JP2021179665A - Sentence creation device - Google Patents

Abstract

To provide a sentence creation device capable of more easily creating a bilingual corpus for simultaneous interpretation.SOLUTION: A sentence creation device 1 includes: an acquisition unit 10 for acquiring a first sentence T11 of a first language and a second sentence T21 of a second language being a translation of the first sentence T11; an analysis unit 20 for performing morphological analysis to each of the first sentence T11 and the second sentence T21; a detection unit 30 for detecting all of phrases having morphemes in the second sentence T21; a mapping unit 40 for mapping morphemes of the first sentence T11 and morphemes of the second sentence T21 on the basis of meanings of respective morphemes; a deriving unit 50 for deriving a feature quantity in each phrase of the second sentence T21 on the basis of indexes indicating positions of the morphemes, being mapped to the morphemes of the second sentence T21, of the first sentence T11 in the first sentence T11; and a creation unit 60 for creating a third sentence T31 by arranging phrases of the second sentence T21 on the basis of the feature quantity of each phrase of the second sentence T21.SELECTED DRAWING: Figure 1

Description

The present invention relates to a writing device.

Patent Document 1 describes a rearrangement model generator that converts a second language sentence into a word order of the first language. The rearrangement model generator associates each word constituting the first language sentence with a word constituting the second language sentence having the same meaning as each word constituting the first language sentence, and then converts each phrase constituting the second language sentence into the first language sentence. Corresponds to only one word that composes, and rearranges the phrases that compose the second language sentence based on the order of the words associated with each phrase in the first language sentence. The sort model generator erases the correspondence of the function words included in the phrase of the second language sentence in order to associate each phrase constituting the second language sentence with the only word constituting the first language sentence. , Leave the correspondence of the independent words that are most likely to be associated.

Japanese Unexamined Patent Publication No. 2013-117888

By the way, in the existing bilingual corpus, translations between two languages suitable for sequential interpretation are aggregated, so translations suitable for simultaneous interpretation are not aggregated. When simultaneous interpretation is performed using a translation model constructed based on an existing bilingual corpus, the waiting time from the sequential input of sentences in the target language to the output of the translated text may increase.

Here, it is conceivable to create a parallel translation corpus for simultaneous interpretation by using the rearrangement model generator described in Patent Document 1. The rearrangement model generation device of Patent Document 1 selects an independent word in a phrase in the mapping, and rearranges the phrase based on the independent word. Therefore, the rearrangement model generator of Patent Document 1 performs a process of discriminating an independent word or a function word in a second language sentence and a process of leaving an association of the independent words with the most probable association. Therefore, the processing is complicated. For this reason, there is a demand for a sentence creation device that can more easily create a bilingual corpus for simultaneous interpretation.

An object of the present invention is to provide a sentence creation device capable of more easily creating a bilingual corpus for simultaneous interpretation.

The sentence creation device according to one aspect of the present invention includes an acquisition unit for acquiring a first sentence in the first language and a second sentence in the second language which is a translation of the first sentence, and the first sentence and the second sentence. An analysis unit that analyzes each morpheme, a detection unit that detects all clauses that have morphemes in the second sentence, and a mapping unit that associates the morpheme of the first sentence with the morpheme of the second sentence based on the meaning of each morpheme. Based on the index indicating the position of the morpheme of the first sentence in the first sentence associated with the morpheme of the second sentence, the derivation unit for deriving the feature amount in each clause of the second sentence, and each of the second sentences. It is provided with a creation unit for arranging the phrases of the second sentence and creating the third sentence based on the feature amount of the phrase.

According to this sentence creation device, the phrases of the second sentence are arranged according to the feature amount of each phrase of the second sentence, and the third sentence is created. Since the feature amount is derived based on the index of the morpheme of the first sentence, complicated processing is not required to arrange the clauses of the second sentence, and the time to create the third sentence is shortened. Can be done. Since the third sentence is a sentence considering the position of each morpheme of the first sentence, for example, the third sentence can be created as a sentence in which the first sentence is simultaneously interpreted in a second language. Therefore, it is possible to more easily create a bilingual corpus for simultaneous interpretation.

According to the present invention, it is possible to more easily create a bilingual corpus for simultaneous interpretation.

FIG. 1 is a diagram showing a configuration of a text creation device according to an embodiment. FIG. 2 is a diagram showing an example of morphological analysis processing by the analysis unit shown in FIG. FIG. 3 is a diagram showing an example of a phrase-separating process by the detection unit shown in FIG. FIG. 4 is a diagram showing an example of a morpheme-to-morpheme mapping process by the mapping section shown in FIG. 1 and an example of a feature quantity derivation process by the derivation section shown in FIG. FIG. 5 is a diagram showing an example of the sorting process by the creating unit shown in FIG. FIG. 6 is a flowchart showing a series of processes of the sentence creation method executed by the sentence creation device shown in FIG. FIG. 7 is a diagram showing a configuration of a text creation device according to another embodiment. FIG. 8 is a diagram showing a hardware configuration of the text creation device according to the embodiment.

An embodiment of the present invention will be described with reference to the accompanying drawings. When possible, the same parts are designated by the same reference numerals, and duplicate description is omitted.

FIG. 1 is a diagram showing a configuration of a text creation device according to an embodiment. The sentence creation device 1 shown in FIG. 1 is a device that creates a third sentence T31 in a second language by rearranging the second sentence T21 in the second language in the word order of the first sentence T11 in the first language. .. The second sentence T21 is a sentence (translated sentence) obtained by translating the first sentence T11 into a second language. The first sentence T11, the second sentence T21, and the third sentence T31 are, for example, text data.

The first language and the second language are different languages from each other. For example, the first language is English and the second language is Japanese. The sentence pattern (word order) of the sentence composed in the first language may be different from the sentence pattern (word order) of the sentence composed in the second language. For example, the order of the subject (S: Subject), the object (O: Object), and the verb (V: Verb) is different between the sentence of the first language and the sentence of the second language. For example, a typical sentence pattern in a sentence in a first language is an SVO type, and a typical sentence pattern in a sentence in a second language is an SOV type. For example, the sentence pattern of the third sentence T31 is different from the sentence pattern of the second sentence T21 and is the same as the sentence pattern of the first sentence T11. Even when the first sentence T11 includes a complement (C: Complement) or a modifier (M: Modifier), the sentence pattern of the third sentence T31 is the same as the sentence pattern of the first sentence T11.

The text creation device 1 is realized by, for example, a server device. The text creation device 1 may be realized by a plurality of server devices, that is, a computer system. The text creation device 1 is configured to be communicable with the first translation corpus 82 and the second translation corpus 84 provided outside the text creation device 1.

The first translation corpus 82 and the second translation corpus 84 are functional units that function as databases for storing information, respectively. The first bilingual corpus 82 and the second bilingual corpus 84 are each implemented, for example, by a database, server, or other suitable medium containing at least one of memory and storage.

The first bilingual corpus 82 is, for example, a bilingual corpus for sequential interpretation. The second bilingual corpus 84 is, for example, a bilingual corpus for simultaneous interpretation. A bilingual corpus is a corpus (sentence database) of bilingual data in which sentences and sentences in different languages are collected in the form of bilingual translation, which is constructed for use as learning data for machine translation. That is, the first bilingual corpus 82 includes a plurality of first bilingual data, and each of the first bilingual data includes a first sentence T11 associated with each other and a second sentence T21 which is a translation of the first sentence. It is a combination. The second bilingual corpus 84 includes a plurality of second bilingual data, and each second bilingual data has a phrase of the first sentence T11 associated with each other and the phrase of the second sentence T21 which is a translation of the first sentence. It is a combination with the changed third sentence T31.

Subsequently, the function of the text creation device 1 according to the present embodiment will be described. As shown in FIG. 1, the sentence creation device 1 functionally includes an acquisition unit 10, an analysis unit 20, a detection unit 30, an association unit 40, a derivation unit 50, and a creation unit 60. ..

The acquisition unit 10 is a functional unit that acquires the first bilingual data of the first sentence T11 and the second sentence T21. The acquisition unit 10 is configured to be able to acquire information from the first translation corpus 82. The acquisition unit 10 acquires, for example, the first translation data from the first translation corpus 82. The acquisition unit 10 acquires, for example, the first translation data including the first sentence T11 that is not stored in the second translation corpus 84 from the first translation corpus 82. The acquisition unit 10 outputs the acquired first sentence T11 and second sentence T21 to the analysis unit 20. The acquisition unit 10 outputs the first sentence T11 to the creation unit 60.

The analysis unit 20 is a functional unit that morphologically analyzes each of the first sentence T11 and the second sentence T21. The morphological analysis is, for example, a process of dividing text data into columns of morphemes (Morpheme) based on information such as grammar and part of speech of a certain target language, and discriminating the part of speech of each morpheme. A morpheme is, for example, the smallest linguistic unit that has meaning. The analysis unit 20 analyzes each of the first sentence T11 and the second sentence T21 by using, for example, a known morphological analysis method. The known morphological analysis method is, for example, a method based on a conditional random field (CRF), a Hidden Markov Model (Hidden Markov Model), or a recurrent neural network. The known morphological analysis method may be, for example, a rule-based method. As a specific morphological analysis tool, the analysis unit 20 applies, for example, NLTK (Natural Language Toolkit) to the first sentence T11 and MeCab (Yet Another Part-of-Speech and Morphological Analyzer) to the second sentence T21. Apply. The analysis unit 20 may execute the analysis process using words instead of morphemes.

FIG. 2 is a diagram showing an example of morphological analysis processing by the analysis unit shown in FIG. As shown in FIG. 2, the analysis unit 20 morphologically analyzes each of the first sentence T11 and the second sentence T21 acquired by the acquisition unit 10. For example, the analysis unit 20 performs morphological analysis on the first sentence T11, and performs morphological elements 2a, 2b, 2c, 2d, 2e, 2f, 2g, 2h, 2i in order from the beginning of the sentence to the end of the first sentence T11. Extract. For example, the analysis unit 20 performs morphological analysis on the second sentence T21, and the morphemes 3a, 3b, 3c, 3d, 3e, 3f, 3g, 3h, 3i, in order from the beginning to the end of the second sentence T21. Extract 3j and 3k. The analysis unit 20 extracts all the morphemes constituting the first sentence T11 and the second sentence T21 as described above.

The analysis unit 20 assigns an index 2p to the morphemes 2a to 2i. The index 2p indicates the position (order) of the morphemes in the morphemes 2a to 2i in the first sentence T11. The index 2p may be an ascending number increasing from the beginning of the first sentence T11 toward the end of the sentence, or a descending number decreasing from the beginning of the first sentence T11 toward the end of the sentence. In the present embodiment, the analysis unit 20 assigns ascending numbers to each of the morphemes 2a to 2i as the index 2p. The analysis unit 20 assigns "0" to the morpheme 2a located at the beginning of the sentence in the first sentence T11, for example. The analysis unit 20 assigns a number obtained by adding 1 to the number assigned to the immediately preceding morpheme for each of the subsequent morphemes 2b to 2i.

The analysis unit 20 outputs the array of the morphemes 2a to 2i and the index 2p to the association unit 40 as the first analysis sentence T12. The analysis unit 20 outputs the arrangement of the morphemes 3a to 3k to the detection unit 30 as the second analysis sentence T22.

The detection unit 30 is a functional unit that detects all the clauses in the second sentence T21. The detection unit 30 detects the phrase of the second sentence T21 by, for example, using the second analysis sentence T22 and performing a phrase delimiter process for dividing between the phrases. By separating the clauses of the second analysis sentence T22 by the clause break processing by the detection unit 30, between two adjacent breaks, between the beginning and the first break, and between the last break and the end of the sentence. Each phrase is detected as a phrase. The first delimiter is the delimiter closest to the beginning of the sentence among the plurality of delimiters. The last delimiter is the delimiter closest to the end of the sentence among the delimiters. The detection unit 30 executes the phrase-breaking process by using, for example, a known phrase-breaking method. The known phrase-separating method is, for example, a method based on a support vector machine or a neural network. As a specific phrase delimiter tool, the detection unit 30 uses, for example, CaboCha (Yet Another Japanese Dependency Structure Analyzer).

FIG. 3 is a diagram showing an example of a phrase-separating process by the detection unit shown in FIG. As shown in FIG. 3, the detection unit 30 performs a phrase delimiter process on the second analysis sentence T22, for example. Specifically, the detection unit 30 divides the second analysis sentence T22 by inserting the division 4 into the second analysis sentence T22. The detection unit 30 detects an array of morphemes (morphemes 3a, 3b) located between the beginning of the sentence and the first delimiter 4 as a clause 5a. The detection unit 30 detects an array of morphemes (morphemes 3c, 3d, 3e) located between the first delimiter 4 and the second delimiter 4 as clause 5b. The detection unit 30 detects an array of morphemes (morphemes 3f, 3g) located between the second delimiter 4 from the beginning of the sentence and the last delimiter 4 as the clause 5c. The detection unit 30 detects an array of morphemes (morphemes 3h, 3i, 3j, 3k) located between the last delimiter 4 and the end of the sentence as the clause 5d. As described above, the detection unit 30 detects all the clauses 5a to 5d in the second analysis sentence T22.

The detection unit 30 outputs the array of the clauses 5a to 5d to the association unit 40 as the detection sentence T23. In addition, the detection unit 30 may execute the phrase delimiter process for the second sentence T21 instead of the second analysis sentence T22. The detection unit 30 may detect only a part of the second analysis sentence T22 instead of all the phrases.

The association unit 40 is a functional unit that associates the morphemes 2a to 2i of the first sentence T11 with the morphemes 3a to 3k of the second sentence T21 based on the meaning of each morpheme. The mapping unit 40 performs, for example, a mapping process (word alignment process). The word alignment process is a process for establishing a correspondence between morphemes 2a to 2i and morphemes 3a to 3k. For example, in the word alignment process, if the meaning of one of the morphemes 2a to 2i and the meaning of at least one of the morphemes 3a to 3k are similar, it is determined that there is a correspondence between the morphemes. Will be done. The association unit 40 executes the word alignment process by using, for example, a known word alignment method. The known word alignment method is, for example, a probability model or a heuristic-based alignment method. As a specific word alignment tool, the detection unit 30 uses, for example, the IBM model 2. The detection unit 30 may use, for example, an IBM model 1, an IBM model 3, an IBM model 4, or an IBM model 5 as a word alignment tool.

FIG. 4 is a diagram showing an example of a morpheme-to-morpheme mapping process by the mapping section shown in FIG. 1 and an example of a feature quantity derivation process by the derivation section shown in FIG. As shown in FIG. 4, the mapping unit 40 executes the mapping process using, for example, the first analysis sentence T12 and the detection sentence T23. The mapping unit 40 may execute the mapping process using the first analysis sentence T12 and the second analysis sentence T22. For example, one morpheme among the morphemes 2a to 2i is associated with one morpheme included in the morphemes 3a to 3k. The morphemes 2a to 2i are associated with one or more morphemes contained in the morphemes 3a to 3k, but may not be associated with any of the morphemes.

For example, the associating unit 40 searches for morphemes having similar meanings to the morphemes 2b from the morphemes 3a to 3k, determines that the morphemes 3c are similar, and associates the morphemes 2b with the morphemes 3c. For example, the associating unit 40 searches for morphemes having similar meanings to the morphemes 2a from the morphemes 3a to 3k, determines that the morphemes 3e and the morphemes 3i are similar, associates the morphemes 2a with the morphemes 3e, and associates the morphemes 2a with the morphemes 3e. The 2a and the morpheme 3i are associated with each other. In the example shown in FIG. 4, the associating unit 40 associates all the morphemes of the morphemes 3a to 3k with any of the morphemes 2a to 2i, for example. The association unit 40 does not have to associate any of the morphemes 3a to 3k with the morphemes 2a to 2i. The association unit 40 outputs information indicating the correspondence between the morphemes, the first analysis sentence T12 and the detection sentence T23 to the derivation unit 50.

The derivation unit 50 is a functional unit that derives the feature quantities of the clauses 5a to 5d. The feature amount is derived based on the index 2p of the morphemes 2a to 2i associated with the morphemes included in each clause. The derivation unit 50 extracts the index 2p of the morpheme of the first sentence T11 associated with the plurality of morphemes included in the clause for each of the clauses 5a to 5d, and sets the minimum value among the extracted combinations of the index 2p. Derived as a feature quantity.

Hereinafter, the feature quantity derivation process will be described in detail. As shown in FIG. 4, the derivation unit 50 extracts the index 2p of the morphemes 2a to 2i associated with the morphemes 3a to 3k for each clause of the clauses 5a to 5d. For example, the first clause, clause 5a, includes morphemes 3a and morphemes 3b. The morpheme 3a is associated with the morpheme 2h, and the morpheme 3b is associated with the morpheme 2c. Therefore, the derivation unit 50 extracts "7" which is the index 2p of the morpheme 2h and "2" which is the index 2p of the morpheme 2c for the clause 5a, and combines these indexes 2p (7, 2). ) Is obtained as the extraction amount 7a. The association unit 40 performs the same processing on the clauses 5b, 5c, and 5d to obtain the extraction amounts 7b, 7c, and 7d of the clauses 5b, 5c, and 5d. The extraction amount 7b of the clause 5b is (1,2,0), the extraction amount 7c of the clause 5c is (5,4), and the extraction amount 7d of the clause 5d is (3,0,8,8). ..

The derivation unit 50 derives the feature quantities 8a to 8d based on the extraction amounts 7a to 7d, respectively. In the example shown in FIG. 4, the derivation unit 50 derives the minimum value of the index 2p included in the extraction amount as the feature amount. For example, the derivation unit 50 derives the minimum value “2” from the extraction amount 7a (7, 2) as the feature amount 8a of the clause 5a. The derivation unit 50 derives the feature amount 8b of the clause 5b, the feature amount 8c of the clause 5c, and the feature amount 8d of the clause 5d by performing the same processing for the clauses 5b, 5c, and 5d. The feature amount 8b of the clause 5b is "0", the feature amount 8c of the clause 5c is "4", and the feature amount 8d of the clause 5d is "0". The derivation unit 50 outputs the detection text T23 and the feature quantities 8a to 8d to the creation unit 60.

The creating unit 60 is a functional unit that creates the third sentence T31 by arranging the clauses 5a to 5d based on the feature quantities 8a to 8d. The creation unit 60 executes, for example, a rearrangement process for rearranging the clauses 5a to 5d to create the third sentence T31. In the third sentence T31, the first sentence T11 composed of the first language is translated into the second language, and the clauses 5a to 5d described in the second language are arranged according to the sentence pattern of the first sentence T11. It becomes. For example, the creating unit 60 arranges the clauses 5a to 5d so that the features 8a to 8d correspond to the order of the indexes 2p of the morphemes 2a to 2i arranged from the beginning to the end of the first sentence T11 to form the third sentence T31. create. For example, when the indexes 2p are assigned to the morphemes 2a to 2i in ascending order from the beginning to the end of the first sentence T11, the creating unit 60 sets the clauses 5a to 5d so that the feature quantities 8a to 8d are in ascending order. The third sentence T31 is created by sorting.

When the feature quantities in the plurality of clauses have the same value, the creating unit 60 arranges the plurality of clauses according to the order of the clauses in the second sentence T21. That is, when the feature quantities in the plurality of clauses have the same value, the order of the plurality of clauses in the third sentence T31 retains the order of the plurality of clauses in the second sentence T21.

FIG. 5 is a diagram showing an example of the sorting process by the creating unit shown in FIG. As shown in FIG. 5, the creating unit 60 sorts the feature quantities 8a to 8d in ascending order. In the example shown in FIG. 5, the feature amount 8a is "2", the feature amount 8b is "0", the feature amount 8c is "4", and the feature amount 8d is "0". Therefore, the creating unit 60 rearranges the feature amount 8b, the feature amount 8d, the feature amount 8a, and the feature amount 8c in this order. Then, the creating unit 60 arranges (rearranges) the clauses 5a to 5d in the order of the clause 5b, the clause 5d, the clause 5a, and the clause 5c from the beginning of the sentence to the end of the sentence according to the order of the features, so that the third sentence T31 To create. Since the feature amount 8b and the feature amount 8d have the same value, the creating unit 60 keeps the order of the original clauses and arranges them in the order of clauses 5b and 5d.

In FIG. 5, for convenience of explanation, the delimiter 4 is left in the third sentence T31, but the creating unit 60 creates the third sentence T31 by removing the delimiter 4 and rearranging the clauses 5a to 5d. You may. The creating unit 60 outputs the combination of the first sentence T11 and the third sentence T31 acquired from the acquisition unit 10 to the second bilingual corpus 84 as the second bilingual data, and stores it in the second bilingual corpus 84.

FIG. 6 is a flowchart showing a series of processes of the sentence creation method executed by the sentence creation device shown in FIG. The series of processes shown in FIG. 6 is started, for example, by storing new first translation data in the first translation corpus 82. As shown in FIG. 6, first, the acquisition unit 10 executes the acquisition process (step S10). In step S10, the acquisition unit 10 acquires the first translation data including the first sentence T11 and the second sentence T21 from the first translation corpus 82. Then, the acquisition unit 10 outputs the first parallel translation data to the analysis unit 20.

Subsequently, the analysis unit 20 executes the morphological analysis process (step S20). In step S20, when the analysis unit 20 receives the first bilingual data from the acquisition unit 10, it morphologically analyzes each of the first sentence T11 and the second sentence T21, thereby performing the first analysis sentence T12 and the second analysis sentence T22. To create. Then, the analysis unit 20 outputs the first analysis sentence T12 to the association unit 40, and outputs the second analysis sentence T22 to the detection unit 30.

Subsequently, the detection unit 30 executes the phrase delimiter process (step S30). In step S30, when the detection unit 30 receives the second analysis sentence T22 from the analysis unit 20, the detection unit 30 inserts a delimiter 4 into the second analysis sentence T22 to delimit the second analysis sentence T22. By this process, the detection unit 30 detects the clauses 5a to 5d and creates the detection sentence T23. Then, the detection unit 30 outputs the detection sentence T23 to the association unit 40.

Subsequently, the mapping unit 40 executes the mapping process (step S40). In step S40, when the matching unit 40 receives the first analysis sentence T12 from the analysis unit 20 and the detection sentence T23 from the detection unit 30, the morphemes 2a to 2i of the first analysis sentence T12 and the morpheme 3a of the detection sentence T23 ~ 3k are associated with each other based on the meaning of each morpheme. Then, the mapping unit 40 outputs the information indicating the correspondence between the morphemes, the first analysis sentence T12 and the detection sentence T23 to the derivation unit 50.

Subsequently, the derivation unit 50 executes the feature amount derivation process (step S50). In step S50, when the derivation unit 50 receives the first analysis sentence T12 and the detection sentence T23 from the association unit 40, the derivation unit 50 derives the feature quantities 8a to 8d of the clauses 5a to 5d, respectively. Then, the derivation unit 50 outputs the detection text T23 and the feature quantities 8a to 8d to the creation unit 60.

Subsequently, the creating unit 60 executes the sorting process (step S60). In step S60, when the creating unit 60 receives the detected sentences T23 and the feature amounts 8a to 8d from the derivation unit 50, the creating unit 60 arranges the clauses 5a to 5d based on the feature amounts 8a to 8d to create the third sentence T31.

Subsequently, the creation unit 60 executes the output process (step S70). In step S70, the creating unit 60 outputs the combination of the third sentence T31 and the first sentence T11 received from the acquisition unit 10 to the second bilingual corpus 84 as the second bilingual data. With this output, the creation unit 60 stores the second translation data in the second translation corpus 84. This completes a series of processing of the sentence creation method.

The text creation device 1 executes, for example, the processing of the flowchart shown in FIG. 6 for all the first translation data in the first translation corpus 82. When the processing for the first translation data in the text creation device 1 is completed, the text creation device 1 determines whether or not the unprocessed first translation data is in the first translation corpus 82. When the sentence creation device 1 determines that there is unprocessed first bilingual data, the acquisition unit 10 acquires the unprocessed first bilingual data from the first bilingual corpus 82. The sentence creation device 1 creates a second bilingual data corresponding to each of the first bilingual data, and stores the second bilingual data in the second bilingual corpus 84. With this configuration, the sentence creation device 1 can create a second translation corpus 84 corresponding to the first translation corpus 82. When the sentence creation device 1 determines that the unprocessed first translation data is not in the first translation corpus 82, the sentence creation device 1 ends the processing for the first translation corpus 82.

In the flowchart shown in FIG. 6, if the sentence creation device 1 is after the acquisition process (step S10) and before the feature amount derivation process (step S50), the sentence segmentation process (step S30) is performed. You may run it at any time.

In the sentence creating device 1 described above, the phrases of the detected sentence T23 are arranged by the respective feature quantities 8a to 8d of the phrases 5a to 5d of the detected sentence T23, and the third sentence T31 is created. For example, compared to the process of discriminating an independent word or a function word for a sentence in a second language and the process of leaving the correspondence of the independent words with the most probable correspondence, the sentence is created. The device 1 does not require complicated processing up to the rearrangement process (step S60) for arranging the clauses 5a to 5d of the detected sentence T23, and can shorten the time until the third sentence T31 is created. Since the third sentence T31 is a sentence considering the positions of the morphemes 2a to 2i of the first sentence T11, the third sentence T31 is created as a sentence obtained by simultaneously interpreting the first sentence T11 in a second language. Therefore, it is possible to more easily create a second translation corpus 84, which is a translation corpus for simultaneous interpretation. A translation model constructed based on the second translation corpus 84 (translation corpus for simultaneous interpretation) in which the second translation data is stored can, for example, simultaneously translate a sentence in the first language into a second language. become.

As described above, the creating unit 60 sets the clauses so that the feature quantities 8a to 8d of the clauses 5a to 5d correspond to the order (ascending order) of the indexes 2p of the morphemes 2a to 2i arranged from the beginning to the end of the first sentence T11. The third sentence T31 is created by arranging each of 5a to 5d. Since the third sentence T31 is a sentence corresponding to the order of the indexes 2p of the morphemes 2a to 2i, the sentence creation device 1 creates the third sentence T31 as a sentence obtained by simultaneously interpreting the first sentence T11 in the second language. Can be done.

As described above, the derivation unit 50 sets the minimum value of the index 2p of the morpheme of the first analysis sentence T12 associated with the morpheme included in each of the clauses 5a to 5d as the feature amount of each of the clauses 5a to 5d. .. In the above embodiment, the indexes 2p are assigned in ascending order from the beginning to the end of the first analysis sentence T12 (first sentence T11), so that the creating unit 60 arranges the feature quantities 8a to 8d in ascending order. , The third sentence T31 is created by arranging the clauses 5a to 5d. In other words, the clauses are sorted based on the smallest value (index 2p) of the extraction amount of each clause. With this configuration, the phrases corresponding to the morphemes near the beginning of the first sentence T11 can be arranged in order. Therefore, the created second translation corpus 84 (translation corpus for simultaneous interpretation) can train the translation model so that the morphemes of the first sentence T11 that are sequentially input can be translated first. The morpheme contained in the first sentence T11 needs to be translated earlier as it is closer to the beginning of the sentence. Therefore, in the translation model, the time from the input of the first sentence T11 to the translation of the third sentence T31 can be shortened.

It should be noted that all the clauses in the third sentence T31 do not necessarily have to be arranged in order from the clause corresponding to the morpheme close to the beginning of the first sentence T11. For example, as shown in FIG. 5, all the morphemes (morphemes 2e and morphemes 2f) of the first sentence T11 corresponding to the clause 5c among the clauses 5a to 5d are at least the morphemes of the first sentence T11 corresponding to the clause 5a. Even if it is located closer to the beginning of the sentence than one (morpheme 2h), the phrase 5c may be located closer to the end of the sentence than the phrase 5a in the third sentence T31. Even in this case, at least one or more clauses (for example, clauses 5b and 5d) among the clauses 5a to 5d are arranged in order from the clause corresponding to the morpheme close to the beginning of the first sentence T11. Therefore, the created second translation corpus 84 (translation corpus for simultaneous interpretation) can train the translation model so that at least a part of the first sentence T11 that is sequentially input can be translated first. For example, when the translation model is trained using the first translation corpus 82, in the example of FIG. 5, the translation model cannot output anything until the morpheme 2h is input. On the other hand, when the translation model is trained using the second translation corpus 84, in the example of FIG. 5, when the morpheme 2b is input, the translation model is the first of the first clauses (phrases 5b) of the translated sentence. The morpheme 3c can be output. Therefore, in the translation model, it is possible to shorten the time from the input of the first sentence T11 to the translation of at least a part of the third sentence T31.

The present invention is not limited to the above-described embodiment. For example, the feature amount does not have to be the minimum value of the index 2p of the morpheme of the first analysis sentence T12 associated with the morpheme in the clauses 5a to 5d. The feature amount may be, for example, the maximum value of the index 2p of the morpheme of the first analysis sentence T12 associated with the morpheme included in each of the clauses 5a to 5d.

In this case, the derivation unit 50 outputs the maximum value of each of the extraction amounts 7a to 7d as the feature amount. In the example shown in FIG. 4, the derivation unit 50 derives the maximum value “7” from the extraction amount 7a (7, 2) as the feature amount 8a of the clause 5a. The derivation unit 50 derives the feature amount 8b of the clause 5b, the feature amount 8c of the clause 5c, and the feature amount 8d of the clause 5d by performing the same processing for the clauses 5b, 5c, and 5d. In this case, the feature amount 8b of the clause 5b is "2", the feature amount 8c of the clause 5c is "5", and the feature amount 8d of the clause 5d is "8".

As described above, the derivation unit 50 uses the maximum value of each of the extraction amounts of the clauses 5a to 5d as the feature amount, and the creation unit 60 sets the feature amount of each of the clauses 5a to 5d from the beginning of the first analysis sentence T12. The third sentence T31 may be created by arranging the clauses 5a to 5d so as to correspond to the order of the indexes 2p of the morphemes 2a to 2i arranged at the end of the sentence. In this modification, the indexes 2p are assigned in ascending order from the beginning to the end of the first analysis sentence T12 (first sentence T11), so that the creating unit 60 arranges the feature quantities 8a to 8d in ascending order. , The clauses 5a to 5d are arranged to create the third sentence T31. In other words, the clauses are sorted based on the largest value (index 2p) of the extraction amount of each clause. With this configuration, the clauses are arranged in the order in which all the associated morphemes are aligned from the beginning to the end of the first sentence T11. Therefore, in the created second translation corpus 84 (translation corpus for simultaneous interpretation), when the first sentence T11 is sequentially input from the beginning of the sentence, priority is given to the order in which all the morphemes corresponding to the clauses of the second sentence T21 are prepared. The translation model can be trained so that it can be output to. In this translation model, since the phrase is translated after all the morphemes of the first sentence T11 corresponding to the phrase are prepared, the accuracy of translation in the second language for the first sentence T11 represented by the first language is correct. Can be improved.

Hereinafter, another embodiment will be described. FIG. 7 is a diagram showing a configuration of a text creation device according to another embodiment. As shown in FIG. 7, the text creation device 1A is mainly different from the text creation device 1 in that the weighting unit 70 is further provided and the feature quantity derivation process of the derivation unit 50 is provided.

The weighting unit 70 is a functional unit that sets weights for the morphemes 2a to 2i of the first analysis sentence T12 as a weighting process. The weight indicates, for example, the importance of each morpheme appearing in a sentence. The higher the weight, the higher the importance of the morpheme. The weight is, for example, a value of 0 or more and 1 or less. The weighting unit 70 executes, for example, a weighting process for setting weights by using a known weighting method. The known weighting method is, for example, a method based on the appearance frequency of morphemes in a document (text) or the inverse document frequency of morphemes in a document (text). The document contains multiple sentences. The frequency of appearance of a morpheme represents the ratio of the number of appearances of a target morpheme to the number of appearances of all words in a document or sentence. The reciprocal of a morpheme represents the reciprocal of the proportion of documents containing a certain morpheme in a set of documents, and the more morphemes appear in other documents, the smaller the value. Specific weighting tools include, for example, TF-IDF and Okapi BM25. As a weighting method, weighting by a user operation may be executed.

The weighting unit 70 executes a weighting process on the first analysis sentence T12 that has been morphologically analyzed by the analysis unit 20. For example, in the first analysis sentence T12, the weighting unit 70 sets a larger weight for the morpheme 2b, the morpheme 2d, the morpheme 2f, and the morpheme 2h than for the other morphemes. The weighting unit 70 outputs the weights of the first analysis sentence T12 and the morphemes 2a to 2i to the derivation unit 50.

The derivation unit 50 derives each feature amount of the clauses 5a to 5d by using the result of the weighting process by the weighting unit 70 described above. The derivation unit 50 derives the index 2p of the morpheme having the largest weight among the morphemes of the first analysis sentence T12 associated with the morphemes included in each of the clauses 5a to 5d as the feature quantities of the clauses 5a to 5d. do. For example, the derivation unit 50 is the index 2p assigned to the morpheme 2h, which is the morpheme (highly important morpheme) to which the largest weight is set among the morphemes (morpheme 2c and morpheme 2h) associated with the clause 5a. "7" is derived as the feature quantity of the clause 5a.

The derivation unit 50 performs the same processing on the clauses 5b, 5c, and 5d to obtain the feature quantities of the clauses 5b, 5c, and 5d, respectively. The feature amount of the clause 5b is "1", the feature amount of the clause 5c is "5", and the feature amount of the clause 5d is "3".

In the flowchart of FIG. 6, the weighting unit 70 executes the weighting process, for example, after the mapping process (step S40) and before the feature amount derivation process (step S50).

The sentence creation device 1 may execute the weighting process (step S45) at any time after the morphological analysis process (step S20) and before the feature amount derivation process (step S50). The association unit 40 may output the detection sentence T23 to the derivation unit 50 without outputting it to the weighting unit 70.

As described above, the sentence creation device 1A also has the same effect as the sentence creation device 1. Further, in the sentence creation device 1A, the index 2p of the morpheme having the highest weight among the morphemes of the first analysis sentence T12 associated with the morphemes included in each of the clauses 5a to 5d is a feature of each of the clauses 5a to 5d. Derived as a quantity. Then, the third clauses 5a to 5d are arranged so that the feature quantities of the clauses 5a to 5d correspond to the order of the indexes 2p of the morphemes 2a to 2i arranged from the beginning to the end of the first analysis sentence T12. The sentence T31A is created. In the sentence creation device 1, the phrases of the detected sentence T23 associated with the morpheme are rearranged in the order in which the morphemes of high importance appear from the beginning of the sentence to the end of the first analysis sentence T12. The created second translation corpus 84 (translation corpus for simultaneous interpretation) is translated so that the morpheme of the first sentence T11, which is of high importance, can be translated into the third sentence T31A among the first sentences T11 that are sequentially input. You can train the model. Therefore, the translation model can shorten the time until the phrase including the morpheme that translates the highly important morpheme of the first sentence T11 is output.

The derivation unit 50 sets an index 2p of the morpheme of the first analysis sentence T12 whose weight is equal to or more than a predetermined threshold among the morphemes of the first analysis sentence T12 associated with the morphemes included in each of the clauses 5a to 5d. It may be derived as a feature amount for each of the clauses 5a to 5d. When there are a plurality of morphemes whose weights are equal to or greater than a predetermined threshold value in a certain clause of the first analysis sentence T12, the derivation unit 50 may, for example, use the minimum value or the maximum value of the index 2p assigned to the plurality of morphemes. May be derived as a feature quantity.

When the minimum value of the index 2p is used as the feature quantity, the second translation corpus 84 (translation corpus for simultaneous interpretation) has the morphemes that are sequentially input among the morphemes of high importance in the first sentence T11. You can train the translation model so that you can translate quickly. When the maximum value of the index 2p is used as the feature quantity, the second translation corpus 84 (translation corpus for simultaneous interpretation) is the first sentence T11 that is sequentially input among the highly important morphemes of the first sentence T11. The translation model can be trained so that the phrase of the second sentence T21 corresponding to the morpheme of can be output in order from the phrase.

In another embodiment described above, the weighting unit 70 may set weights for the morphemes 3a to 3k of the detection sentence T23 as the weighting process. For example, in the detection sentence T23, the weighting unit 70 sets a larger weight for the morpheme 3a, the morpheme 3c, the morpheme 3f, and the morpheme 3h than for the other morphemes.

In this case, the derivation unit 50 derives the index 2p of the morpheme of the first analysis sentence T12 associated with the morpheme having the largest weight contained in each of the clauses 5a to 5d as the feature quantity of each of the clauses 5a to 5d. .. The derivation unit 50 extracts, for example, the morpheme 3a which is the morpheme (the morpheme of high importance) to which the largest weight is set among the morpheme 3a and the morpheme 3b in the clause 5a. The derivation unit 50 derives "7" of the index 2p assigned to the morpheme 2h associated with the morpheme 3a as a feature amount of the clause 5a.

The derivation unit 50 performs the same processing on the clauses 5b, 5c, and 5d to obtain the feature quantities in each of the clauses 5b, 5c, and 5d. The feature amount of the clause 5b is "1", the feature amount of the clause 5c is "5", and the feature amount of the clause 5d is "3".

In the sentence creation device 1, the detection sentence T23 associated with the morpheme associated with the morpheme of high importance of the second sentence T21 appears from the beginning of the sentence to the end of the first analysis sentence T12. The clauses are rearranged. The created second translation corpus 84 (translation corpus for simultaneous interpretation) is a morpheme of the first sentence T11 associated with a morpheme of high importance of the second sentence T21 among the first sentences T11 that are sequentially input. The translation model can be trained so that it can be translated into the third sentence T31A. Therefore, the translation model can shorten the time until the phrase including the highly important morpheme of the second sentence T21 is output.

The writing devices 1 and 1A may include at least one of the first translation corpus 82 and the second translation corpus 84.

The block diagram used in the description of the above embodiment shows a block of functional units. These functional blocks (components) are realized by any combination of at least one of hardware and software. The method of realizing each functional block is not particularly limited. That is, each functional block may be realized using one physically or logically coupled device, or two or more physically or logically separated devices can be directly or indirectly (eg, for example). , Wired, wireless, etc.) and may be realized using these plurality of devices. The functional block may be realized by combining the software with the one device or the plurality of devices.

Functions include judgment, decision, judgment, calculation, calculation, processing, derivation, investigation, search, confirmation, reception, transmission, output, access, solution, selection, selection, establishment, comparison, assumption, expectation, and assumption. These include broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, and assigning. Not limited to functions. For example, a functional block (constituent unit) that causes transmission to function is referred to as a transmitting unit or a transmitter. In each case, as described above, the realization method is not particularly limited.

For example, the text creation devices 1 and 1A in one embodiment of the present disclosure may function as a computer that performs information processing of the present disclosure. FIG. 8 is a diagram showing an example of the hardware configuration of the text creation devices 1 and 1A according to the embodiment of the present disclosure. The above-mentioned sentence creating devices 1, 1A may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, and the like. ..

In the following description, the word "device" can be read as a circuit, a device, a unit, or the like. The hardware configuration of the writing devices 1 and 1A may be configured to include one or more of the devices shown in the figure, or may be configured not to include some of the devices.

For each function in the text creation devices 1 and 1A, the processor 1001 performs an operation by loading predetermined software (program) on the hardware such as the processor 1001 and the memory 1002, and controls the communication by the communication device 1004. It is realized by controlling at least one of reading and writing of data in the memory 1002 and the storage 1003.

Processor 1001 operates, for example, an operating system to control the entire computer. The processor 1001 may be configured by a central processing unit (CPU) including an interface with a peripheral device, a control device, an arithmetic unit, a register, and the like. For example, each function in the above-mentioned sentence creating device 1 may be realized by the processor 1001.

The processor 1001 reads a program (program code), a software module, data, and the like from at least one of the storage 1003 and the communication device 1004 into the memory 1002, and executes various processes according to these. As the program, a program that causes a computer to execute at least a part of the operations described in the above-described embodiment is used. For example, each function in the sentence creating devices 1 and 1A may be realized by a control program stored in the memory 1002 and operating in the processor 1001. Although it has been described that the various processes described above are executed by one processor 1001, they may be executed simultaneously or sequentially by two or more processors 1001. Processor 1001 may be mounted by one or more chips. The program may be transmitted from the network via a telecommunication line.

The memory 1002 is a computer-readable recording medium, for example, by at least one of ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), RAM (Random Access Memory), and the like. It may be configured. The memory 1002 may be referred to as a register, a cache, a main memory (main storage device), or the like. The memory 1002 can store a program (program code), a software module, or the like that can be executed to perform information processing according to the embodiment of the present disclosure.

The storage 1003 is a computer-readable recording medium, for example, an optical disk such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, an optical magnetic disk (for example, a compact disk, a digital versatile disk, a Blu-ray). It may consist of at least one such as a (registered trademark) disk), a smart card, a flash memory (eg, a card, stick, key drive), a floppy (registered trademark) disk, and a magnetic strip. The storage 1003 may be referred to as an auxiliary storage device. The first translation corpus 82 and the second translation corpus 84 may be, for example, a database, server, or other suitable medium containing at least one of the memory 1002 and the storage 1003.

The communication device 1004 is hardware (transmission / reception device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as, for example, a network device, a network controller, a network card, or a communication module. For example, the above-mentioned acquisition unit 10 and the like may be realized by the communication device 1004.

The input device 1005 is an input device (for example, a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.) that receives an input from the outside. The output device 1006 is an output device (for example, a display, a speaker, an LED lamp, etc.) that performs output to the outside. The input device 1005 and the output device 1006 may have an integrated configuration (for example, a touch panel).

Each device such as the processor 1001 and the memory 1002 is connected by a bus 1007 for communicating information. The bus 1007 may be configured by using a single bus, or may be configured by using a different bus for each device.

The writing devices 1 and 1A include hardware such as a microprocessor, a digital signal processor (DSP), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), and an FPGA (Field Programmable Gate Array). It may be configured by, and a part or all of each functional block may be realized by the hardware. For example, processor 1001 may be implemented using at least one of these hardware.

The processing procedures, sequences, flowcharts, and the like of the embodiments / embodiments described in the present disclosure may be rearranged in order as long as there is no contradiction. For example, the methods described in the present disclosure present elements of various steps using exemplary order, and are not limited to the particular order presented.

The input / output information and the like may be stored in a specific place (for example, a memory), or may be managed using a management table. Information to be input / output may be overwritten, updated, or added. The output information and the like may be deleted. The input information or the like may be transmitted to another device.

The determination may be made by a value represented by 1 bit (0 or 1), by a true / false value (Boolean: true or false), or by comparing numerical values (for example, a predetermined value). It may be done by comparison with the value).

The embodiments / embodiments described in the present disclosure may be used alone, in combination, or switched with each other in practice. Further, the notification of predetermined information (for example, the notification of "being X") is not limited to the explicit notification, but is implicitly (for example, by not notifying the predetermined information). May be good.

Although the present disclosure has been described in detail above, it is clear to those skilled in the art that the present disclosure is not limited to the embodiments described in the present disclosure. The present disclosure may be implemented as amendments and modifications without departing from the spirit and scope of the present disclosure as defined by the claims. Therefore, the description of the present disclosure is for purposes of illustration and has no limiting meaning to the present disclosure.

Software, whether called software, firmware, middleware, microcode, hardware description language, or other names, is an instruction, instruction set, code, code segment, program code, program, subprogram, software module. , Applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, features, etc. should be broadly interpreted.

Software, instructions, information and the like may be transmitted and received via a transmission medium. For example, the software may use at least one of wired technology (coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), etc.) and wireless technology (infrared, microwave, etc.) to create a website. When transmitted from a server or other remote source, at least one of these wired and wireless technologies is included within the definition of transmission medium.

The information, signals, etc. described in this disclosure may be represented using any of a variety of different techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that may be referred to throughout the above description are voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, light fields or photons, or theirs. It may be represented by any combination.

The terms described in the present disclosure and the terms necessary for understanding the present disclosure may be replaced with terms having the same or similar meanings.

The terms "system" and "network" used in this disclosure are used interchangeably.

The information, parameters, etc. described in the present disclosure may be expressed using absolute values, relative values from a predetermined value, or other corresponding information. It may be represented.

The terms "determining" and "determining" as used in this disclosure may include a wide variety of actions. "Judgment" and "decision" are, for example, judgment (judging), calculation (calculating), calculation (computing), processing (processing), derivation (deriving), investigation (investigating), search (looking up, search, inquiry). It may include (eg, searching in a table, database or another data structure), ascertaining as "judgment" or "decision". "Judgment" and "decision" are receiving (for example, receiving information), transmitting (for example, transmitting information), input (input), output (output), and accessing (accessing). ) (For example, accessing the data in the memory) may be regarded as "judgment" or "decision". "Judgment" and "decision" may include resolving, selecting, choosing, establishing, comparing, etc. as being regarded as "judgment" and "decision". .. That is, "judgment" and "decision" may include considering some action as "judgment" and "decision". "Judgment (decision)" may be read as "assuming", "expecting", "considering", or the like.

The terms "connected", "coupled", or any variation thereof, mean any direct or indirect connection or connection between two or more elements and each other. It can include the presence of one or more intermediate elements between two "connected" or "combined" elements. The combination or connection between the elements may be physically performed, logically performed, or a combination thereof. For example, "connection" may be read as "access". As used in the present disclosure, the two elements use at least one of one or more wires, cables and printed electrical connections, and, as some non-limiting and non-comprehensive examples, radio frequencies. It can be considered to be "connected" or "coupled" to each other using electromagnetic energy having wavelengths in the region, microwave region and light (both visible and invisible) regions.

The phrase "based on" as used in this disclosure does not mean "based on" unless otherwise stated. In other words, the statement "based on" means both "based only" and "at least based on".

Any reference to elements using designations such as "first" and "second" as used in the present disclosure does not generally limit the quantity or order of those elements. These designations can be used in the present disclosure as a convenient way to distinguish between two or more elements. Therefore, references to the first and second elements do not mean that only two elements can be adopted, or that the first element must somehow precede the second element.

The "part" in the configuration of each of the above devices may be replaced with a "circuit", a "device" or the like.

When "include", "including" and variations thereof are used in the present disclosure, these terms are as inclusive as the term "comprising". Is intended. Moreover, the term "or" used in the present disclosure is intended not to be an exclusive OR.

In the present disclosure, if articles are added by translation, for example, "a", "an", and "the" in English, the present disclosure is a plural form of the noun following these articles. May include that.

In the present disclosure, the term "A and B are different" may mean "A and B are different from each other". The term may mean that "A and B are different from C". Terms such as "separate" and "combined" may be interpreted in the same way as "different".

1,1A ... Sentence writing device, 2a, 2b, 2c, 2d, 2e, 2f, 2g, 2h, 2i, 3a, 3b, 3c, 3d, 3e, 3f, 3g, 3h, 3i, 3j, 3k ... Morphological elements, 2p ... Index, 5a, 5b, 5c, 5d ... Phrase, 8a, 8b, 8c, 8d ... Feature quantity, 10 ... Acquisition unit, 20 ... Analysis unit, 30 ... Detection unit, 40 ... Correspondence unit, 50 ... Derivation unit , 60 ... Creation unit, 70 ... Weighting unit, 82 ... First translation corpus, 84 ... Second translation corpus, 1001 ... Processor, 1002 ... Memory, 1003 ... Storage, 1004 ... Communication device, 1005 ... Input device, 1006 ... Output Device, 1007 ... Bus, T11 ... 1st sentence, T21 ... 2nd sentence, T31 ... 3rd sentence.

Claims (7)

An acquisition unit that acquires the first sentence of the first language and the second sentence of the second language, which is a translation of the first sentence, and
An analysis unit that morphologically analyzes each of the first sentence and the second sentence,
A detection unit that detects all clauses having morphemes in the second sentence, and
A matching unit that associates the morpheme of the first sentence with the morpheme of the second sentence based on the meaning of each morpheme.
A derivation unit for deriving a feature amount in each clause of the second sentence based on an index indicating the position of the morpheme of the first sentence in the first sentence associated with the morpheme of the second sentence.
A creation unit that creates a third sentence by arranging the phrases of the second sentence based on the feature amount of each phrase of the second sentence.
A writing device equipped with. The creating unit arranges each clause of the second sentence so that the feature amount of each clause of the second sentence corresponds to the order of the index of the morpheme arranged from the beginning of the sentence to the end of the sentence. 3. The sentence creation device according to claim 1, which creates sentences. The derivation unit according to claim 2, wherein the minimum value among the indexes of the morphemes of the first sentence associated with the morphemes in the phrase of the second sentence is the feature amount of the phrase of the second sentence. Writing device. The derivation unit according to claim 2, wherein the maximum value of the index of the morpheme of the first sentence associated with the morpheme in the phrase of the second sentence is the feature amount of the phrase of the second sentence. Writing device. Further, a weighting unit for setting a weight for the morpheme of the first sentence is provided.
The derivation unit sets the index of the morpheme of the first sentence having the highest weight among the morphemes of the first sentence associated with the morpheme in each clause of the second sentence in each clause of the second sentence. The text creation device according to claim 2, which is the feature amount. Further, a weighting unit for setting a weight for a morpheme in the phrase of the second sentence is provided.
The derivation unit sets the index of the morpheme of the first sentence associated with the morpheme having the highest weight among the morphemes in each clause of the second sentence as the feature of each clause of the second sentence. The text-creating device according to claim 2, which is a quantity. When the feature quantities in the plurality of clauses of the second sentence have the same value, the creating unit arranges the plurality of clauses so as to maintain the order of the clauses arranged from the beginning to the end of the second sentence. The sentence creating device according to any one of claims 1 to 6, wherein the third sentence is created.

Images