Sentence simplification system, sentence simplification method, and computer program

Abstract

To simplify sentences including "let's" in a predicate, more properly than before.SOLUTION: A conversion server 1 is provided with: a directory 120 which, for each of a plurality of first phrases using "let's", stores a second phrase plainly expressing the first phrase; and a paraphrase complementary section 105 which, when one of the plurality of first phrases is included in a document 52 in a honorific language form, creates a paraphrase complementary document 53 by replacing the first phrase in the document 52 in the honorific language form, with a second phrase of the first phrase being stored in the directory 120.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to a technique for simplifying sentences written in Japanese. [Background technology]

[0002] In recent years, the number of foreigners staying in Japan and learning Japanese has been increasing. However, Japanese documents are written using many different types of characters, such as kanji, hiragana, and katakana, and are written with grammar that is significantly different from that of languages ​​such as English and French. Therefore, Japanese documents can be difficult for foreigners to understand.

[0003] Not being able to fully understand Japanese documents can cause problems in your life in Japan. In particular, not being able to obtain information about disasters can be life-threatening.

[0004] For example, the Great Hanshin-Awaji earthquake that occurred in January 1995 affected not only Japanese people but also many foreigners residing in Japan, some of whom were unable to fully understand Japanese and therefore were unable to receive the information they needed.

[0005] Therefore, rules for "easy Japanese" were devised to help such foreigners act appropriately in the event of a disaster.

[0006] "Easy Japanese" includes rules such as avoiding difficult words and using simple words, shortening sentences, and separating sentences. By creating Japanese documents in accordance with these rules, necessary information can be conveyed more reliably to foreigners who are not yet able to fully understand Japanese. Currently, these rules are compiled by administrative agencies, as described in Non-Patent Document 1.

[0007] It is possible to create documents from scratch according to the rules of easy Japanese, but since it requires learning the rules, this is surprisingly difficult even for native Japanese speakers. Also, creating documents both in easy Japanese and in other languages ​​requires time and effort. This is especially true in emergencies such as disasters, when such time is of the essence.

[0008] Therefore, a technique for simplifying documents has been proposed. The sentence simplification system described in Patent Document 1 simplifies an original sentence written in Japanese and ending with "mashou" as follows: The structure of the original sentence is analyzed by morphological analysis and dependency analysis. Then, based on the analyzed structure, the ending of the original sentence is converted to "masu" in the first case where the case particle "ga" is used in the subject of the original sentence, and to "kudasai" in the second case where the subparticle "ha" is used in the subject and in the third case where the subject is not indicated in the original sentence, thereby generating a simplified sentence. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] Japanese Patent Publication No. 2021-121977 [Non-patent literature]

[0010] [Non-Patent Document 1] Guidelines for easy Japanese for supporting residents in Japan, etc., Agency for Cultural Affairs website, retrieved on July 7, 2023, https: / / www.bunka.go.jp / seisaku / kokugo_nihongo / kyoiku / 92484001.html Summary of the Invention [Problem to be solved by the invention]

[0011] "Masho" used at the end of a sentence is a phrase consisting of the polite auxiliary verb "mashou" in its potential form and the auxiliary verb "u". As the auxiliary verb "u" expresses either will, invitation, or conjecture, "mashou" also expresses either one of these. As it has multiple meanings, it is difficult for non-native Japanese speakers to grasp the exact nuance of "mashou". Therefore, it is desirable to more accurately simplify sentences that contain "mashou" as a predicate.

[0012] In view of the above problems, the present invention aims to simplify sentences that include the predicate "mashou" more accurately than ever before. [Means for solving the problem]

[0013] A sentence simplification system according to one embodiment of the present invention is a sentence simplification system that simplifies an original sentence having a predicate using "mashou" and includes: a storage means that stores, for each of a plurality of first phrases using "mashou", a second phrase that expresses the first phrase in plain language; and a generation means that, when any of the plurality of first phrases is included in the original sentence, generates a paraphrase sentence by replacing the first phrase included in the original sentence with the second phrase of the included first phrase stored in the storage means.

[0014] Preferably, the apparatus has a second storage means for storing, for each of a plurality of additional words which are specific particles or symbols, a rephrasing method when the additional word is used immediately after "maesho", and the generating means generates the rephrased sentence by paraphrasing the preceding descriptive word using the rephrasing method of the additional word used, which is stored in the second storage means, when none of the plurality of first words is included in the original sentence but any of the plurality of additional words is used immediately after "maesho" in the preceding descriptive word.

[0015] Alternatively, when none of the plurality of first words and phrases is included in the original sentence but the verb used in the preceding descriptive word is an aspectless verb, the generation means generates the paraphrase sentence by converting the preceding descriptive word so that the verb expresses an intention if it is any of a plurality of specified verbs.

[0016] Alternatively, when none of the plurality of first words and phrases is included in the original sentence but the subject of the preceding descriptive word is indicated by the case particle "ga," if the subject noun falls into one of a plurality of types, the generation means generates the paraphrase sentence by converting the preceding descriptive word to express will.

[0017] Alternatively, when none of the multiple first words are included in the original sentence but the subject of the preceding descriptive word is indicated by the connecting particle "mo," the generation means generates the paraphrase sentence by converting the preceding descriptive word to express an invitation if the subject is single and includes a title, but converting it to express an intention if it does not, and converting the preceding descriptive word to express an invitation if the subject is plural and includes a second person noun, but converting it to express an intention if it does not. Effect of the Invention

[0018] According to the present invention, sentences including "mashou" can be simplified more accurately than ever before. [Brief description of the drawings]

[0019] [Figure 1] FIG. 1 is a diagram illustrating an example of an overall configuration of a document simplification system. [Diagram 2] FIG. 2 illustrates an example of a hardware configuration of a conversion server. [Diagram 3] FIG. 2 illustrates an example of a functional configuration of a conversion server. [Figure 4] FIG. 4 is a diagram showing an example of first dictionary data. [Diagram 5] 13 is a flowchart showing an example of the flow of a sentence-end paraphrase process for "mashou" (let's do this). [Figure 6] FIG. 11 is a diagram illustrating an example of second dictionary data. [Figure 7] FIG. 13 is a diagram showing an example of paraphrase method data. [Figure 8] 13 is a flowchart showing an example of the flow of a no-aspect verb process. [Figure 9] 13 is a flowchart showing an example of the flow of an aspect verb process. [Figure 10] 11 is a flowchart illustrating an example of an overall flow of processing by a document simplification program. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] Fig. 1 is a diagram showing an example of the overall configuration of a document simplification system 4. Fig. 2 is a diagram showing an example of the hardware configuration of a conversion server 1. Fig. 3 is a diagram showing an example of the functional configuration of the conversion server 1.

[0021] As shown in FIG. 1, the document simplification system 4 is made up of a conversion server 1, one or more terminal devices 2, a communication line 3, and the like.

[0022] The conversion server 1 and each terminal device 2 can communicate with each other via a communication line 3. As the communication line 3, the Internet, a public line, a LAN (Local Area Network), or the like is used.

[0023] According to the document simplification system 4, it is possible to provide a service to a user in which a document is converted into a simpler document and output.

[0024] The conversion server 1 converts a document into a simpler document. As shown in Fig. 2, the conversion server 1 is composed of a computing unit 10a, a RAM (Random Access Memory) 10b, a ROM (Read Only Memory) 10c, an auxiliary storage device 10d, and a network adapter 10e. A so-called server machine is used as the conversion server 1. Alternatively, a system combining a plurality of computers may be used, or a so-called cloud server may be used.

[0025] The ROM 10c or the auxiliary storage device 10d stores an operating system, web server software, a document simplification program 100 (see FIG. 3) described later, and the like. Furthermore, the auxiliary storage device 10d is provided with a directory 120 for storing data used by the document simplification program 100. A hard disk or an SSD (Solid State Drive) or the like is used as the auxiliary storage device 10d.

[0026] The RAM 10b is the main memory of the conversion server 1, and the programs stored in the ROM 10c or the auxiliary storage device 10d are loaded into the RAM 10b as required.

[0027] The arithmetic unit 10a is a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit), and executes a program loaded into the RAM 10b.

[0028] The network adapter 10e is a device for communicating with the terminal device 2 and the like using protocols such as TCP / IP (Transmission Control Protocol / Internet Protocol). A network interface card (NIC) is used as the network adapter 10e.

[0029] The terminal device 2 outputs the plain document obtained by the conversion server 1 for presentation to a user. As the terminal device 2, a personal computer having a web browser, a tablet computer, a smartphone, or the like is used.

[0030] The document simplification program 100 is a program that provides the above-mentioned services and runs on web server software, and is therefore written in a language compatible with the web server software.

[0031] The document simplification program 100 is composed of software modules such as an original document data acquisition unit 101, a sentence analysis unit 102, a sentence division unit 103, a sentence-final expression unification unit 104, a paraphrase completion unit 105, a ruby ​​character division unit 106, and a plain document transmission unit 107, all of which are shown in FIG.

[0032] In addition, the document simplification program 100 is an extension of the document simplification program described in the publicly known document 1 (JP Patent Publication No. 2021-121976), and in particular, can simplify sentences in documents that include "mashou" attached to the conjunctive form of a verb or auxiliary verb more accurately than the document simplification program described in the publicly known document 1. Below, the functions of each part shown in FIG. 3 will be explained, focusing on the function of simplifying sentences that include "mashou."

[0033] FIG. 4 is a diagram showing an example of first dictionary data 801. FIG. 5 is a flowchart showing an example of the flow of sentence-end paraphrasing processing for "mashou." FIG. 6 is a diagram showing an example of second dictionary data 802. FIG. 7 is a diagram showing an example of paraphrasing method data 803. FIG. 8 is a flowchart showing an example of the flow of non-aspect verb processing. FIG. 9 is a flowchart showing an example of the flow of aspect verb processing.

[0034] The original document data acquisition unit 101 acquires text data of an original document 50, which is an original document (original text) including one or more sentences, from the terminal device 2 by a known method such as that described in the publicly known document 1.

[0035] When the text data of the original document 50 is acquired by the original document data acquisition unit 101, the sentence analysis unit 102 performs morphological analysis and dependency analysis of the original document 50 based on known algorithms such as MeCab, ChaSen, or CaboCha, thereby detecting the words and phrases contained in each sentence of the original document 50 and determining the structure of the sentence (dependencies, relationships between words, relationships between phrases, etc.).

[0036] The sentence dividing unit 103 generates the shortened document 51 by appropriately dividing the original document 50 into each sentence contained in the original document 50. These processes may be performed by a known method such as that described in the publicly known document 1.

[0037] The sentence-end expression unification unit 104 unifies the expression style at the end of each sentence included in the shortened document 51 into a polite style (so-called "desu / masu style"), thereby generating a polite document 52. These processes may be performed by a known method such as that described in the publicly known document 1.

[0038] The paraphrase completion unit 105 generates the paraphrase completion document 53 by replacing words included in the polite document 52 with other words or by completing words in the polite document 52. Specifically, the paraphrase completion unit 105 executes end-of-sentence paraphrase processing, number paraphrase processing, difficult word paraphrase processing, and completion processing.

[0039] The end-of-sentence paraphrase process is a process of paraphrasing a predicate at the end of a sentence to prevent misunderstanding of the sentence or to aid in understanding. The end-of-sentence paraphrase process is executed as follows based on the first dictionary data 801 shown in FIG.

[0040] The paraphrase completion unit 105 detects sentences in which any of the target phrases shown in the first dictionary data 801 is included in the predicate at the end of the sentence, among the sentences included in the polite document 52. Then, the paraphrase completion unit 105 replaces the target phrase at the end of the detected sentence with an alternative phrase corresponding to the target phrase. However, a serial code in a predetermined format (for example, a four-digit alphanumeric string starting with "A") may be shown in the first dictionary data 801 as an alternative phrase corresponding to the target phrase. In this case, the processing of a routine corresponding to this serial code is executed as the sentence-end paraphrase processing.

[0041] For example, the sentence "It is expected that the rain will continue this week" is converted to "It will continue this week." In addition, a sentence that includes the target phrase "mashou" as part of the predicate is processed by the routine corresponding to the serial code "A001." This process is executed according to the procedure shown in Figure 5.

[0042] Incidentally, "let's" means either a will, an invitation, or a conjecture. That is, for example, in the sentence "Let's go to Tokyo," "let's" expresses a will, and this sentence can be rephrased as "I'm going to Tokyo." Also, in the sentence "Let's dance with me," "let's" expresses an invitation, and this sentence can be rephrased as "Please dance with me." Also, in the sentence "Let's be extremely hot this summer," "let's" expresses a conjecture, and this sentence can be rephrased as "This summer will be extremely hot."

[0043] As "mashou" has three meanings, it can be difficult for people whose native language is not Japanese to understand sentences that contain "mashou." Therefore, paraphrasing sentences that contain "mashou" is important in simplifying sentences. Therefore, the conversion server 1 employs the process shown in Figure 5 as a sentence-end paraphrase process for "mashou." This process makes it possible to simplify sentences that contain "mashou" more reliably than conventional methods.

[0044] The outline of the sentence-end paraphrase processing for "mashou" is as follows: When any of the target words and phrases shown in the second dictionary data 802 in Fig. 6 is included in the sentence (Yes in #721 in Fig. 5), the paraphrase completion unit 105 executes the exception processing as the sentence-end paraphrase processing (#722).

[0045] However, if the sentence does not contain any target phrases shown in the second dictionary data 802 (No in #721), and if any of the additional words (final particles, formal nouns, or symbols, etc.) shown in the paraphrase method data 803 in FIG. 7 is attached immediately after “mashou” included in the sentence (Yes in #723), the paraphrase completion unit 105 executes the additional word-containing verb processing as the end-of-sentence paraphrase processing (#724).

[0046] However, if the verb immediately preceding "mashou" in the sentence is not immediately followed by an additional word indicated in the paraphrase method data 803 (No in #723), and if the verb immediately preceding "mashou" is a non-aspect verb (Yes in #725), the paraphrase completion unit 105 executes a non-aspect verb process as a sentence-end paraphrase process (#726). On the other hand, if the verb is not a non-aspect verb, that is, if it is an aspect verb (No in #725), it executes an aspect verb process as a sentence-end paraphrase process (#727).

[0047] Here, exception processing, adjunct-containing verb processing, aspectless verb processing, and aspect verb processing will be explained in turn.

[0048] The paraphrase completion unit 105 executes exception processing as follows. The paraphrase completion unit 105 replaces a target word contained in a sentence with an alternative word corresponding to the target word, based on the second dictionary data 802 (see FIG. 6). However, if there are multiple corresponding alternative words, one of them is selected at random to replace the target word. Alternatively, the target word may be replaced with a predetermined one.

[0049] As a result, for example, the sentence "Tomorrow will be a good day for an outing" is converted to "Tomorrow will be a good day for an outing."

[0050] In addition, the sentence "If you have a fever of 37.5 degrees or more, see a doctor" is converted to "If you have a fever of 37.5 degrees or more, see a doctor" or "If you have a fever of 37.5 degrees or more, see a doctor."

[0051] According to the latter conversion, not only is "maeshou" replaced by "kudasai," but "ikusai o karu" is replaced by "ikusai o tsuite" (to receive a doctor's diagnosis). In other words, "maeshou" is replaced by another phrase expressing will, invitation, or conjecture, and difficult phrases are replaced by simpler phrases. Although difficult phrases can be simplified by difficult word paraphrase processing, the accuracy of simplifying the entire sentence can be improved by performing the paraphrase processing at the end of the sentence together in the exception processing as in this example.

[0052] Furthermore, the paraphrase completion unit 105 executes the additional word-containing verb process as follows. Based on the paraphrase method data 803 (see FIG. 7), the paraphrase completion unit 105 replaces "let's" and the additional word with another phrase based on the paraphrase method corresponding to the additional word. As the paraphrase method, only an alternative phrase may be indicated. In this case, "let's" and the additional word are replaced with the alternative phrase.

[0053] As a result, for example, the sentence "Let's win tomorrow" is converted to "We will win tomorrow" by replacing "will" and "also" with "masu". Similarly, the sentence "What strategy shall we use tomorrow?" is converted to "What strategy will you use tomorrow?" Similarly, the sentence "I promise to keep myself in good condition" is converted to "I promise to keep myself in good condition."

[0054] Alternatively, alternative phrases corresponding to each of a plurality of cases may be displayed as a paraphrase method. In this case, the paraphrase completion unit 105 replaces "mashou" and the additional word with an alternative phrase corresponding to the case to which the sentence fits among the plurality of cases.

[0055] For example, a sentence that contains either "ne" or "yo" as an adjunct is rephrased to express an intention if the predicate containing "mashou" (the phrase that has a dependent relationship with "mashou"), i.e., the subject, is in the first person and its case particle is "ga," and is rephrased to express an invitation in all other cases.

[0056] As a result, for example, the sentence "We will win" is converted to "We will win." However, the sentence "Yamada-san, let's study" is converted to "Yamada-san, please study."

[0057] Alternatively, a set of a conjugated form and an alternative phrase may be displayed as a paraphrase method. In this case, the paraphrase completion unit 105 changes the verb immediately before "mashou" to the conjugated form, and replaces "mashou" and the additional word with the alternative phrase.

[0058] As a result, for example, the sentence "The weather will improve tomorrow, and it will be a good day to go out" is converted to "The weather will improve tomorrow, and it will be a good day to go out." Also, the sentence "The heavy rain will continue, so please stay away from the river" is converted to "The heavy rain will continue, so please stay away from the river." Also, the sentence "Please wear a hat, as the sun will get stronger" is converted to "Please wear a hat, as the sun will get stronger."

[0059] In addition, when processing verbs containing adjuncts and when processing substitutions, the conjugation ending of the verb immediately before "mashou" is appropriately changed to a phonetic suffix. Also, the "te" in "kudasai" is appropriately changed to "de."

[0060] For example, the sentence "We will not advance to the final league, whether we win or lose the next game" is converted to "We will not advance to the final tournament, whether we win or lose the next game" by using the double consonant. Also, the sentence "Let's proceed calmly" is converted to "Please proceed calmly."

[0061] Moreover, the paraphrase completion unit 105 executes a no-aspect verb process in the procedure shown in Fig. 8. The paraphrase completion unit 105 converts the sentence as follows depending on whether the verb immediately before "mashou" is a specific verb or not.

[0062] If the immediately preceding verb is a sa-hen verb (a verb of the sa-row irregular conjugation), the progressive form of a sa-hen verb (``○○ shieru'' or ``○○ shite oru''), or the conjunctive form of ``aru'' (``Yes'' in #731 of FIG. 8), the paraphrase completion unit 105 converts the immediately preceding verb and ``mashou'' into an expression expressing a will, such as ``○○ shimasu,'' ``○○ shimasu,'' or ``arimasu'' (``are'') (#732).

[0063] For example, the sentence "Tomorrow's game will be worth three regular games" is converted to "Tomorrow's game is worth three regular games." Also, the sentence "That means this game is not just one game" is converted to "That means this game is not just one game." Also, the sentence "It will be worth that much" is converted to "It is worth that much."

[0064] On the other hand, if the sentence is not one of these three conjunctive forms (No in #731), paraphrase completion section 105 converts the sentence to express a conjecture based on a predetermined rule (#733).

[0065] For example, the sentence "This product will be inferior to that product" is converted to "This product will be inferior to that product." Also, the sentence "The two people's ways of thinking will be completely different" is converted to "The two people's ways of thinking will be completely different."

[0066] Moreover, the paraphrase completion unit 105 executes the aspectless verb process in the procedure shown in Fig. 9. The paraphrase completion unit 105 converts the sentence as follows mainly based on the rules for the case particle "ga" of the modifier of the predicate including "mashou" or the rules for the modifier particle "mo".

[0067] If the case particle "ga" is used in the subject of the predicate (Yes in #741 of FIG. 9), the paraphrase completion unit 105 converts the sentence as follows according to the type of the noun that forms a clause together with the case particle "ga" (i.e., the subject of this predicate): If the noun is a proper noun other than the second person (including cases where a title is attached, such as "Yamada-kun," "Suzuki-san," or "Kato-sensei"), a first person noun, or a noun representing an animal or a person (Yes in #742), the sentence is converted to express an intention (#743). Otherwise (No in #742), the sentence is converted to express an invitation (#744).

[0068] As a result, for example, the sentence "President Sato will announce the start of the tournament" is converted to "President Sato will announce the start of the tournament." Similarly, the sentence "I will go to Tokyo" is converted to "I am going to Tokyo." In addition, the sentence "Pochi will walk in the park" is converted to "Pochi is walking in the park."

[0069] However, the sentence "You shall go on a business trip to Tokyo" is converted to "You please go on a business trip to Tokyo." Also, the sentence "Kimura-san shall go on a business trip to Tokyo" is converted to "Kimura-san please go on a business trip to Tokyo" if "Kimura-san" is the listener of the sentence, that is, corresponds to a second person such as "you." Whether or not a proper noun is second person may be specified by the user, may be determined by natural language processing, or may be determined based on the context.

[0070] When the connecting particle "mo" is used instead of the case particle "ga" (No in #741, Yes in #745), the paraphrase completion unit 105 converts the sentence in the same way as in the case of the case particle "ga", in principle, but there are exceptions.

[0071] If there are multiple nouns (i.e., the subject of this predicate) that form a clause together with the connecting particle "mo" (Yes in #746), and if a second-person noun is included among those nouns (Yes in #747), and if a first-person noun is included (Yes in #748), the paraphrase completion unit 105 converts a part of the form {first-person noun} + "mo" + {second-person noun} + "mo" or a part of the form {second-person noun} + "mo" + {first-person noun} + "mo" to a form {first-person noun} + "together" + {second-person noun} + "mo", and converts "mashou" to mean an invitation (#749).

[0072] As a result, for example, a sentence such as "Let's go to Tokyo together" or "Let's go to Tokyo together" is converted to "Please go to Tokyo together with me." Or, it may be converted to "Please go to Tokyo together with me."

[0073] Also, if a second person noun is included (Yes in #747) but a first person noun is not included (No in #748), the paraphrase completion unit 105 converts "maeshou" to mean an invitation (#744).

[0074] As a result, for example, the sentence "Let's go to Tokyo together." is converted to "Please go to Tokyo together." As in step #749, it may be converted to "Please go to Tokyo together with him."

[0075] If the multiple nouns forming a phrase together with the connecting particle "mo" do not include a second person noun (No in #747), the paraphrase completion unit 105 converts "mashou" to mean will (#743).

[0076] As a result, for example, a sentence such as "Let's go to Tokyo together" is converted to "We are both going to Tokyo together."

[0077] If there is one noun that forms a phrase together with the particle "mo" (No in #746), and if that noun has a title attached to it (Yes in #750), the paraphrase completion unit 105 converts "mashou" to mean an invitation (#744). If not (No in #750), the sentence is converted in the same way as in the case of the particle "ga" (#742-#744).

[0078] As a result, for example, a sentence such as "Yamada, let's go to Tokyo too" is converted to "Yamada, please go to Tokyo too." Also, a sentence such as "I'll go with you too" is converted to "I'll go with you too."

[0079] If neither the case particle "ga" nor the functional particle "mo" is used (No in #741, No in #745), paraphrase completion unit 105 converts the sentence to express an invitation (#744).

[0080] As a result, for example, a sentence such as "Let's evacuate immediately" is converted to "Please evacuate immediately."

[0081] As described above, the paraphrase completion unit 105 further executes number replacement processing, difficult word replacement processing, and completion processing. The number replacement processing is processing for simplifying difficult words and phrases related to numbers. The difficult word replacement processing is processing for simplifying difficult words and phrases other than numbers. The completion processing is processing for completing words that explain difficult words and phrases in simple terms. The method described in the publicly known document 1 is used as the method for each processing. By these processes, a paraphrase completion document 53 is generated from the polite document 52.

[0082] When the paraphrase completion unit 105 generates the paraphrase completion document 53, the ruby ​​separator unit 106 separates each sentence included in the paraphrase completion document 53 into phrases and adds ruby ​​to each kanji included in the paraphrase completion document 53, thereby generating a plain text document 54. In this embodiment, hiragana or katakana characters that indicate the reading of the kanji are added as ruby. The reading of the phrases and the kanji can be identified by the known morphological analysis algorithm and dictionary exemplified above. After the word separation, the full stop may be deleted.

[0083] When the plain document 54 is generated by the ruby ​​separator 106 , the plain document transmitter 107 transmits the HTML data of the plain document 54 to the terminal device 2 .

[0084] When the terminal device 2 receives the HTML data, the terminal device 2 uses a web browser to display the plain text 54. Furthermore, the terminal device 2 may read out the plain text 54 using a known voice synthesis function.

[0085] FIG. 10 is a flowchart illustrating an example of the overall flow of processing by the document simplification program 100. As shown in FIG.

[0086] Next, we will explain the overall processing flow of the conversion server 1. The conversion server 1 executes processing based on the document simplification program 100 in the procedure shown in FIG.

[0087] When the conversion server 1 receives the text data of the original document 50 from the terminal device 2 (#11 in FIG. 10), it performs morphological analysis and dependency analysis on the original document 50 to detect the words and phrases contained in each sentence of the original document 50 and determine the structure of each sentence (#12). Furthermore, it divides each sentence of the original document 50 as necessary (#13). As a result, one or more shortened documents 51 are obtained.

[0088] The conversion server 1 standardizes the endings of all sentences contained in the shortened document 51 to polite form (#14). As a result, a polite form document 52 is obtained.

[0089] Furthermore, the conversion server 1 executes end-of-sentence replacement processing, number replacement processing, difficult word replacement processing, and completion processing on the polite document 52 (#15). This results in a paraphrase completion document 53. In particular, for a sentence that includes "mashou," end-of-sentence replacement processing is executed according to the procedure shown in FIG. 5.

[0090] Furthermore, the conversion server 1 adds ruby ​​notation to the kanji characters contained in the paraphrase complement document 53, and converts the paraphrase complement document 53 into word-spaced text (#16). As a result, a plain text 54 is obtained.

[0091] Then, the conversion server 1 transmits the HTML data of the plain document 54 to the terminal device 2 (#17). Upon receiving the HTML data, the terminal device 2 displays the plain document 54.

[0092] According to this embodiment, a sentence including the predicate "mashou" can be simplified more accurately than ever before.

[0093] In particular, by executing exception processing first among the four processing methods, namely, exception processing, processing of verbs with adjuncts, processing of verbs without aspect, and processing of verbs with aspect, it is possible to more accurately rephrase "mashou" used in predicates or verb phrases that are difficult to classify than before.

[0094] In this embodiment, after the exception processing, the three processes, the adjunct-containing verb processing, the non-aspect verb processing, and the aspect verb processing, are executed in this order, but the three processes do not have to be executed in this order. For example, the adjunct-containing verb processing may be executed after the two processes, the non-aspect verb processing and the aspect verb processing, are executed in any order.

[0095] Incidentally, the sentence "Sperm whales may be lurking in the deep sea" can be interpreted as either expressing a will or a conjecture. However, it cannot be interpreted as expressing an invitation. The same is true when the connecting particle "mo" is used instead of the case particle "ga." In this way, when only a third-person noun is the subject, the conversion server 1 need only convert the sentence so that it expresses either a will or a conjecture (i.e., so that it does not express an invitation).

[0096] In this case, whether to adopt will or inference can be determined based on the environment in which the document simplification system 4 is used or the characteristics of the sentence. For example, when the document simplification system 4 is used in an environment in which asserting the content would raise a question of responsibility or in an environment in which nuance is important, the system can be set to convert sentences to express inference. Alternatively, when the document simplification system 4 is used in an environment in which it is acceptable to lose nuance but clarity is important, the system can be set to convert sentences to express will.

[0097] Alternatively, it may be determined in advance for each verb or verb phrase. In this case, it is sufficient to register it in the second dictionary data 802.

[0098] Additionally, the configuration of the document simplification system 4 and the conversion server 1 as a whole or each part, the content of the processing, the order of processing, the content of the data, etc. may be modified as appropriate in accordance with the spirit of the present invention. [Explanation of symbols]

[0099] 1 Conversion server (computer) 4. Document Simplification System 50 Original document (original sentence) 52 Polite sentences (original sentences) 53 Paraphrased sentences 54 Plain text (paraphrased sentences) 100 Document simplification program (computer program) 105 Paraphrase completion section (generation section) 120 Directory (storage means) 802 Second Paraphrase Data 803 Paraphrase Method Data

Claims

1. A sentence simplification system for simplifying an original sentence having a predicate including "mashou" (let's do this), comprising: a storage means for storing a second phrase that expresses the first phrase in a simpler form for each of a plurality of first phrases that use "mashou"; a generating means for generating a paraphrase sentence by replacing the first phrase included in the original sentence with the second phrase of the first phrase included in the original sentence, the second phrase being stored in the storage means, when any of the plurality of first phrases is included in the original sentence; A sentence simplification system comprising:

2. a second storage means for storing, for each of a plurality of additional words which are specific particles or symbols, a paraphrase method when the additional word is used immediately after "mashou"; having the generating means, when none of the plurality of first words and phrases is included in the original sentence but any of the plurality of additional words is used immediately after "mashou" in the preceding descriptive word, generates the paraphrased sentence by paraphrasing the preceding descriptive word using the paraphrasing method of the additional word used, which is stored in the second storage means; 2. The sentence simplification system of claim 1.

3. the generating means generates the paraphrased sentence by replacing "mashou" and the "tomo" in the preceding descriptive word with "masu" when the additional word used is "tomo", replacing the "mashou" and the "ga," with "temo," when the additional word used is "ga," replacing the "mashou" and the "ga," with "temo," when the additional word used is "shi" or "kara," converting the verb or auxiliary verb immediately before the "mashou" to a conclusive form and replacing the "mashou" and the "shi" or "kara" with "deshoushi" or "deshokara," when the additional word used is "ne" or "yo," replacing the "mashou" and the "ne" or "yo" with "masu" if the subject of the preceding descriptive word is first person, or with "kudasai" if not, and replacing the "mashou" and the "ka?" or "?" with "masu mashou?" when the additional word used is "ka?" or "?" 3. The sentence simplification system of claim 2.

4. the generating means, when none of the plurality of first words and phrases is included in the original sentence but a verb used in the preceding descriptive word is a non-aspect verb, generates the paraphrased sentence by converting the preceding descriptive word so that the verb expresses an intention if the verb is any one of a plurality of predetermined verbs; 2. The sentence simplification system of claim 1.

5. The plurality of predetermined verbs are sa-row irregular conjugated verbs and their progressive forms, as well as "aru" (aru), 5. A sentence simplification system according to claim 4.

6. the generating means generates the paraphrased sentence by converting the preceding descriptive word so as to express an invitation if the verb used in the preceding descriptive word is not any of the plurality of predetermined verbs; 6. A sentence simplification system according to claim 4 or claim 5.

7. the generating means generates the paraphrased sentence by converting the preceding descriptive word so as to express an intention if the subject of the preceding descriptive word corresponds to one of a plurality of types when none of the plurality of first words and phrases is included in the original sentence but the subject of the preceding descriptive word is indicated by the case particle "ga".

2. The sentence simplification system of claim 1.

8. The types are non-second person proper nouns, first person nouns, animal names, and nouns representing people.

8. A sentence simplification system according to claim 7.

9. the generating means generates the paraphrased sentence by converting the preceding descriptive word so as to express an invitation if the subject does not fall into any of the plurality of types; 9. A sentence simplification system according to claim 7 or claim 8.

10. the generating means, when none of the plurality of first words and phrases is included in the original sentence but the subject of the preceding descriptive word is indicated by the connecting particle "mo", converts the preceding descriptive word to express an invitation if the subject is single and includes a title, but converts it to express an intention if the subject is not included, and converts the preceding descriptive word to express an invitation if the subject is plural and includes a second person noun, but converts it to express an intention if the subject is not included, thereby generating the paraphrased sentence; 2. The sentence simplification system of claim 1.

11. said generating means, if the subject of the preceding descriptive word is not indicated by either the case particle "ga" or the connecting particle "mo", generates said paraphrase sentence by converting the preceding descriptive word so as to express an invitation; A sentence simplification system according to claim 7 or claim 10.

12. A sentence simplification method for simplifying an original sentence having a predicate using "mashou" comprising the steps of: a second phrase expressing the first phrase in a simpler form is stored in a storage means for each of a plurality of first phrases using "mashou"; causing a computer to execute a process of generating a paraphrase sentence by replacing, when any of the plurality of first words and phrases is included in the original sentence, the first word and phrase included in the original sentence with the second word and phrase stored in the storage means of the included first word and phrase; 13. A sentence simplification method comprising the steps of:

13. A computer program for use in a computer that can access a storage means for simplifying an original sentence having a predicate using "mashou" and storing, for each of a plurality of first phrases using "mashou," a second phrase that expresses the first phrase in a simpler form, causing the computer to execute a process of generating a paraphrase sentence by replacing, when any of the plurality of first words and phrases is included in the original sentence, the first word and phrase included in the original sentence with the second word and phrase stored in the storage means of the included first word and phrase; A computer program comprising:

Images