above-mentioned example, and various modifications are possible.

[0176] The emotion identification model 59 as an emotion engine may determine the emotion of the user according to a specific mapping. Specifically, the emotion identification model 59 may determine the emotion of the user according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the emotion of the robot, and the identification processing unit 290 may perform identification processing using the emotion of the robot.

[0177] FIG. 9 is a diagram showing an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive emotions are arranged. The more outside the concentric circles, the more emotions that represent states and actions that arise from a state of mind are arranged. Emotions are a concept that includes emotions and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions that occur in the brain are arranged. On the right side of the concentric circles, emotions that are generally induced by situational judgment are arranged. On the upper and lower sides of the concentric circles, emotions that are generally generated from reactions that occur in the brain and are induced by situational judgment are arranged. In addition, on the upper side of the concentric circles, emotions of "pleasure" are arranged, and on the lower side, emotions of "discomfort" are arranged. In this way, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[0178] These emotions are distributed in the 3 o'clock direction of emotion map 400 and usually fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[0179] The inside of emotion map 400 represents what is going on inside one's mind, and the outside of emotion map 400 represents behavior, so the further out you go on emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[0180] Here, human emotions are based on various balances such as posture and blood sugar level, and when these balances are far from the ideal, it indicates an unpleasant state, and when they are close to the ideal, it indicates a pleasant state. Emotions can also be created for robots, cars, motorcycles, etc., based on various balances such as posture and battery level, so that when these balances are far from the ideal, it indicates an unpleasant state, and when they are close to the ideal, it indicates a pleasant state. The emotion map may be generated, for example, based on the emotion map of Dr. Mitsuyoshi (Research on speech emotion recognition and emotion brain physiological signal analysis system, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). On the left half of the emotion map, emotions belonging to an area called "reaction" where sensation is dominant are lined up. On the right half of the emotion map, emotions belonging to an area called "situation" where situation recognition is dominant are lined up.

[0181] The emotion map defines two emotions that promote learning. The first is the negative emotion around the middle of "repentance" or "remorse" on the situation side. In other words, this is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the positive emotion around "desire" on the response side. In other words, this is when the robot has positive feelings such as "I want more" or "I want to know more."

[0182] The emotion identification model 59 inputs the user input to a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the emotion of the user. This neural network is pre-trained based on multiple learning data that are combinations of the user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in Fig. 10. Fig. 10 shows an example in which multiple emotions, "relief," "calm," and "encouraging," have similar emotion values.

[0183] In the above embodiment, an example was given in which a specific process is performed by one computer 22, but the technology disclosed herein is not limited to this, and distributed processing of the specific process may be performed by multiple computers, including computer 22.

[0184] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable computer-readable non-transitory storage medium such as a Universal Serial Bus (USB) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[0185] In addition, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 upon request from the data processing device 12.

[0186] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[0187] As the hardware resource for executing the specific process, various processors as shown below can be used. An example of the processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing the specific process by executing software, i.e., a program. Another example of the processor is a dedicated electric circuit, which is a processor having a circuit configuration designed exclusively for executing a specific process, such as a Field-Programmable Gate Array (FPGA), a Programmable Logic Device (PLD), or an Application Specific Integrated Circuit (ASIC). Each processor has a built-in or connected memory, and each processor executes the specific process by using the memory.

[0188] The hardware resource that executes the specific process may be one of these various processors, or may be a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific process may be a single processor.

[0189] As an example of a configuration using one processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a configuration using a processor that realizes the functions of the entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[0190] Furthermore, more specifically, the hardware structure of these various processors can be an electric circuit that combines circuit elements such as semiconductor elements. The specific processes described above are merely examples. It goes without saying that unnecessary steps may be deleted, new steps may be added, or the order of processes may be changed without departing from the spirit of the invention.

[0191] In the above example, the first to fourth embodiments are described separately, but some or all of these embodiments may be combined. The smart device 14, the smart glasses 214, the headset terminal 314, and the robot 414 are only examples, and they may be combined with each other, or may be other devices. In the above example, the first and second embodiments are described separately, but they may be combined.

[0192] The above description and illustrations are detailed descriptions of the parts related to the technology of the present disclosure, and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, function, action, and effect is an example of the configuration, function, action, and effect of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replaced with respect to the description and illustrations shown above, within the scope of the gist of the technology of the present disclosure. In addition, in order to avoid confusion and to facilitate understanding of the parts related to the technology of the present disclosure, the description and illustrations shown above omit explanations of technical common sense that do not require particular explanation in enabling the implementation of the technology of the present disclosure.

[0193] All publications, patent applications, and standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, and standard was specifically and individually indicated to be incorporated by reference.

[0194] (Appendix 1) A reading section that reads the answers written by students, a summarizing section that summarizes the answer sheets read by the reading section; a similarity calculation unit that calculates a similarity between the summary generated by the summarization unit and the model answer; a providing unit that converts the similarity calculated by the similarity calculation unit into a score and provides a scoring result; A system characterized by: (Appendix 2) The summary section is Analyze the content of the answer sheet, extract key points, and create a summary 2. The system of claim 1, (Appendix 3) The similarity calculation unit is Compare your answer with the model answer and calculate the similarity 2. The system of claim 1, (Appendix 4) The provision department The similarity is scored and the scoring results are provided 2. The system of claim 1, (Appendix 5) The apparatus further includes a training data collection unit for collecting data for the summary unit. 2. The system of claim 1, (Appendix 6) Further include a section for providing feedback to students based on the results of their assessments. 2. The system of claim 1, (Appendix 7) The reading unit is Infer student emotions and adjust the timing of loading attempts based on the estimated student emotions 2. The system of claim 1, (Appendix 8) The reading unit is Analyze students' past submission history and choose the best way to load them 2. The system of claim 1, (Appendix 9) The reading unit is Filter attempts based on students' current learning status and interests when loading 2. The system of claim 1, (Appendix 10) The reading unit is Infer student sentiment and prioritize which attempts to load based on the estimated student sentiment 2. The system of claim 1, (Appendix 11) The reading unit is While loading attempts, the geographic location of the student is taken into consideration to prioritize loading the most relevant attempts. 2. The system of claim 1, (Appendix 12) The reading unit is While loading attempts, analyze students' social media activity and load relevant attempts 2. The system of claim 1, (Appendix 13) The summary portion comprises: Estimate the student's emotions and adjust the way the summary is presented based on the estimated student emotions. 2. The system of claim 1, (Appendix 14) The summary portion comprises: When generating a summary, adjust the detail of the summary based on the importance of the attempt 2. The system of claim 1, (Appendix 15) The summary portion comprises: When generating summaries, apply different summarization algorithms depending on the category of the answer 2. The system of claim 1, (Appendix 16) The summary portion comprises: Estimate student sentiment and adjust the length of the summary based on the estimated student sentiment 2. The system of claim 1, (Appendix 17) The summary portion comprises: When generating summaries, prioritize summaries based on when attempts were submitted 2. The system of claim 1, (Appendix 18) The summary portion comprises: When generating summaries, adjust the order of summaries based on the relevance of attempts 2. The system of claim 1, (Appendix 19) The similarity calculation unit is Estimate the student's emotions and adjust the criteria for similarity calculation based on the estimated student's emotions. 2. The system of claim 1, (Appendix 20) The similarity calculation unit is Improve accuracy of similarity calculation by considering correlations between answers when calculating similarity. 2. The system of claim 1, (Appendix 21) The similarity calculation unit is When calculating similarity, the attribute information of the answer submitter is taken into account. 2. The system of claim 1, (Appendix 22) The similarity calculation unit is Estimate the student's emotions and adjust the order in which the results of the similarity calculation are displayed based on the estimated student's emotions. 2. The system of claim 1, (Appendix 23) The similarity calculation unit is When calculating similarity, the geographical distribution of answers is taken into account. 2. The system of claim 1, (Appendix 24) The similarity calculation unit is When calculating similarity, the accuracy of the similarity calculation is improved by referring to related literature of the answer. 2. The system of claim 1, (Appendix 25) The providing unit is Infer student sentiment and adjust how grading results are presented based on the estimated student sentiment 2. The system of claim 1, (Appendix 26) The providing unit is When providing grading results, the system selects the optimal display method by referring to the student's past performance history. 2. The system of claim 1, (Appendix 27) The providing unit is When providing marks, provide customized feedback based on the student's current learning progress 2. The system of claim 1, (Appendix 28) The providing unit is Infer student emotions and prioritize grading results based on the estimated student emotions 2. The system of claim 1, (Appendix 29) The providing unit is When providing grade results, the most appropriate display method is selected taking into account the student's geographic location information. 2. The system of claim 1, (Appendix 30) The providing unit is Analyze students' social media activity and provide feedback when providing grades 2. The system of claim 1, (Appendix 31) The training data collection unit includes: Estimate student emotions and select training data based on the estimated student emotions. 2. The system of claim 1, (Appendix 32) The training data collection unit includes: When collecting training data, reference past data to optimize the collection algorithm 2. The system of claim 1, (Appendix 33) The training data collection unit includes: Estimate student emotions and adjust the frequency of training data collection based on the estimated student emotions. 2. The system of claim 1, (Appendix 34) The training data collection unit includes: When collecting training data, weight the data based on when the attempt was submitted 2. The system of claim 1, (Appendix 35) The feedback unit is Estimate the student's emotions and tailor the feedback based on the estimated student emotions 2. The system of claim 1, (Appendix 36) The feedback unit is When providing feedback, look at the student's past feedback history to provide the most appropriate feedback 2. The system of claim 1, (Appendix 37) The feedback unit is Estimate student sentiment and prioritize feedback based on estimated student sentiment 2. The system of claim 1, (Appendix 38) The feedback unit is When providing feedback, consider student device information to provide optimal feedback 2. The system of claim 1, [Explanation of symbols]

[0195] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot

Claims

1. A reading section that reads the answers written by students, a summarizing section that summarizes the answer sheets read by the reading section; a similarity calculation unit that calculates a similarity between the summary generated by the summarization unit and a model answer; a providing unit that scores the similarity calculated by the similarity calculation unit and provides a scoring result. A system characterized by:

2. a training data collection unit for collecting data for the summary unit; 2. The system of claim 1 .

3. The system further includes a feedback unit for providing feedback to the student based on the grading result.

2. The system of claim 1 .

4. The reading unit is Infer student emotions and adjust the timing of loading attempts based on the estimated student emotions 2. The system of claim 1 .

5. The reading unit is Analyze students' past submission history and choose the best way to load them 2. The system of claim 1 .

6. The reading unit is Filter attempts based on students' current learning status and interests when loading 2. The system of claim 1 .

7. The reading unit is Infer student sentiment and prioritize which attempts to load based on the estimated student sentiment 2. The system of claim 1 .

Images