Method for evaluating statements

Abstract

[Problem] To provide a method for evaluating sentences using a computer, said method being capable of evaluating the context of a text. [Solution] A method for evaluating a sentence using a computer, the method comprising: a sentence input step for inputting a sentence to be evaluated into the computer; an in-statement word extraction step in which the computer extracts in-statement words, which are words included in the statement to be evaluated; a keyword sentence extraction step in which a computer extracts keyword sentences included in the sentence to be evaluated, the keyword sentences including one or more words in the sentence and one or more assistant words; and a statement evaluation step in which the computer evaluates the statement on the basis of the keyword sentence, the statement evaluation step including: a reading step in which evaluation information about the keyword sentence is read from the storage unit, the evaluation information being information about whether or not the statement to be evaluated, which is the keyword sentence, is correct, or information about which category the keyword sentence belongs to.

Description

Technical Field

[0001] The present invention relates to a method for evaluating sentences using a computer. Background Art

[0002] Japanese Patent No. 7049010 describes a demonstration evaluation system.

[0003] This evaluation system evaluates the content of a conversation or the person who is conducting the conversation based on the number of keywords, the number of related words, the combination of keywords, or the combination of related words. In this case, even if the conversation is incorrect in terms of the context of the text, it may be evaluated as correct if the keywords are included in the conversation.

[0004] Prior art literature

[0005] Patent Literature

[0006] Patent Document 1: Japanese Patent No. 7049010 Summary of the invention

[0007] Problems to be solved by the invention

[0008] The object of the present invention is to provide a method for evaluating sentences using a computer, which can also evaluate the context of the article.

[0009] An object of the present invention is to provide a learning support method using the above-mentioned evaluation method.

[0010] The purpose of the present invention is to provide a method for obtaining the continuation of a sentence so that role-playing can be performed after evaluating the context of the article.

[0011] Technical solutions to problems

[0012] This method relates to a method for evaluating a sentence, which can correctly grasp the context of an article by basically evaluating the sentence using key words and phrases including auxiliary words.

[0013] The method is a method for evaluating sentences based on key words or sentences by a computer.

[0014] The example of the present method includes a sentence input step (S101), a word extraction step (S102), a key word extraction step (S103), and a sentence evaluation step (S104).

[0015] The sentence input step ( S101 ) is a step of inputting a sentence as an evaluation target into a computer.

[0016] The sentence word extraction step ( S102 ) is a step in which a computer extracts words contained in a sentence that is an evaluation target, that is, words in the sentence.

[0017] The key word extraction step (S103) is a step in which a computer extracts key words contained in a sentence, which is an evaluation target. Key words include one or more words in the sentence and one or more particles.

[0018] The sentence evaluation step (S104) is a step in which a computer evaluates a sentence based on a key word or a phrase.

[0019] Examples of the sentence input step ( S101 ) include a step of inputting speech into a computer and a step of having the computer analyze the speech input into the computer to obtain a sentence as an evaluation target.

[0020] Examples of the evaluation object, i.e., the sentence, are explanatory texts or answer texts.

[0021] An example of a sentence word is a word stored in a sentence word storage unit that stores sentence words associated with a sentence that is an evaluation target.

[0022] The example of the key words and phrases is stored in a key word and phrase storage unit that stores key words and phrases associated with the evaluation target sentence.

[0023] Examples of key words and phrases may be a group of words that exist at consecutive positions or a group of words that exist at separate positions in the sentence that is the evaluation target.

[0024] An example of application of the above method is a learning support method using a computer.

[0025] Another application example of the above method relates to a method for creating a continuation sentence using a computer.

[0026] The method further includes a continuation sentence creation step (S105). The continuation sentence creation step (S105) is a step of obtaining a sentence that is a continuation sentence of the evaluation object, i.e., the sentence based on the evaluation. The computer has a continuation sentence storage unit that stores the continuation sentence corresponding to the evaluation. Then, in the continuation sentence creation step (S105), the computer uses the evaluation of the evaluation object, i.e., the sentence, to read the continuation sentence corresponding to the evaluation.

[0027] This specification also discloses a system for evaluating statements using a computer.

[0028] The present system 1 is a system for evaluating sentences by a computer and includes a sentence input unit 3 , a word extraction unit 5 , a key word extraction unit 7 , and a sentence evaluation unit 9 .

[0029] The sentence input unit 3 is an element for inputting a sentence which is an evaluation object.

[0030] The sentence word extraction unit 5 is an element for extracting words contained in the sentence which is the evaluation target, that is, words in the sentence.

[0031] The key phrase extraction unit 7 is an element for extracting key phrases contained in the sentence that is the evaluation object. The key phrases include one or more words in the sentence and one or more particles.

[0032] The sentence evaluation unit 9 is an element for evaluating a sentence based on a key word or sentence.

[0033] This specification also discloses a learning support system using the above system.

[0034] This specification also discloses a system for creating a continuation sentence using the above system. The system for creating a continuation sentence includes a continuation sentence creation unit.

[0035] This specification also discloses a program for causing a computer to execute the above-mentioned method or to cause a computer to function as the above-mentioned system, and a computer-readable non-transitory information recording medium recording the program.

[0036] Effects of the Invention

[0037] The present invention can provide a method for evaluating sentences using a computer, which can also evaluate the context of the article.

[0038] The present invention can provide a learning support method using the above-mentioned evaluation method.

[0039] The present invention can provide a method for obtaining the continuation of a sentence, so that role-playing can be performed after evaluating the context of the article. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] [ Figure 1 ] Figure 1 This is a flowchart for explaining a method for evaluating a sentence using a computer.

[0041] [ Figure 2 ] Figure 2 This is a block diagram showing an example of the configuration of a system for evaluating sentences using a computer.

[0042] [ Figure 3 ] Figure 3 This is a diagram showing an overview of the experiment in Example 1.

[0043] [ Figure 4 ] Figure 4 This is a diagram showing an implementation example of iRolePlay (registered trademark).

[0044] [ Figure 5 ] Figure 5This is a graph that replaces the attached figure and shows the verification of the memory retention degree of word memorization learning.

[0045] [ Figure 6 ] Figure 6 This is a graph instead of a drawing showing the retention rate of a word description text learned by pronunciation.

[0046] [ Figure 7 ] Figure 7 This is a concept map that illustrates the concept of role-playing.

[0047] [ Figure 8 ] Figure 8 This is a conceptual diagram showing an example of role-playing in which the AI ​​doctor's response may differ depending on the MR answer. DETAILED DESCRIPTION

[0048] Hereinafter, the embodiment of the present invention will be described with reference to the drawings. The present invention is not limited to the embodiment described below, and includes contents that can be appropriately modified within a range obvious to those skilled in the art from the embodiment below.

[0049] The present invention basically relates to a method for evaluating a sentence, which can correctly grasp the context of an article by evaluating a sentence using key words and phrases including particles. In the present method, a computer evaluates a sentence based on key words and phrases, or based on key words and phrases.

[0050] The so-called "evaluation sentence" can be any one of explaining whether the sentence is a correct answer, explaining which category the sentence belongs to, and seeking the degree of goodness of the sentence performance.

[0051] Figure 1 This is a flowchart for explaining a method for evaluating a statement using a computer. Figure 1 As shown, the example of the method includes a sentence input step (S101), a word extraction step (S102), a key phrase extraction step (S103), and a sentence evaluation step (S104). The method may further include a continuous sentence creation step (S105). S means step (step).

[0052] Figure 2 FIG. 1 is a block diagram showing an example of a system for evaluating a statement using a computer. Figure 2As shown, the system 1 includes a sentence input unit 3, a word extraction unit 5 in the sentence, a key phrase extraction unit 7, and a sentence evaluation unit 9. The system may further include any one or more of a word storage unit 6 in the sentence, a key phrase storage unit 8, and a sentence evaluation information storage unit 10. The system 1 may further include a continuation sentence creation unit 11. In the case where the system 1 has the continuation sentence creation unit 11, the system 1 may further include a continuation sentence storage unit 13. The sentence input unit 3 is used to input the evaluation object, that is, the element of the sentence. The word extraction unit 5 in the sentence is used to extract the words contained in the evaluation object, that is, the sentence, that is, the elements of the words in the sentence. The key phrase extraction unit 7 is used to extract the key phrases contained in the evaluation object, that is, the sentence. The sentence evaluation unit 9 is used to evaluate the sentence based on the key phrases, or based on the key phrases. The continuation sentence creation unit 11 is used to obtain the sentence that continues the evaluation object, that is, the sentence, that is, the element of the continuation sentence. The system 1 is a computer system for executing the above method.

[0053] The computer has an input unit, an output unit, a control unit, a computing unit, and a storage unit, and each element is connected by a bus or the like, and is configured to be able to send and receive information. For example, the storage unit can store both a control program and various information. When predetermined information is input from the input unit, the control unit reads the control program stored in the storage unit. Then, the control unit appropriately reads the information stored in the storage unit and transmits it to the computing unit. In addition, the control unit appropriately transmits the input information to the computing unit. The computing unit uses the received various information to perform computing processing and stores it in the storage unit. The control unit reads the computing results stored in the storage unit and outputs it from the output unit. In this way, various processing or various processes are performed. Those that perform these various processing are the various units or units. The computer has a processor, and the processor can also perform various functions or various processes. The computer can be independent. The computer can also be distributed to a server and a terminal as part of the function. In the above case, the server and the terminal are preferably configured to be able to send and receive information through a network such as the Internet or an intranet.

[0054] The computer can also receive information transmitted wirelessly from various external devices or information output as optical signals, convert it into electrical signals for use. In addition to on-off keying, optical signals or electrical signals can also include various modulation signals. In addition, the storage unit can store various input information based on the presence or absence of electric charge, or based on multiple physical states such as quantum states.

[0055] When extracting words or key words, the computer compares the words or key words stored in the storage unit with the input sentence. When the words or phrases contained in the input sentence are consistent with the words or key words stored in the storage unit, the consistent words or key words can also be extracted. Alternatively, machine learning can be performed in advance to construct a learning model, and the input sentence and the words or key words stored in the storage unit are input into the learning model to output specific words or key words.

[0056] For example, the computer may also be equipped with a machine learning engine, through machine learning, using teacher data to construct a model for completing learning, and inputting various information into the model for completing learning to obtain various information. The various information obtained may also be stored in a storage unit. The model for completing learning can improve its accuracy by feeding back the various information obtained. Again, by repeatedly inputting teacher data and correct answer data, teacher data and incorrect answer data, the accuracy of the completed learning model can be improved. For example, in the present invention, various sentences and evaluation values ​​may also be used as teacher data, and machine learning and learning models may be constructed to obtain the evaluation of sentences. Again, one or more key words and evaluation values ​​(evaluation values ​​for how excellent the sentence is) may also be used as teacher data to carry out machine learning and construct learning models to obtain the evaluation of sentences. Again, for sentences that have obtained evaluation values, one or more key words and evaluation values ​​may also be obtained, and one or more key words and evaluation values ​​may be input into the model for completing learning to obtain evaluation values. Then, the evaluation values ​​obtained in this way may also be compared with the evaluation values ​​of the sentences themselves, and the comparison results may be fed back as teacher data to improve the accuracy of the learning model.

[0057] The sentence input process (S101) is a process for inputting the evaluation object, i.e., the sentence, into the computer. It is sufficient as long as the sentence input unit 3 in the system 1 inputs the evaluation object, i.e., the sentence into the system. For example, the sentence can be input into the computer using an input device (such as a keyboard and a mouse). In addition, a voice input device such as a microphone can be used to input the voice that will be the source of the sentence into the computer. The voice input into the computer can also be voice recognized by a known method to obtain the evaluation object, i.e., the sentence. In this case, the computer performs an operation to digitize the voice, and appropriately stores the digitized voice in a storage unit. Then, the computer can also have a voice analysis unit that analyzes the words or terms contained in the voice. The voice analysis unit can read the voice analysis program and the digitized voice from the storage unit, analyze the voice, and obtain the evaluation object, i.e., the sentence. The obtained evaluation object, i.e., the sentence, can also be appropriately stored in the storage unit.

[0058] When the evaluation object, i.e., the sentence, is input into the computer, in the case where the sentence is associated with something, information about the thing associated with the sentence can also be input into the computer. In this way, the system can use various information about the thing associated with the sentence to perform various operations. Examples of operation processing are voice analysis, word extraction, sentence extraction, and various operations in the analysis of consecutive sentences. Examples of things associated with the sentence are presentation materials, pages of presentation materials, reports, conference materials, medicine, questions, questions, questionnaires, troubleshooting collections, telephone answer records, and chat robot operation manuals. For example, when the presentation material is started at a certain terminal, information associated with the presentation material can be input into the system. In this case, as described later, it will become possible to use a dictionary or storage unit associated with the presentation material. The evaluation object, i.e., the sentence, can be a variety of sentences. Examples of the evaluation object, i.e., the sentence, are explanatory texts or answer texts. That is, the system can be used to evaluate the sentence when someone explains to others, or automatically evaluate the answer text related to a certain question. Furthermore, the system can automatically interpret input questions and efficiently create answers to them.

[0059] For example, a MR (Medical Representative) inputs information about a medicine A into the system, or starts explanatory materials about a medicine A and displays the explanatory materials on the display unit. Then, the system stores the subsequent utterances or conversations as relevant information about medicine A in the storage unit.

[0060] Then, the voice of a certain MR “I am going to sing a little song” is input into the microphone.

[0061] The system analyzes the above speech with reference to the term conversion dictionary for medicine A (the conversion term storage unit for medicine A). In the term conversion dictionary for medicine A, words including "pregnant woman", "pregnancy", "possible", "female", "treatment", "treatment", "beneficial", "risk", "exceeding", "judgment", "situation", and "administration" are stored as words with high priority. For example, since "female" is stored as a word with a higher priority than "female surname" in this dictionary, "female" will be read first for the above input speech "nǚxìng".

[0062] In this way, in the system, a sentence is input regarding drug A: "Please administer the drug to women who are pregnant or may become pregnant when it is judged that the benefits of the treatment outweigh the risks."

[0063] The word extraction process (S102) in a sentence is a process in which a computer extracts words contained in the evaluation object, i.e., the sentence, i.e., the words in the sentence. The word extraction unit 5 in a sentence extracts words contained in the evaluation object, i.e., the sentence, i.e., the words in the sentence. The system has a dictionary about words in the sentence, and can also read the evaluation object, i.e., the sentence, from the storage unit, and extract the words in the sentence from the evaluation object, i.e., the sentence by comparing it with the words in the dictionary. In addition, when information about things associated with the sentence is input into the computer, by using a dictionary about things associated with the sentence, i.e., a dictionary for extracting words in the sentence (the word storage unit 6 in the sentence), it becomes easy to extract words in the sentence. In this case, an example of a word in the sentence is a word stored in a word storage unit in the sentence, and the word storage unit in the sentence stores words in the sentence associated with the evaluation object, i.e., the sentence. The system (the word extraction unit in the sentence 5) reads the words in the sentence stored in the word storage unit 6 in the sentence, and compares it with the evaluation object, i.e., the sentence, so as to extract the words in the sentence contained in the evaluation object, i.e., the sentence. The words in the sentence extracted by the system can also be appropriately stored in the storage unit by the system. The system can obtain the words in the sentence in this way.

[0064] The system has a word storage unit 6 for sentences related to medicine A. In the word storage unit 6 for sentences related to medicine A, words including "pregnant woman", "pregnancy", "female", "benefit", "risk", "exceed", and "drug administration" are stored as words in sentences with high priority.

[0065] Therefore, from the sentence about medicine A input into the system, "For women who are pregnant or may become pregnant, please administer the medicine when it is judged that the benefits of the treatment outweigh the risks", the words "pregnant woman", "pregnancy", "female", "benefit", "risk", "outweigh", and "administer" in the sentence are extracted. In addition, in the case of the voice input system, although the term conversion dictionary and the sentence word storage unit 6 can be the same, they are preferably different elements. That is, the term used for term conversion and the word extracted as the sentence word can be the same, but they can also be different due to different purposes.

[0066] The key phrase extraction process (S103) is a process in which a computer extracts key phrases contained in the evaluation object, i.e., the sentence. For example, the key phrase extraction unit 7 extracts key phrases contained in the evaluation object, i.e., the sentence. The key phrase contains one or more words in the sentence, and contains one or more auxiliary words. The key phrase may also contain one or more words in the sentence. Then, in the case where there are more than two words in the sentence contained in a key phrase, these words in the sentence may be words in the sentence that exist continuously in the evaluation object, or there may be one or more other words in the sentence or nouns before these words in the sentence appear.

[0067] In the above-mentioned word extraction process (S102) in the sentence, the words contained in the evaluation object, i.e., the sentence, are extracted. However, even if only words are used, there are situations where it is impossible to judge whether it is a correct sentence (explanatory text or answer text). Therefore, the system 1 extracts key words and phrases through the key word and phrase extraction process (S103). Key words and phrases refer to sections used to evaluate whether a sentence is a correct explanatory text or a correct answer text. Key words and phrases usually contain two or more words. Key words and phrases can be words and phrases stored in the key word and phrase storage unit 8 that stores key words and phrases associated with the evaluation object, i.e., the sentence, or words and phrases extracted from the evaluation object, i.e., the sentence as a word A in a certain sentence and a particle associated with the word A in the sentence contained in the evaluation object, i.e., the sentence. In addition, the so-called word extraction process (S102) in the sentence and the key word and phrase extraction process (S103) are for convenience, and these processes can also be performed simultaneously. In this case, words in the sentence are also extracted when the key word and phrase extraction process (S103) is performed.

[0068] In the former case, the system (keyword sentence extraction unit 7) reads one or more key words and phrases associated with the evaluation object, i.e., the sentence and stored in the key word sentence storage unit 8, performs a comparison operation with the evaluation object, i.e., the sentence, and extracts the key words and phrases contained in the evaluation object, i.e., the sentence. The system may also appropriately store the extracted key words and phrases contained in the evaluation object, i.e., the sentence in the storage unit.

[0069] In the latter case, for example, the system (keyword sentence extraction unit 7) stores the keyword sentence words stored in association with the evaluation object, i.e., the sentence, in the keyword sentence storage unit 8. Then, the system may also read the keyword sentence words from the keyword sentence storage unit 8, and read the keyword sentence words and the particle following them from the evaluation object, i.e., the sentence, to extract the keyword sentence from the evaluation object, i.e., the sentence. The system may also appropriately store the extracted keyword sentence contained in the evaluation object, i.e., the sentence, in the storage unit.

[0070] For example, medicine A is administered to pregnant women or women who may become pregnant when the benefits outweigh the risks. However, by simply extracting the words in the sentence, there is a possibility that the evaluation object, that is, the sentence, means that the risks outweigh the benefits. Such a sentence is not correct as an explanatory text about medicine A. Therefore, the key word storage unit 8 stores key words ("therapeutic benefits"... "outweigh the risks"). In addition, the key word storage unit 8 may also store key words ("the benefits outweigh"), ("the risks outweigh"), ("the risks are low"). The key word storage unit 8 may also store the words "pregnant women" and "pregnancy" in the sentence for the medication object, and store key words containing any of the above for the efficacy.

[0071] Therefore, from the sentence "For women who are pregnant or likely to become pregnant, please administer the drug only when it is judged that the benefits of the treatment outweigh the risks", for example, key words and phrases are extracted as the subject (either or both of "pregnant woman" and "pregnancy") and about the efficacy of the drug ("therapeutic benefits", "outweigh the risks").

[0072] The sentence evaluation process (S104) is a process in which a computer evaluates a sentence based on key words and phrases. For example, the sentence evaluation unit 9 evaluates the evaluation object, i.e., the sentence, based on the key words and phrases. The system 1 may further include a sentence evaluation information storage unit 10. The sentence evaluation information storage unit 10 stores, for example, key words and phrases and evaluation information about the key words and phrases. Examples of evaluation information are information about whether the key words and phrases as the evaluation object, i.e., the sentence, are correct, information about which category the key words and phrases belong to, and evaluation values ​​associated with the key words and phrases. The system (sentence evaluation unit 9) uses the key words and phrases extracted in the key word and phrase extraction process (S103) to read the evaluation of the extracted key words and phrases from the sentence evaluation information storage unit 10. In this way, the system 1 can evaluate the evaluation object, i.e., the sentence.

[0073] The evaluation value associated with the key words and phrases may also be an evaluation value (score) associated with each key word and phrase and stored in the storage unit. For example, it may also be configured so that the higher the evaluation value, the more appropriate key words and phrases are used. As a result, it may also be configured so that the evaluation value of a sentence using appropriate key words and phrases becomes higher and a high evaluation is obtained. In the case of obtaining multiple evaluation values, these evaluation values ​​may also be summed up to obtain the evaluation value of the sentence (an indicator of how excellent it is).

[0074] For example, the sentence evaluation information storage unit 10 is associated with medicine A, and is associated with key words and phrases including (either or both of "pregnant women" and "pregnancy"), ("benefit of treatment"), and ("outweighs risk"), and stores "correct" as an evaluation. In this case, the sentence evaluation unit 9 uses the above key words and phrases to read the evaluation "correct" from the sentence evaluation information storage unit 10. The sentence evaluation unit 9 may also store the read evaluation "correct" as the evaluation object, that is, the evaluation of the sentence in the storage unit. In addition, the system 1 may also output the evaluation "correct". In addition, it is assumed that the sentence evaluation information storage unit 10 is associated with key words and phrases including (either or both of "pregnant women" and "pregnancy"), ("benefit of treatment"), and ("outweighs risk"), and stores the category "medication prescription". In this case, the system 1 uses the extracted key words and phrases to associate with medicine A, and reads the category "medication prescription" as one of the evaluations. In this case, the system 1 is associated with medicine A, and the evaluation "correct" can also be obtained for the category "medication prescription".

[0075] The word storage unit 6 in the sentence can also store the classification of words in association with the words in the sentence. Examples of classification are affirmation, negation and ambiguity. Table 1 shows the classification and examples of words in the sentence for each classification. In addition, in the case where the key words and phrases include words in the sentence, the key words and phrases storage unit 8 can also store the classification of the words in the sentence. The sentence evaluation information storage unit 10 can also store the classification of the words in the sentence included in the key words and phrases. The system (sentence evaluation unit 9) uses the key words and phrases extracted in the key words and phrases extraction step (S103) to read the evaluation of the key words and phrases extracted from the sentence evaluation information storage unit 10. For example, the system 1 uses the extracted key words and phrases to read the information about the classification of the words in the sentence included in the key words and phrases (for example, affirmation, negation and ambiguity) from the sentence evaluation information storage unit 10 (key words and phrases storage unit 8 or word storage unit 6 in the sentence). Then, in the case where the classification of the words in the sentence included in the read key words and phrases is negative (or ambiguous), the evaluation of the evaluation object sentence can also be evaluated as wrong (inappropriate). This can prevent situations where ambiguous statements are used, and can improve the ability to answer.

[0076] In addition, an invention in which words in a sentence are extracted from a sentence to be evaluated, rather than from a keyword sentence, and the sentence to be evaluated is evaluated based on the classification of the words in the extracted sentence, is also an invention described in this specification. In this invention, the keyword sentence extraction step (S103) and the keyword sentence extraction unit 7 are no longer necessary.

[0077] [Table 1]

[0078]

[0079] The continuation sentence creation process (S105) is a process for obtaining a sentence, i.e., a continuation sentence, that is, a sentence that is a continuation of the evaluation object, i.e., a sentence, based on the evaluation. The continuation sentence creation unit 11 can also obtain a sentence, i.e., a continuation sentence, that is, a sentence that is a continuation of the evaluation object, i.e., a sentence. For example, the computer has a continuation sentence storage unit 13 that stores continuation sentences corresponding to the evaluation. Then, in the continuation sentence creation process (S105), the computer uses the evaluation of the evaluation object, i.e., the sentence, to read the continuation sentence corresponding to the evaluation. Information about the thing associated with the sentence is input into the computer, and an appropriate continuation sentence can also be read from the continuation sentence storage unit 13 corresponding to the information about the thing associated with the sentence. For example, when the information about the thing associated with the sentence is "a certain question" and the evaluation is "correct" (correct answer), the evaluation "very good answer" and the explanation about the question are stored in the continuation sentence storage unit 13. The system (continuation sentence creation unit 11) reads the evaluation from the storage unit, reads the continuation sentence corresponding to the evaluation, that is, the evaluation "very good answer" and the explanation about the question from the continuation sentence storage unit 13, and stores them appropriately in the storage unit. The read continuation sentence can also be output appropriately. In the case where the information about the thing associated with the sentence is a chat robot about a certain product, and the evaluation is "determined topic" (category), it is sufficient to read the answer to the evaluation object, that is, the sentence, from the continuation sentence storage unit 13 based on the evaluation, and obtain an appropriate answer. In this way, the system can obtain a continuation sentence. For example, in the case where the evaluation of the sentence is "interrogative sentence" or "question", it will become possible to obtain an answer to the sentence.

[0080] When system 1 inputs the sentence "For women who are pregnant or may become pregnant, please administer the medicine only when it is determined that the benefits of the treatment outweigh the risks" about medicine A, system 1 extracts words or key phrases from this sentence, associates it with medicine A, and obtains the evaluation "correct" for the category "medication prescription". For example, in the continuation sentence storage unit 13, the continuation sentence "The frequency of administration of medicine A is one pill a day" is stored in association with the above-mentioned association with medicine A, the category "medication prescription", and the evaluation "correct". The system (continuation sentence creation unit 11) can also read the above evaluation from the storage unit, read the above continuation sentence from the continuation sentence storage unit 13, and store it in the storage unit. In addition, the system can also output the above continuation sentence. In this way, it will become "The frequency of administration of medicine A is one pill a day." and be output as the continuation sentence of the sentence.

[0081] Learning support using computers

[0082] The use example of the above method and system is a learning support method using a computer. This specification also discloses a learning support system using the above system. The invention of this form can also be provided as a learning support application in a downloadable form. In this case, various dictionaries about each question are stored in a server or installed in a mobile terminal. For example, this system stores a certain question A from a storage unit and displays it on the display unit of the terminal. The user answers the question with voice. The user's voice is input into the terminal via the input unit of the terminal. The input voice (user's answer) is stored in the terminal as a sentence of evaluation object. The system including either or both of the terminal and the server has a term conversion dictionary about the question A, a word storage unit 6 in the sentence, a key phrase storage unit 8, a sentence evaluation information storage unit 10, and any one or more of the continuation sentence storage unit 13. Therefore, the system can evaluate the user's answer. In addition, the system can read the continuation sentence corresponding to the user's answer and display it on the display unit of the terminal. As described later, compared with simply memorizing words, memorizing sentences can not only retain the memory, but also improve communication skills. Therefore, the learning support system or the learning support method of the present invention can provide users with high learning effects.

[0083] Chatbot or role-playing system

[0084] Examples of the use of the above-mentioned system are chat robots or role-playing systems.

[0085] The chat robot or role-playing system has one or more of a word conversion dictionary for a certain product or service, a word storage unit 6 in a sentence, a key phrase storage unit 8, a sentence evaluation information storage unit 10, and a follow-up sentence storage unit 13. In this way, when there is an input to the system for a question or the like inputted by voice input via a telephone or automatically inputted via the Internet, the system searches for key phrases, obtains an evaluation of the product or service, and can appropriately obtain a follow-up sentence. In this way, it becomes possible to automatically output an appropriate answer to the question. In this case, for example, for an input of "the wheel of product AA does not move.", the system can also use the storage unit for product AA to obtain a follow-up sentence "Is the power on?" and output it to the questioner. In addition, when the user inputs "yes" for this answer, the system can also obtain a follow-up sentence "Please clean the periphery of the wheel with a brush. If the wheel still does not move, please inform the service center and we will go to collect it." and output it to the questioner.

[0086] Conversation or presentation support system

[0087] The use example of the above system is a conversation or demonstration support system. For example, the system has any one or more of the following: explanatory materials, demonstration materials, a term conversion dictionary for each page of the demonstration materials, a word storage unit 6 in a sentence, a key phrase storage unit 8, a sentence evaluation information storage unit 10, and a continuation sentence storage unit 13. In this way, in the case of a conversation or demonstration based on the materials or demonstration materials, key phrases can be obtained, and the use of the key phrases can obtain appropriate evaluations from the sentence evaluation information storage unit 10, and appropriate continuation sentences can be obtained from the continuation sentence storage unit 13. In the above situation, the evaluation can be something like "very good", "excellent", or "good". An example of a continuation sentence can be "Instead of AAA, using BBB will get a higher evaluation." This kind of explanation helps to improve the conversation or demonstration.

[0088] This specification includes instructions for causing a computer to execute the above method, and also discloses a program for causing a computer to function as the above system, or a computer-readable non-temporary information recording medium (such as a CD-ROM, DVD, SD card, and USB memory) recording the program.

[0089] Example 1

[0090] An investigation of a learning model using vocal learning

[0091] It can be imagined that after the three steps of "words and explanations are retained in memory", "replacement of retained memories with one's own language" and "ability to provide easily understandable explanations", it will become possible to provide easily understandable explanations. In this embodiment, the verification results of "words and explanations are retained in memory" and "replacement of retained memories with one's own language" are examined. The following three verification items were verified in this study. In the following experiments, the application for realizing a learning support system using a computer in this specification was used as a role-playing application. The overview is shown in Figure 3 .

[0092] (i) Verification of memory retention of word recitation learning

[0093] (ii) Measurement of retention rate of word description texts learned using vocalization

[0094] (Ten men and women aged between 20 and 50 were selected as subjects and asked questions related to pharmaceutical products. They were divided into two groups and the same content was administered.)

[0095] (iii) Measuring retention and learning time on iPad(TM) role-playing applications

[0096] (6 male and female subjects aged between 30 and 50 were selected to answer questions related to the Internet.)

[0097] Verification of memory retention in word recitation learning

[0098] First, in order to verify the extent to which words can be memorized by silent reading, we selected 21 professional terms related to the pharmaceutical industry that are not used in daily life and created questions. The summary of the words and the description is shown in Table 2.

[0099] [Table 2]

[0100]

[0101] On the first day of the empirical experiment, A4 sheets of paper with words and explanations were given to the subjects, who were asked to memorize the words in ten minutes other than reading them aloud (e.g. visually, writing them on paper, etc.). Ten minutes later, the explanations for the words were verbally read out as interrogative sentences to the subjects. In the case of incorrect answers, the correct answers were verbally fed back to the subjects on the spot. Only the test was conducted on the second and third days to observe how the memory retention changed, and the number of correct answers of the subjects was counted.

[0102] Measuring the retention rate of word description texts learned using vocalization

[0103] After the three-day word test, in order to measure the extent to which the subjects could recall the explanation text for the words, a test was conducted to see if they could speak the word explanation text. As part of the test, the subjects were asked to verbally ask "Please explain ○○" for 21 words, and the subjects' answers were recorded (for the sake of efficiency, the iPad (registered trademark) voice recognition function was used to record). Afterwards, as shown in Table 3, the same test was repeated regularly over a total of 19 days (due to business reasons, the implementation schedule was not uniform, please understand), and the results were recorded to measure the memory retention rate of the word explanation text.

[0104] [Table 3]

[0105]

[0106] In order to seek the memory retention effect of vocal learning, the subjects were given three vocal learning sessions at intervals of about 3 to 4 days from the first day of measurement. The content of the vocal learning is: the A4 paper with words and descriptions shown in Table 1 is given to the subjects in the same way as the above verification, and they are asked to read the words and descriptions aloud and memorize them within 10 minutes on the spot. The "memory test" in Table 2 refers to a test of how many words and descriptions can be said, and the "confirmation test" refers to a test used to confirm how many words and descriptions can be said immediately after vocal learning.

[0107] Measuring retention and learning time with iPad(R) role-playing apps

[0108] The memory and learning time of words and explanatory texts were measured using the iPhone(registered trademark) and iPad(registered trademark) application "iRolePlay(registered trademark)" developed by Interactive Solutions(registered trademark).

[0109] iRolePlay (registered trademark) is a new-age retraining tool that can "learn the ability to explain and propose" by objectively analyzing and evaluating the user's speech content and providing suggestions through AI using voice recognition and the neural network engine equipped in iPhone (registered trademark) and iPad (registered trademark). It consists of two modes (learning mode / challenge mode). The former displays the answers to questions read out by voice and allows voice training ( Figure 4 The latter can acquire practical explanation skills by practicing answering questions read out loud.

[0110] The subjects were asked to solve 21 questions about Internet words that they were not familiar with in daily life. Two courses were prepared: one for learning words and one for learning explanatory texts. The subjects were given iPads (registered trademark) and were urged to study independently during the gaps between tasks within three days. Figure 4 As shown, the subjects learned in the learning mode, that is, in the form of role-playing in which the question sentences and answers are displayed on the iPad (registered trademark). In this application, the next question can be asked by correctly saying the correct word or explanation. The test method is a challenge mode that uses voice questions to ask questions about words or word explanations, and determines whether the questions are answered correctly through voice recognition.

[0111] Experimental results and investigation

[0112] Verification of memory retention in word recitation learning

[0113] Although it has been mentioned that reciting words aloud is effective, the correct answer rate is 78.6% even if the memorization method is visual or written on paper within 10 minutes. In addition, the correct answer rate on the second day without setting a recitation time is 81.9%, and the correct answer rate on the third day is 87.6%, showing an upward trend ( Figure 5 ).

[0114] Investigation

[0115] As can be seen from the "Ebbinghaus Forgetting Curve", it is said that general memory is forgotten over time. However, in this verification, the answers to incorrect answers were given to the subjects on the spot, which was also easy to retain as memory, and it is believed that this is the reason for the increase in test scores. In the actual listening to the subjects, three people felt that the correct answers to the questions they got wrong in the previous test were still retained in their memories. This shows that in vocabulary memory, immediately giving feedback on the correct answer to incorrect answers is effective for memory retention.

[0116] Measuring the retention rate of word description texts learned using vocalization

[0117] Although the correct answer rate for the word test was 87.6%, the correct answer rate for the word explanation test was 43.3%, which was less than half. This result proves that even if you can answer a word in a question-and-answer format, it does not mean that you can explain the word.

[0118] Then, let's look at the results of memorizing the word description text. In the word description test, if we compare the results of visual learning and vocal learning, the latter has a 64% higher correct answer rate than the former. In addition, the change in the correct answer rate of the word description test, that is, the change in the retention rate of memory, is as follows: Figure 6 As shown in the figure, the correct answer rate has been increasing throughout the whole schedule by regularly conducting vocalization training. Even after 19 days from the initial explanation test, the answer rate is still greater than 70%.

[0119] Investigation

[0120] The subject can utter answers close to the example answers by repeating the vocalization learning. Figure 3 (Example of explanation shown). When analyzing the answers recorded by the subject's voice recognition, the proportion of professional terms in the answers increased through repeated vocalization learning. When listening to the subjects, there are such sounds, even if they are professional terms that are not usually used, speaking them out or listening to the sounds with the ears has an advantage when reading them out loud and answering them. As the number of tests increases, it gradually becomes possible to speak the answers fluently while mixing in professional terms, and at the same time the correct answer rate also slowly increases. In order to develop the ability to explain, it is necessary to express appropriately and sometimes talk to the other party while mixing in professional terms. In that regard, this verification can be thought of as professional terms being retained in memory through vocalization, and getting used to saying the words, so that it becomes possible to explain fluently as one's own language.

[0121] In addition, this verification also confirmed that the explanation text memorized once is easy to be remembered for a long time. If we look at the answer results of each 21 questions, the correct answer rate of questions that have been answered once does not decrease even after a few days. When listening to the subjects, two people felt that they were able to correctly explain the words in the word explanation test by memorizing the words that would become keywords in the explanation text in advance.

[0122] Measuring retention and learning time with iPad(R) role-playing applications

[0123] Although the iPad (registered trademark) role-playing application is still under verification, the change in the correct answer rate of the word explanation text shows the same trend. As an educator, the only time spent on measurement is sending the application, which is the biggest advantage. In the above verification, if all the test questions, feedback on the correct answers to wrong answers, and scoring of the test results are taken into account, it will take 2480 minutes to present the results of 10 people.

[0124] Another advantage is that the subjects can study at any time during their spare time. Since the application log can be obtained, the study time can be accurately measured, which is expected to provide useful study guidance for educators. To what extent the subjects can effectively use their spare time and shorten the period in order to improve the correct answer rate of vocabulary explanation texts is a topic to be verified in the future.

[0125] In this study, the ability to explain is defined as "the ability to explain to the other party with appropriate expressions", and it is believed that there are three steps to retain the ability to explain. In the verification of the memory retention of word recitation learning, it was shown that immediate feedback of the correct answer to an incorrect answer is effective for memory retention. On the other hand, it was proved that even if the answer to a word in a question-and-answer format can be achieved, it still does not mean that the explanation of the word can be achieved.

[0126] The retention rate of word explanation texts using vocalization learning was measured, and it was found that the retention rate of word explanation texts increased with repeated vocalization learning. In addition, the word explanation test contents of the subjects showed that the correct answer rate increased with the three steps.

[0127] Example 2

[0128] role play

[0129] Figure 7 This is a conceptual diagram that explains the concept of role-playing. The "brackets" are the categories of evaluation.

[0130] Figure 8 This is a conceptual diagram showing an example of role-playing in which the AI ​​doctor's response may differ depending on the MR answer.

[0131] The developed application is actually installed on the computer, and the computer is made to execute instructions based on the application. In the following, Q is the input evaluation object sentence, and A is the subsequent sentence to it.

[0132] QWhat is the patient's medical history?

[0133] A suffered a myocardial infarction three years ago

[0134] QWhat are the other symptoms?

[0135] A seems to have a headache and dizziness

[0136] QObservations before administration?

[0137] A thinks it is stressful. Q What is your current coping method?

[0138] ADiet therapy and taking antihypertensive drugs

[0139] QWhat is the connection with Amprolium?

[0140] AAmpro tablets are easy to prescribe because they have few side effects on the respiratory system.

[0141] In this example, if the doctor inputs a question, the appropriate follow-up sentence is displayed on the display unit by first referring to the term conversion dictionary about the medical scene and the target patient, the word storage unit 6 in the sentence, the key sentence storage unit 8, the sentence evaluation information storage unit 10, and the follow-up sentence storage unit 13. In the process, "amprodine" appears as a word in the sentence. Therefore, in this system, the appropriate follow-up sentence is obtained by referring to the key sentence storage unit 8, the sentence evaluation information storage unit 10, and the follow-up sentence storage unit 13 about amprodine. Since this system performs key sentence analysis, the chatbot and role-playing become conversational styles, and it will become a natural conversation even when using a computer. In addition, unlike the conversation system that uses deep learning so far, it will become a system that can be constructed very easily and without time in a conversation in a narrow area. Role-playing using this system is effective in cultivating developmental conversation skills.

[0142] Industrial Applicability

[0143] This method can be used in the information industry, education industry, etc.

[0144] Description of Reference Numerals

[0145] 1: System

[0146] 3: Sentence input section

[0147] 5: Word extraction part in the sentence

[0148] 6: Word storage part in the sentence

[0149] 7: Keyword extraction department

[0150] 8: Keyword phrase storage unit

[0151] 9: Sentence Evaluation Department

[0152] 10: Sentence evaluation information storage unit

[0153] 11: Continuation statement creation section

[0154] 13: Continuation statement storage unit

Claims

1. A method for evaluating a sentence using a computer, It is characterized in that Include: A sentence input step of inputting the evaluation object, i.e., the sentence, into the computer; A word extraction step in a sentence, wherein the computer extracts words contained in the sentence, i.e., the evaluation object; A key word extraction step, wherein the computer extracts key words contained in the evaluation object, i.e., the sentence, wherein the key words contain one or more words in the sentence and one or more auxiliary words; as well as a sentence evaluation step, wherein the computer evaluates the sentence based on the key words and sentences, The sentence evaluation step includes a reading step of reading evaluation information about the key words and sentences from a storage unit, The evaluation information is information on whether the key word or sentence as the evaluation object, ie, the sentence, is correct, or information on which category the key word or sentence belongs to.

2. The method for evaluating a sentence according to claim 1, It is characterized in that The sentence input process comprises: The step of inputting speech into the computer; and The computer analyzes the speech to obtain the sentence as the evaluation object.

3. The method for evaluating a sentence according to claim 1, It is characterized in that The evaluation object, i.e., the sentence, is an explanatory text or an answer text.

4. The method for evaluating a sentence according to claim 1, It is characterized in that The evaluation information is information on which category the key word phrase belongs to.

5. The method for evaluating a sentence according to claim 1, It is characterized in that The key words and phrases may be a group of words that exist at consecutive positions or may be a group of words that exist at separate positions in the sentence that is the evaluation target.

6. A learning support method using the computer, It is characterized in that The method includes the step of evaluating the sentence as the evaluation object based on the method according to claim 1.

7. A system (1) for evaluating a sentence using a computer, It is characterized in that Include: A sentence input unit (3) which inputs the evaluation object, i.e., the sentence; A sentence word extraction unit (5) extracts the words contained in the evaluation object, i.e., the sentence, i.e., the words in the sentence; A keyword and phrase extraction unit (7) extracts the keyword and phrase contained in the evaluation object, i.e., the sentence, wherein the keyword and phrase includes one or more words in the sentence and one or more auxiliary words; as well as A sentence evaluation unit (9) is used to evaluate the sentence, which is the evaluation object, based on the key words or sentences, or to read evaluation information about the key words or sentences from a storage unit based on the key words or sentences. The evaluation information is information on whether the key word or sentence as the evaluation object, ie, the sentence, is correct, or information on which category the key word or sentence belongs to.

8. A program for causing a computer to execute a method of evaluating a statement, It is characterized in that The method comprises: A sentence input step of inputting the evaluation object, i.e., the sentence, into the computer; A word extraction step in a sentence, wherein the computer extracts words contained in the sentence, i.e., the evaluation object; A key word extraction step, wherein the computer extracts key words contained in the evaluation object, i.e., the sentence, wherein the key words contain one or more words in the sentence and one or more auxiliary words; as well as a sentence evaluation step, wherein the computer evaluates the sentence as the evaluation object based on the key words and sentences, The sentence evaluation process includes a reading step of reading evaluation information about the key words and phrases from a storage unit, The evaluation information is information on whether the key word or sentence as the evaluation object, ie, the sentence, is correct, or information on which category the key word or sentence belongs to.

9. A computer-readable non-transitory information recording medium, It is characterized in that The program according to claim 8 is stored.

Images