Text proofreading support device, text proofreading method, and computer program

Abstract

This will make it easier than before to check for errors in texts written by learners in languages ​​other than their native tongue. [Solution] An English composition area 500 is displayed on the screen, which includes a first box 501 for inputting the main body, a second box 502 for inputting the predicate verb, and third boxes 503 to fifth boxes 505 for inputting modifying phrases related to the predicate verb. The original text, formed by arranging the words and phrases entered into the first box 501 to fifth box 505 in box order, is then proofread by AI (Artificial Intelligence) to obtain a proofread text, which is then displayed on the screen.

Description

Technical Field

[0001] The present invention relates to a technique for checking errors and the like in a sentence created by a learner, and particularly to a technique for checking errors and the like in a sentence created in a language other than the native language.

Background Art

[0002] Conventionally, there are various problems in learning or using a foreign language. Among them, one of the major problems when a learner whose native language is Japanese writes an English composition is the difference in word order between Japanese and English.

[0003] In order to solve this problem, a method called the "semantic order" method as described in Non-Patent Document 1 has been proposed. Note that "semantic order" is a registered trademark.

[0004] According to this method, an English sentence is completed by entering phrases corresponding to each heading into five horizontally arranged boxes (regions) provided on a sheet and labeled in order with the headings "Who", "Does (Do)", "Who / What", "Where", and "When". For example, when writing an English composition for "My son plays baseball on weekends.", one should enter "My son", "plays", "baseball", and "on weekends." into the boxes corresponding to "Who", "Does (Do)", "Who / What", and "When", respectively. The phrases entered into each box arranged in the order of the boxes is the target English sentence.

[0005] One of the advantages of this method is that it enables writing an English composition without paying much attention to various grammars of English represented by five sentence patterns (the first sentence pattern to the fifth sentence pattern).

Prior Art Documents

Non-Patent Documents

[0006]

Non-Patent Document 1

[0007] However, when writing a sentence like "My son plays baseball on weekends" in English, one might forget to add the third-person singular 's' to "play," or use "do" instead of "do," or forget to add "on" or the plural 's' to "on weekends." It is especially difficult for beginners in English to notice these subtle differences in usage.

[0008] Therefore, traditionally, detailed usage had to be checked by an English instructor or a Japanese-English bilingual, or someone who understands English.

[0009] However, if you don't have someone like this (a checker) nearby, it's difficult to get the opportunity to have your work checked. Furthermore, even if you do have a checker nearby, having your work checked every time you write in English is a burden for both the learner and the checker.

[0010] In view of these problems, the present invention aims to make it easier than before to check for errors in texts written by learners in a language other than their native language. [Means for solving the problem]

[0011] A text proofreading support device according to one embodiment of the present invention is a text proofreading support device that assists in proofreading a text created by a learner, and comprises: an input area display means for displaying on a display an input area in which a main body area for inputting the main body, a predicate verb area for inputting the predicate verb of the main body, and a modifier area for inputting modifier phrases relating to the predicate verb; a proofreading text acquisition means for obtaining a proofread text of the original text by having AI (Artificial Intelligence) proofread the original text, which is formed by arranging a first input phrase input by the learner into the main body area, a second input phrase input by the learner into the predicate verb area, and a third input phrase input by the learner into the modifier area in the order of the respective areas; and a proofreading text display means for displaying the proofread text on the display.

[0012] Preferably, the proofreading text display means displays the proofreading text in such a way that the differences between the original text and the proofreading text are emphasized, or displays the first input phrase, the second input phrase, or the third input phrase in such a way that the differences are emphasized.

[0013] Alternatively, the AI ​​is an interactive AI that applies an LLM (Large Language Model), and the proofreading text acquisition means provides the AI ​​with a prompt that includes a command to enclose the proofreading text with predetermined symbols, and acquires the portion enclosed by the predetermined symbols from the text acquired from the AI ​​as the proofreading text. [Effects of the Invention]

[0014] According to the present invention, errors and other issues in texts written by learners in a language other than their native language can be checked more easily than before. [Brief explanation of the drawing]

[0015] [Figure 1] This diagram shows an example of the overall structure of an English composition learning support system. [Figure 2] This figure shows an example of the hardware configuration of a terminal device. [Figure 3] This is a diagram showing an example of the functional configuration of a terminal device. [Figure 4] This is a diagram showing an example of the initial state of the main window. [Figure 5] This is a diagram showing an example of an input to the English composition area. [Figure 6] This is a diagram showing an example of a state where a set of the first box to the sixth box is added to the main window. [Figure 7] This is a diagram showing an example of a method for obtaining a Japanese sentence. [Figure 8] This is a diagram showing an example of a state where a Japanese sentence is arranged in the main window. [Figure 9] This is a diagram showing an example of a method for obtaining a hint sentence. [Figure 10] This is a diagram showing an example of a hint screen. [Figure 11] This is a diagram showing an example of a hint screen. [Figure 12] This is a diagram showing an example of hint data. [Figure 13] This is a diagram showing an example of an input to the English composition area. [Figure 14] This is a diagram showing an example of a method for obtaining a corrected sentence. [Figure 15] This is a diagram showing an example of a state where a corrected sentence is arranged in the main window. [Figure 16] This is a diagram showing an example of a template screen. [Figure 17] This is a diagram showing an example of proposed data. [Figure 18] [[ID=4b]]This is a diagram showing an example of a method for presenting a corrected sentence. [Figure 19] This is a diagram showing an example of an input to the English composition area. [Figure 20] This is a diagram showing an example of a proposal screen. [Figure 21] This is a flowchart for explaining an example of the overall processing flow by a learning support program.

Mode for Carrying Out the Invention

[0016] 〔1. Overall Configuration〕 Figure 1 shows an example of the overall configuration of the English composition learning support system 1. Figure 2 shows an example of the hardware configuration of terminal device 2. Figure 3 shows an example of the functional configuration of terminal device 2.

[0017] As shown in Figure 1, the English composition learning support system 1 consists of a terminal device 2, a server 3, and a communication line 4, among other things.

[0018] English composition learning support system 1 is a system that combines a "meaning-based" method with AI (Artificial Intelligence) to support the learning of English composition.

[0019] Terminal device 2 is a computer used by learner 10 in English composition. Terminal device 2 can be a personal computer, tablet computer, or smartphone. The following explanation will use the case where a laptop-type personal computer is used as terminal device 2 as an example.

[0020] Server 3 generates answers to questions indicated in text data received from terminal device 2 using AI and sends them back to terminal device 2. For example, an existing server providing conversational AI using LLM (Large Language Model) can be used as Server 3. The following explanation will use the example of a server with OpenAI's ChatGPT-4o as Server 3. "GPT" is an abbreviation for "Generative Pretrained Transformer".

[0021] Terminal device 2 and server 3 can communicate with each other via communication line 4. The communication line 4 can be the internet, a LAN (Local Area Network), or a public network.

[0022] As shown in Figure 2, terminal device 2 consists of a main processor 20, RAM (Random Access Memory) 21, ROM (Read Only Memory) 22, auxiliary storage device 23, display 24, network interface 25, keyboard 26, pointing device 27, microphone 28, and speaker 29, among others.

[0023] The ROM 22 or auxiliary storage device 23 has the operating system and other computer programs such as the learning support program 200 (see Figure 3) installed on it. A hard disk or SSD (Solid State Drive) is used as the auxiliary storage device 23.

[0024] According to the learning support program 200, the functions shown in Figure 3, such as the original text receiving unit 201, the Japanese text acquisition unit 202, the hint presentation unit 203, the proofread text acquisition unit 204, the difference presentation unit 205, the alternative text acquisition unit 206, the hint data storage unit 211, and the suggestion data storage unit 212, are realized. The functions of each unit will be explained sequentially later.

[0025] RAM21 is the main memory of terminal device 2. Computer programs stored in ROM22 or auxiliary storage device 23 are loaded into RAM21 as needed.

[0026] The main processor 20 executes the computer program loaded into RAM 21. The main processor 20 can be a GPU (Graphics Processing Unit) or a CPU (Central Processing Unit), among others.

[0027] Display 24 displays screens for entering commands or data, and screens showing the results of calculations performed by server 3, etc.

[0028] The network interface 25 communicates with other devices using protocols such as TCP / IP (Transmission Control Protocol / Internet Protocol). A NIC (Network Interface Card) or a Wi-Fi communication device can be used as the network interface 25.

[0029] The keyboard 26 and pointing device 27 are input devices for the learner 10 to input commands or data, etc.

[0030] The microphone 28 acquires the learner's voice 10 and converts it into audio data. The speaker 29 outputs the audio.

[0031] [2. Details of the processing performed by terminal device 2 and server 3] The following describes the processing of each part of terminal device 2, the processing of server 3, and how to operate terminal device 2.

[0032] [2.1 Receipt of original text 600] Figure 4 shows an example of the initial state of the main window 50 displayed on the display 24. Figure 5 shows an example of input into the English composition area 500. Figure 6 shows an example of the main window 50 with a set of first boxes 501 through sixth boxes 506 added.

[0033] The source text receiving unit 201 in Figure 3 processes the English text created by the learner 10 (hereinafter referred to as "source text 600") as follows, for example.

[0034] When learner 10 inputs a predetermined command to terminal device 2, the source text receiving unit 201 displays the main window 50 shown in Figure 4 on the display 24.

[0035] The main window 50 includes an English composition area 500, an add button 511, a Japanese translation button 512, a hint button 513, a proofreading button 514, an alternative sentence button 515, a Japanese text area 521, and a proofread text area 522, among others.

[0036] The English composition area 500 is the area where learner 10 writes in English. The English composition area 500 is provided with a first box 501 to a sixth box 506, which are areas for inputting words and phrases. Learner 10 basically writes in English by inputting English words and phrases into one or more of the first box 501 to the sixth box 506 using the keyboard 26 or pointing device 27.

[0037] These boxes are, in principle, filled with phrases such as the following: The first box, 501, is where the subject, or "who" or "what," is entered. The subject may consist of only one word, such as "people" or "I," or it may consist of multiple words, such as "a child" or "the dogs walking in the park."

[0038] The second box, 502, is where the predicate verb (to do / is) is entered. The predicate verb can consist of only one word, such as "is" or "swim," or it can consist of multiple words, such as "look at" or "can jump."

[0039] The third box, 503, is where modifying phrases that modify the predicate verb are entered, particularly phrases that represent a person (who) or a thing (what). Often, these include complements in the second sentence pattern, objects in the third sentence pattern, direct and indirect objects in the fourth sentence pattern, or objects and complements in the fifth sentence pattern, but they are not limited to these. For example, when the English sentence "I gave a pen to him." is entered, "a pen to him" is entered in the third box, 503.

[0040] The fourth box, 504, is where you enter a phrase indicating location, such as "in the building" or "there." The fifth box, 505, is where you enter a phrase indicating time, such as "at that time" or "today." The sixth box, 506, is where you enter a phrase that precedes the subject.

[0041] Thus, each of the first box 501 through the sixth box 506 has a role. However, learner 10 can input what they want to express in English words and phrases without being too conscious of the roles of each of the first box 501 through the sixth box 506 or English grammar, by following the headings "Who," "Does," "Who / What," "Where," "When," and "Treasure Box" written above each box.

[0042] For example, to express in English, "We met Nancy at the station this morning," learner 10 can simply input the words as shown in Figure 5(A).

[0043] In Japanese, the subject is sometimes omitted, as in "I met Nancy at the station this morning." However, following the "meaning order" method, learner 10 can realize that the heading "who" in the first box 501 means that they must enter a word or phrase into the first box 501.

[0044] In other words, if learner 10 creates English sentences that follow Japanese expressions without relying on the "meaning order" method, they may omit the subject, such as "met Nancy at the station this morning," especially if learner 10 is a beginner. However, learner 10 easily notices that they need to input words into the first box 501, indicated by the heading "who," and can input words corresponding to "who," as shown in Figure 5(A).

[0045] Furthermore, learner 10 only needs to be aware of "who," "does (is)," "who / what," "where," and "when," and does not need to be too concerned with the strict distinction between the first box 501 and the fifth box 505. For example, when entering the above sentence "I gave a pen to him," if "him" is considered "who," then "a pen to him" should be entered into the third box 503 as described above. If "him" is considered "where," then "a pen" can be entered into the third box 503 and "to him" into the fourth box 504.

[0046] It is acceptable to make mistakes during input, such as missing articles or periods as shown in Figure 5(B), incorrect spelling or tense, or adding an extra "s" for the third-person singular. These mistakes will be corrected by Server 3, as described later. Boxes that do not require input can be left blank.

[0047] If the English sentence created by learner 10 does not fit into one set of the first to sixth boxes 501 to 506, learner 10 can add another set of the first to sixth boxes 501 to 506 to the English composition area 500 by clicking the add button 511, as shown in Figure 6. Then, learner 10 can input the English sentence using two sets of boxes, for example, as shown in Figure 5(C). By clicking the add button 511 again, learner 10 can add three or more sets of the first to sixth boxes 501 to 506 to the English composition area 500.

[0048] If learner 10 does not understand part of an English word, they may input the corresponding Japanese word, as shown in Figure 5(D).

[0049] Furthermore, the width of the first box 501 through the sixth box 506 expands according to the length of the entered text.

[0050] Furthermore, for details on how to input words into the first box 501 to the sixth box 506 of the English composition area 500, please refer to Non-Patent Document 1, etc.

[0051] The original text receiving unit 201 then accepts the words and phrases entered in the first box 501 through the sixth box 506 of each set within the English composition area 500, arranged in the order of their positions in each box, as the original text 600. For example, if the input is as shown in Figures 5(A) to (D), the unit will accept the sentences "We met Nancy at the station this morning.", "We met Nancy at station this morning," "I was in London, when I met her.", and "We met Nancy at station this morning" as the original text 600.

[0052] [2.2 Acquisition of Japanese 604] Figure 7 shows an example of how to obtain Japanese text 604. Figure 8 shows an example of the state in which Japanese text 604 is placed in the main window 50.

[0053] Learner 10 can check whether what they want to convey is expressed in the original text 600 by looking at the Japanese translation (Japanese text 604) of the original text 600. After entering the original text 600 into the English composition area 500 (see Figures 4 and 6), Learner 10 clicks the Japanese translation button 512.

[0054] The Japanese text acquisition unit 202 (see Figure 3) then acquires the Japanese text 604, which is a translation of the original text 600 into Japanese, for example, using the procedure shown in Figure 7.

[0055] The Japanese text acquisition unit 202 generates a prompt 602 by embedding the source text 600 into a predetermined position in a pre-prepared template 601 (#701 in Figure 7). In the example in Figure 7, "<<Source Text>>" is the predetermined position, and the source text 600 is embedded by replacing "<<Source Text>>" in the template 601 with the source text 600, thereby generating the prompt 602. The Japanese text acquisition unit 202 then sends the prompt 602 to the server 3 (#702).

[0056] Prompt 602 includes not only a command to translate into Japanese but also a request to explicitly state the subject. According to ChatGPT-4o, the subject may be omitted when generating the Japanese text, but since the subject is one of the important elements for learner 10 to check whether what they want to convey is expressed in the original text 600, prompt 602 is generated to include such a request.

[0057] Furthermore, prompt 602 includes a request to enclose the translated portion in a specified symbol (in this example, a single quotation mark). This is to distinguish the translated portion from elements other than the translated text, which may be included in the text generated by ChatGPT-4o.

[0058] When Server 3 receives prompt 602, it generates text according to prompt 602 using ChatGPT-4o and sends it to Terminal Device 2 as response text 603. The exchange of prompt 602 and response text 603 between Terminal Device 2 and Server 3 is conducted via the ChatGPT-4o API (Application Program Interface) provided by OpenAI. The exchange of each prompt and response text described later is also conducted in the same manner.

[0059] In terminal device 2, the Japanese text acquisition unit 202 receives the response text 603 from server 3 (#703), and extracts the Japanese translation portion (in this example, the portion enclosed in single quotes) from the response text 603 as Japanese text 604 (#704). Then, it displays Japanese text 604 in the Japanese text area 521 of the main window 50 (#705). At this time, the main window 50 is updated, for example, as shown in Figure 8.

[0060] Here, learner 10 checks whether what they want to convey is expressed in Japanese text 604, that is, whether what they want to convey matches what is written in Japanese text 604. If they match, they click the proofreading button 514. The process when the proofreading button 514 is clicked will be described later.

[0061] On the other hand, if it is incorrect, learner 10 corrects the words entered in the first box 501 to the sixth box 506 of each set in the English composition area 500, and causes terminal device 2 to retrieve the Japanese text 604 again.

[0062] For example, suppose that learner 10 wants to convey "I met her when I was in London," but because the words were entered as shown in Figure 5(C), the sentence "I was in London when I met her" is obtained and displayed as Japanese sentence 604.

[0063] Learner 10, by reading Japanese text 604, realizes that the position of "when" should be in the sixth box 506 (treasure box) on the first line, not the sixth box 506 (treasure box) on the second line. Therefore, they delete "when" from the sixth box 506 on the second line and enter "When" into the sixth box 506 on the first line.

[0064] The original text reception unit 201 then accepts the words entered in the first box 501 through the sixth box 506 of each set within the English composition area 500, arranged in the order of their positions in each box, namely "When I was in London, I met her.", as the new original text 600.

[0065] Then, when learner 10 clicks the Japanese translation button 512, the Japanese text acquisition unit 202 acquires the Japanese text 604 (for example, the sentence, "I met her when I was in London.") again based on the new source text 600 using the procedure described in Figure 7, and displays it in the Japanese text area 521.

[0066] [2.3 Providing Hints] If learner 10 finds that what they want to convey is not expressed in the original text 600, they may not know how to revise the original text 600 even after looking at the Japanese translation 604. If they have server 3 proofread the original text 600 without knowing how to revise it, they may get a grammatically correct English sentence, but they may not get an English sentence that conveys the meaning that learner 10 intended.

[0067] Therefore, to address this issue, the English composition learning support system 1 is equipped with a function to provide learners 10 with hints for correction. The two methods for providing hints are described below.

[0068] [2.3.1 Method 1] Figure 9 shows an example of how to obtain hint text 614. Figure 10 shows an example of hint screen 53.

[0069] The first method involves obtaining a hint from the server 3 and presenting it to the learner 10, which is carried out, for example, by the procedure shown in Figure 9.

[0070] When the hint button 513 is clicked, the hint display unit 203 (see Figure 3) displays a message such as "Click the box containing the word or phrase for which you need a hint to select it" on top of the main window 50, prompting the learner 10 to select one of the first to sixth boxes 501 through 506 in each set within the English composition area 500 (#711 in Figure 9).

[0071] The hint presentation unit 203 generates a prompt 612 (#712) by embedding the following information into a predetermined position in a pre-prepared template 611 based on the learner's 10 selection.

[0072] The hint display unit 203 is "< <x1>>" will be filled with grammatical terms corresponding to the selected box.

[0073] Specifically, if the selected box is the first box 501, the word "subject" is embedded. If it is the second box 502, the word "verb" is embedded. If it is the third box 503, the word "object or complement" is embedded. If it is the fourth box 504, the word "place" is embedded. If it is the fifth box 505, the word "time" is embedded.

[0074] Furthermore, the hint display unit 203 displays the template 611's "< <x2>The word entered in the selected box will be inserted into the ">" field.

[0075] However, if the selected box is the second box 502, and the verb entered in that box is not in its base form, the verb will be changed back to its base form and embedded. Alternatively, if the verb entered in that box is in the present progressive tense and includes a be-verb, the be-verb will be removed, and the verb immediately following the be-verb will be changed back to its base form and embedded. For example, if "stopped" is entered in the second box 502, "stop" will be embedded. Or, if "am looking forward to" is entered, "look forward to" will be embedded.

[0076] Prompt 612 includes wording that asks about the grammatical usage of the word or phrase entered by learner 10 in the context of the sentence. Furthermore, prompt 612 includes a request to enclose the answer portion (explanatory text) in a specified symbol (in this example, single quotation marks). This is to distinguish between explanatory text and elements other than explanatory text, as the text generated by ChatGPT-4o may contain such elements.

[0077] In the example in Figure 9, the template 611 contains "< <x1>>" and "< <x2>The information is embedded by replacing ">" with a verb and "stop" respectively, thereby generating prompt 612. The hint presentation unit 203 then sends the generated prompt 612 to server 3 (#713).

[0078] When server 3 receives prompt 612, it generates text according to prompt 612 using ChatGPT-4o and sends it to terminal device 2 as response text 613.

[0079] In terminal device 2, the hint presentation unit 203 receives the answer text 613 from server 3 (#714), and extracts the explanatory text (in this example, the part enclosed in single quotes) from the answer text 613 as the hint text 614 (#715). Then, it displays a hint screen 53 showing the hint text 614 on top of the main window 50, as shown in Figure 10 (#716).

[0080] [2.3.2 Second Method] Figure 11 shows an example of the hint screen 54. Figure 12 shows an example of hint data 81. Figure 13 shows an example of input into the English composition area 500.

[0081] In the first method, the server 3 generated hints and presented them to the learner 10. In the second method, however, hints appropriate to the content entered in the English composition area 500 were selected from a pre-prepared set of hints and presented to the learner 10. The second method will be explained below.

[0082] As illustrated in Figures 11(A) to (C), hint screens 54 for each of the multiple hints are prepared in advance.

[0083] The hint data storage unit 211 (see Figure 3) stores hint data 81 (81a, 81b, 81c, ...) for each hint screen 54 (54a, 54b, 54c, ...) as shown in Figure 12. The hint data 81 for each hint screen 54 contains the following information.

[0084] The "screen code" is an identifier used to identify the hint screen 54. The "condition" is a condition regarding the input content in the English composition area 500 for displaying the hint screen 54. In other words, when the words entered in the English composition area 500 match this condition, the hint screen 54 corresponding to the screen code will be displayed.

[0085] When the hint button 513 is clicked, the hint display unit 203 displays the hint screen 54 as follows:

[0086] The hint presentation unit 203 determines whether the content of the English composition area 500 satisfies each condition by comparing the conditions shown in each hint data 81 with the words entered in the English composition area 500.

[0087] For example, hint data 81a shows the condition "BOX_2="stop"". "BOX_n" on the left side means the nth box. Therefore, "BOX_2" means the second box 502. Also, "=" means that the word or phrase enclosed in double quotes on the right side, or its conjugation or third-person singular present tense, is contained in the box on the left side. Therefore, "BOX_2="stop"" means that the word entered in the second box 502 contains "stop", its conjugation (stopped or stopping), or the third-person singular present tense (stops).

[0088] Furthermore, "or" on the right side means that at least one of the words before or after it is included, and "and" means that both of the words before and after it are included. "or" on the left side means that at least one of the boxes before or after it contains the word on the right side.

[0089] Therefore, the hint data 81c, "BOX_3="cloth" or "cloths"", means that the word entered in the third box 503 contains either "cloth" or "cloths". Alternatively, "BOX_5 or BOX_6="these days" or "those days"" means that the word entered in either the fifth box 505 or the sixth box 506 of the same set contains either "these days" or "those days".

[0090] The hint display unit 203 then displays the hint screen 54, which corresponds to the screen code shown in the hint data 81 that has conditions that the current content of the English composition area 500 satisfies, on top of the main window 50.

[0091] For example, if the content of the English composition area 500 is as shown in Figure 13(A), then "stop" is entered in the second box 502, thus satisfying the conditions of the hint data 81a. Therefore, the hint presentation unit 203 displays the hint screen 54, or hint screen 54a, which corresponds to the screen code "G001" shown in the hint data 81a.

[0092] Alternatively, if the content of the English composition area 500 is as shown in Figure 13(B), then "cloth" is entered in the third box 502, and the conditions of hint data 81c are met. Therefore, the hint display unit 203 displays hint screen 54, or hint screen 54c, which corresponds to the screen code "G003" shown in hint data 81c.

[0093] If multiple conditions are met, you can display the hint screen 54 corresponding to each condition. Alternatively, you can pre-determine the priority of each condition and display the hint screen 54 corresponding to the highest priority condition.

[0094] It is also possible to use both the first and second methods interchangeably. For example, one could first try to display a hint screen 54 corresponding to a suitable hint using the second method, and if no such hint screen 54 is found, try to display a template screen 55 using the first method. Alternatively, one could perform both the first and second methods and present the hints obtained by each method to the learner 10.

[0095] [2.4 Proofreading of the original text, page 600] Figure 14 shows an example of how to obtain the proofreading text 624. Figure 15 shows an example of the state in which the proofreading text 624 is placed in the main window 50.

[0096] As described above, if learner 10 finds that the Japanese text 604 matches what they want to convey, they click the proofreading button 514 in the main window 50 (see Figure 8).

[0097] When the proofreading button 514 is clicked, the proofreading acquisition unit 204 (see Figure 3) acquires the proofread text 624 by having the server 3 proofread the original text 600. Furthermore, the difference presentation unit 205 presents the differences between the original text 600 and the proofread text 624. Here, an example of the processing flow for acquiring the proofread text 624 and presenting the differences between the original text 600 and the proofread text 624 will be explained with reference to Figure 14.

[0098] The proofreading text acquisition unit 204 generates prompt 622 (#721) by embedding the original text 600 into a predetermined position in a pre-prepared template 621. In the example in Figure 14, the original text 600 is embedded by replacing "<<original text>>" in the template 621 with the original text 600, thereby generating prompt 622. The proofreading text acquisition unit 204 then sends prompt 622 to the server 3 (#722).

[0099] Prompt 622 includes not only a correction command but also a request to enclose the corrected text in a specified symbol (double quotation marks in this example). This is to distinguish the corrected text from elements other than the text itself, as the text generated by ChatGPT-4o may contain such elements.

[0100] When server 3 receives prompt 622, it generates text according to prompt 622 and sends it to terminal device 2 as response text 623.

[0101] In terminal device 2, the proofreading text acquisition unit 204 receives the response text 623 from server 3 (#723), and extracts the proofreading text portion (in this example, the portion enclosed in double quotes) from the response text 623 as proofreading text 624 (#724).

[0102] The difference presentation unit 205 extracts the differences between the original text 600 and the extracted proofread text 624 by comparing them (#725). Then, in order to highlight the differences, it displays the original text 600 and proofread text 624 as follows (#726).

[0103] The original text 600 is entered in the first box 501 to the sixth box 506 of each set in the English composition area 500. The difference presentation unit 205 highlights and displays the differences in the words and phrases entered in these boxes. Furthermore, it displays the corrected text 624 in the corrected text area 522, highlighting the differences.

[0104] At this time, the main window 50 is updated as shown in Figure 15, for example. In Figure 15, the gray background represents a fluorescent color (for example, a fluorescent yellow). In other words, the differences are highlighted by using a fluorescent color as the background color.

[0105] Furthermore, any unnecessary words or characters in the 500-word English composition section will not only be emphasized but also struck through.

[0106] In the example in Figure 15, the emphasized portion appears only in the corrected text 624 (corrected text area 522), but it may also appear only in the original text 600 (English composition area 500), or in both.

[0107] For example, if the first box 501, the second box 502, and the third box 503 contain "I," "plays," and "baseball." respectively, and the corrected text 624 is "I play baseball.", then the "s" in "plays" in the second box 502 will be emphasized and struck through.

[0108] Alternatively, if "You," "is," and "my best friend." are entered in the first box 501, the second box 502, and the third box 503 respectively, and "You are my best friend" is obtained as the corrected text 624, then "is" in the second box 502 will be highlighted. Furthermore, "You are my best friend" will be displayed in the corrected text area 522 with "are" highlighted.

[0109] Furthermore, you may emphasize the text or other elements that highlight the differences using methods other than using a fluorescent color as the background color. For example, you may make those elements bolder or larger than the other elements. Alternatively, you may use a different font or color than the other elements.

[0110] If there are no differences, the difference presentation unit 205 displays the corrected text 624 in the corrected text area 522 without any emphasis. Furthermore, it may display a message such as "Your English composition is correct." above the main window 50.

[0111] [2.5 Presentation of alternative sentence 634] Figure 16 shows an example of template screen 55. Figure 17 shows an example of proposed data 82. Figure 18 shows an example of how to present proofread text 624. Figure 19 shows an example of input into the English composition area 500. Figure 20 shows an example of proposed screen 56.

[0112] By the way, since proofread text 624 is basically a corrected version of the original text 600, it is assumed that it does not contain any grammatical errors. However, there are cases where proofread text 624 can be expressed more naturally using a different syntax.

[0113] For example, even if the corrected sentence 624 "Two apples are on the table." is obtained from server 3 from the original sentence 600 "Two apples is on the table.", it is more natural to express it as "There are two apples on the table." using the There is / are construction. Or, even if the corrected sentence 624 "To learn three languages ​​at the same time is difficult." is obtained, it is more natural to express it as "It is difficult to learn three languages ​​at the same time." using the It is... (for person) to verb construction.

[0114] Therefore, the English composition learning support system 1 is equipped with a mechanism to suggest sentences using different grammatical structures to learner 10. This mechanism will be explained below.

[0115] As illustrated in Figures 16(A) and (B), template screens 55 for each of the multiple proposals are provided in advance.

[0116] The proposed data storage unit 212 (see Figure 3) stores proposed data 82 (82a, 82b, ...) for each template screen 55 (55a, 55b, ...) as shown in Figure 17. The proposed data 82 for each template screen 55 contains the following information.

[0117] The "screen code" is an identifier used to identify the template screen 55. The "conditions" are the conditions regarding the input content in the English composition area 500 for using the template screen 55. In other words, when the words entered in the English composition area 500 match these conditions, the template screen 55 corresponding to the screen code is used.

[0118] For example, the proposed data 82a includes the condition "BOX_1 = common noun without a definite article and BOX_2 = "be"". This means that the word entered in the first box 501 is a common noun without a definite article (including common noun phrases like "three swimming boys"), and the word entered in the second box 502 of the same pair is the be verb.

[0119] Furthermore, the proposed data 82b includes the condition "BOX_1=To-infinitive phrase and BOX_2="be"". This means that the phrase entered in the first box 501 is a To-infinitive phrase, and the phrase entered in the second box 502 of the same pair is a be verb.

[0120] The "template" is a template used to generate prompts for instructions to Server 3.

[0121] When the alternative text acquisition unit 206 has acquired the proofread text 624 and the alternative text button 515 (see Figure 15) is clicked, it performs the process to display the template screen 55 in the procedure shown in Figure 18.

[0122] The alternative sentence acquisition unit 206 searches for a proposed sentence 82 whose content in the English composition area 500 satisfies the conditions by comparing the conditions shown in each proposed sentence 82 with the words and phrases entered in the English composition area 500 (#731).

[0123] For example, if the content of the English composition area 500 is as shown in Figure 19(A), then the first box 501 contains "Two apples," which is a common noun without a definite article, and the second box 502 contains "are," which is the be verb. As a result of the search, suggested data 82a is found. Note that the determination of whether a noun is a common noun or not can be made using a known method. For example, it can be determined based on a dictionary.

[0124] In this example, the suggested data 82 was searched based on the words entered in the English composition area 500. However, it is also possible to extract words corresponding to "who," "does," "who / what," "where," and "when" from the corrected text 624 using methods such as morphological analysis, and then search the suggested data 82 based on these words.

[0125] The alternative sentence acquisition unit 206 selects the found suggestion data 82 (#732), adopts the template shown in the suggestion data 82 as template 631, and generates prompt 632 by embedding the proofreading text 624 in a predetermined position in template 631 (#733). In the example in Figure 18, "<<proofreading text>>" is the predetermined position, and the proofreading text 624 is embedded by replacing "<<proofreading text>>" in template 631 with the proofreading text 624, thereby generating prompt 632. The alternative sentence acquisition unit 206 then sends prompt 632 to server 3 (#734).

[0126] Prompt 632 includes not only the command to rewrite the text but also a request to enclose the rewritten text (rewritten sentence) in a specified symbol (double quotation marks in this example). This is to distinguish the rewritten text from elements other than the rewritten text, as the text generated by ChatGPT-4o may contain such elements.

[0127] When server 3 receives prompt 632, it generates text according to prompt 632 using ChatGPT-4o and sends it to terminal device 2 as response text 633.

[0128] In terminal device 2, the alternative text acquisition unit 206 receives the response text 633 from server 3 (#735) and extracts the part of the response text 633 that needs to be rewritten (in this example, the part enclosed in double quotes) as the alternative text 634 (#736). Then, by embedding the proofread text 624 and the alternative text 634 in the predetermined positions of the template screen 55 of the screen code shown in the proposal data 82 found in step #731, it generates the proposal screen 56 (#737) and displays it on the main window 50 (#738).

[0129] As a result, for example, if the content of the English composition area 500 is as shown in Figure 19(A), then a suggestion screen 56a like that shown in Figure 20(A) will be generated and displayed as suggestion screen 56. Alternatively, if the content is as shown in Figure 19(B), then a suggestion screen 56b like that shown in Figure 20(B) will be generated and displayed.

[0130] [3. Overall processing flow and effects of this embodiment] Figure 21 is a flowchart illustrating an example of the overall processing flow by the learning support program 200.

[0131] Next, the overall processing flow of terminal device 2 will be explained with reference to the flowchart. Terminal device 2 executes the processing according to the steps in the flowchart shown in Figure 21, based on the learning support program 200.

[0132] When a predetermined command is entered, terminal device 2 displays the main window 50 as shown in Figure 4 (#101 in Figure 21). Then, it executes the following processes in response to the operations performed by learner 10.

[0133] When the add button 511 is clicked (Yes in #102), terminal device 2 adds a set of the first box 501 to the sixth box 506 to the English composition area 500 as shown in Figure 6 (#103).

[0134] Alternatively, when a word is entered into the English composition area 500 (Yes in #104), the entered word is accepted (#105). Once all the necessary words for learner 10 have been entered into the English composition area 500, the original text 600 will be completed.

[0135] Alternatively, when the Japanese translation button 512 is clicked (Yes in #106), the string obtained from server 3, which is the Japanese translation (Japanese text 604) of the original text 600, is created by arranging each word and phrase received in step #105 in the order of the boxes, and is displayed in the Japanese text area 521 of the English composition area 500 as shown in Figure 8 (#107). One example of how to obtain the Japanese text 604 is as explained in Figure 7.

[0136] Alternatively, when the hint button 513 is clicked (Yes in #108), the terminal device 2 retrieves a hint sentence 614 from the server 3, which is a sentence indicating a hint about the original text 600, and displays it on the hint screen 53 as shown in Figure 10 (#109). One example of how to obtain the hint sentence 614 is as explained in Figure 9. Alternatively, a hint appropriate to the content entered in the English composition area 500 may be selected from a pre-prepared set of hints, and the corresponding hint screen 54 (see Figure 11) may be displayed. Alternatively, both methods may be combined.

[0137] Alternatively, when the proofreading button 514 is clicked (Yes in #110), the terminal device 2 obtains a proofread text 624 by having the server 3 proofread the original text 600, and displays it in the proofread text area 522 of the English composition area 500 as shown in Figure 15 (#111). One example of how to obtain the proofread text 624 is as explained in Figure 14.

[0138] Alternatively, when the alternative sentence button 515 is clicked (Yes in #112), terminal device 2 retrieves an alternative sentence 634 from server 3 that is synonymous with the corrected sentence 624 but uses a different syntax, and displays it on a suggestion screen 56 as shown in Figure 20 (#113). One example of how to obtain the alternative sentence 634 is explained in Figure 18.

[0139] Terminal device 2 executes steps #103, #105, #107, #109, #111, and #113 as appropriate at the timings described above. Then, when the termination command is entered (Yes in #114), the learning support program 200 is terminated.

[0140] According to this embodiment, a prompt 622 is generated by embedding the original text 600 (the text created by the learner 10 in the English composition area 500) in a predetermined position in the template 621, and a corrected text 624 is obtained simply by sending this to the server 3. Therefore, the original text 600 can be corrected more easily than in the conventional method.

[0141] Furthermore, by presenting the differences between the original text 600 and the corrected text 624, learners 10 can be made aware of the errors in the original text 600.

[0142] Furthermore, by presenting the Japanese translation of the original text 600, i.e., the Japanese text 604, to the learner 10, the learner 10 can more easily create an English sentence (original text 600) that expresses what they want to convey. If the learner 10 is still unable to successfully create the original text 600, by presenting the hint sentence 614 to the learner 10, the learner 10 can be guided more reliably toward correct English composition.

[0143] Furthermore, by presenting alternative sentence 634 to learner 10, the variety of English compositions learner 10 can produce can be increased.

[0144] [4. Variations] In this embodiment, the use of the English composition learning support system 1 for learning English composition was described as an example, but the English composition learning support system 1 can also be applied to learning composition in other languages.

[0145] Specifically, the order of the elements in the first box 501 through the sixth box 506 is changed in advance to match the word order pattern of the language being learned (SOV, SVO, VSO, etc.).

[0146] Furthermore, change the "English text" in each prompt template 601 and 621 (see Figures 7 and 14) to match the language being learned. For example, if the language being learned is German, change it to "German text" or "German sentence"; if it's French, change it to "French text" or "French sentence"; and if it's Chinese, change it to "Chinese text" or "Chinese sentence".

[0147] Furthermore, hint screens and data are provided as hint screens 54 (see Figure 11) and hint data 81 (see Figure 12), corresponding to the language being learned.

[0148] Furthermore, a template screen 55 (see Figure 16) and proposed data 82 (see Figure 17) are provided, which are screens and data for obtaining alternative sentences corresponding to the language to be learned.

[0149] In this embodiment, the case where learner 10's native language is Japanese was explained as an example, but the English composition learning support system 1 can also be used when the native language is another language. For example, if the native language is Korean, one should change "Japanese" to "Korean" in templates 601 and 611 (see Figure 9), and write the text in Korean in each template and on each screen.

[0150] In this embodiment, the hint sentence 614, the correction sentence 624, and the alternative sentence 634 were presented to the learner 10 by the display 24 displaying them as text, but they may also be presented by the speaker 29 outputting them as audio.

[0151] Furthermore, although the learner 10 input the original text 600 into the terminal device 2 using the keyboard 26, it may also be possible to input it using the microphone 28.

[0152] In this embodiment, the English composition learning support system 1 was implemented by connecting a terminal device 2 and a server 3 via a communication line 4, as shown in Figure 1. However, it may also be implemented on a single computer, for example, by installing ChatGPT-4o on the terminal device 2.

[0153] In this embodiment, as shown in Figure 4, the English composition area 500 and the Japanese composition area 521 are arranged within the main window 50, i.e., the same screen. However, the Japanese composition area 521 may be provided on a separate screen independent of the main window 50.

[0154] In this embodiment, the terminal device 2 performs a process to present a hint to the learner 10 when the hint button 513 is clicked, but this process may be performed automatically when the conditions are met. Similarly, the terminal device 2 performs a process to present an alternative sentence to the learner 10 when the alternative sentence button 515 is clicked, but this process may also be performed automatically when the conditions are met. In this embodiment, the Japanese text 604, the hint text 614, and the proofreading text 624 were obtained from server 3, but they may be obtained from separate servers (AI).

[0155] Furthermore, the overall configuration or individual parts of the English composition learning support system 1 and terminal device 2, the content of processing, the order of processing, the screen configuration, the data configuration, etc., can be modified as appropriate in accordance with the spirit of the present invention. [Explanation of Symbols]

[0156] 1. English Composition Learning Support System 10 Learners 2. Terminal device (document proofreading support device) 200 Learning Support Programs (Computer Programs) 201 Original Text Reception Unit (Input Area Display Means) 202 Japanese text acquisition unit (translation display means) 203 Hint presentation unit (hint display means) 204 Proofreading text acquisition unit (proofreading text acquisition means) 205 Difference presentation section (proofreading text display means) 206 Alternative sentence acquisition unit (alternative sentence display means) 24 displays 3. Server (AI (Artificial Intelligence)) 500 English composition area (input area) 501 First Box (Main Area) 502 Second box (predicate verb area) 503 Third box (modifier phrase area, first modifier phrase area) 504 The fourth box (modifier phrase area, second modifier phrase area) 505 The fifth box (modifier phrase area, third modifier phrase area) 600 original text 604 Japanese text (translated text) 614 Hint Text (Hint) 622 prompt 624 Proofread text 634 Alternative sentence

Claims

1. A text proofreading support device that assists in proofreading texts created by learners, An input area display means for displaying on a display an input area in which a main part area for inputting the main part, a predicate verb area for inputting the predicate verb of the main part, and a modifier area for inputting modifiers related to the predicate verb are arranged, A proofreading text acquisition means for obtaining a proofread version of the original text by having AI (Artificial Intelligence) proofread the original text, which is formed by arranging the first input phrase entered by the learner into the main section, the second input phrase entered by the learner into the pre-descriptive verb section, and the third input phrase entered by the learner into the modifier section in the order of those sections. A proofreading text display means for displaying the proofreading text on the display, A text proofreading support device characterized by having the following features.

2. The proofreading display means displays the proofreading in such a way that the differences between the original text and the proofreading are emphasized, or displays the first input phrase, the second input phrase, or the third input phrase in such a way that the differences are emphasized. The text proofreading support device according to claim 1.

3. The aforementioned AI is an interactive AI that applies LLM (Large Language Model), The proofreading text acquisition means provides the AI ​​with a prompt including a command to enclose the proofreading text with predetermined symbols, and acquires the portion enclosed by the predetermined symbols from the text acquired by the AI ​​as the proofreading text. The text proofreading support device according to claim 2.

4. A translation display means that obtains a translated text obtained by machine-translating the original text into a predetermined natural language and displays it on the display. Having, A text proofreading support device according to any one of claims 1 to 3.

5. Alternative sentence display means, which obtains an alternative sentence in which the corrected sentence is expressed using a syntax different from the syntax used in the corrected sentence and displays it on the display, Having, A text proofreading support device according to any one of claims 1 to 3.

6. The input area display means displays as the input area an area in which a first modifier area for inputting a phrase representing a person or thing, a second modifier area for inputting a phrase representing a place, and a third modifier area for inputting a phrase representing time are arranged. The third input phrase is obtained by arranging the phrases entered in the first modifier phrase area, the second modifier phrase area, and the third modifier phrase area in the order of those areas. A text proofreading support device according to any one of claims 1 to 3.

7. A hint display means that obtains hints on how to use the words entered in the area selected by the learner from the subject area, the pre-descriptive verb area, the first modifier area, the second modifier area, and the third modifier area, and displays them on the display. Having, The text proofreading support device according to claim 6.

8. A text proofreading method for correcting texts written by learners, An input area is displayed on the screen, which includes a subject area for inputting the subject, a predicate verb area for inputting the predicate verb of the subject, and a modifier area for inputting modifiers related to the predicate verb. The original text is obtained by having AI (Artificial Intelligence) proofread the original text, which is formed by arranging the first input phrase entered by the learner in the main section, the second input phrase entered by the learner in the pre-descriptive verb section, and the third input phrase entered by the learner in the modifier section, in the order of each section. The proofreading text is displayed on the display. A method for proofreading text characterized by the following features.

9. A computer program used in a computer to assist in proofreading texts written by learners, To the aforementioned computer, The system executes a process to display on the screen an input area in which a subject area is entered for the subject, a predicate verb area is entered for the predicate verb of the subject, and a modifier area is entered for modifiers relating to the predicate verb. The process involves having AI (Artificial Intelligence) proofread the original text, which is formed by arranging the first input phrase entered by the learner into the main section, the second input phrase entered by the learner into the predicative verb section, and the third input phrase entered by the learner into the modifier section, in the order of these sections, thereby obtaining a proofread version of the original text. The process of displaying the proofread text on the display is executed. A computer program characterized by the following features.

Images