JPH05224687A - Japanese sentence reading word conversion edit processing method - Google Patents

Abstract

PURPOSE:To reduce the throughput of document editing by an editor by previously setting information flags which indicate homonym and polysemy in word information in a Japanese word dictionary and providing a conversion word dictionary which contains substitute conversion words. CONSTITUTION:The system is equipped with the conversion word dictionary 90 which contains the conversion word for homonym substitution for each combination of a polysemy index and a conversion word for polysemy substitution, and a homonym index and the meaning attribute of a case element word. A morphemic analysis process is performed for a Japanese document to be pronounced and a polysemy extracting process part 50 extracts polysemy from a certified word string. Further, a polysemy deciding process part 60 narrows down the polysemy according to the co-occurrence relation between a document decomposition attribute attached to a document file and the meaning attribute of the polysemy when there are plural word candidates meeting grammatical connection requirements to decide the word candidate. Further, a word converting and editing process part 100 retrieves the conversion word dictionary 90 to extract a conversion word and substitute the character string in a source document with it, and writes the character string in a document file 110.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is a Japanese sentence reading word conversion / editing process for converting and editing a Japanese sentence sentence word in an easy-to-understand manner when reading a Japanese sentence breaking news or various information guides and delivering them by communication means. It is about the method.

[0002]

2. Description of the Related Art There is a rapid increase in services in which information data that requires immediacy such as breaking news in Japanese and various information guides are transmitted by communication means and distributed to users through media such as telephones and facsimiles. In particular, when a Japanese sentence is read aloud and delivered by voice data such as a telephone, the reading process is transient, so that it is important to make the read contents easy to hear. However, in Japanese, there are homophones and isomorphic words that have polysemy, and if they are mixed, the intelligibility such as erroneous listening and difficulty in hearing occurs.

On the other hand, in order to increase the intelligibility of information, the user repeatedly performs operations such as reading aloud, adding special reading for word reading distinction and word replacement at the information distribution source, and the word at the information distribution source. It is possible to take actions such as editing the expression conversion.

[0004]

However, there is a problem that it is essentially impossible to repeat reading when the number of operations by the user is increased or when voice data is sent to a plurality of users through a speaker or the like. ..

In addition, as in the example of the grammar proofreading process, special reading is performed so that the original characters can be easily imaged. For example, "pursuit""pursuit""pursuit"
It will be "Tsuikyu Oyobi", "Tsuikyu Kiwame", and "Tsuikyu Motome". However, it is difficult for general information users to hear the special reading, and it becomes difficult to understand because the naturalness is impaired.
In addition, as a means to replace the appearing words with words that are easy to understand, in the case of "Japan-China talks about economic problems between Japan and China", if "Japan-China" → "Japan and China" Will discuss economic issues with Japan and China. " Furthermore, in the case of “taking legal means to take contaminated soil”, “take” → “collect” means “to take legal means to take contaminated soil”, and both are inappropriate expressions. There is a problem that they will be replaced.

[0006] In addition, since it is necessary to convert and edit a large number of homophones and polysemous words within a necessary range at the stage of editing a Japanese sentence sentence, it is a work requiring skill.
In this way, it is possible for the editor to perform editing within a necessary range, such as extracting homophones and polysemous words that are difficult to hear in the reading process of Japanese sentence sentences and converting them into easy-to-understand word expressions with few listening errors. Therefore, it is difficult to process in a short time.

In order to solve the above-mentioned problems in the conventional means, the present invention searches a word conversion dictionary for a homophone or a polysemous word that matches a condition for converting a Japanese sentence sentence from the result of morphological analysis processing. It is intended to be converted into a converted word and edited.

[0008]

In order to achieve the above object, according to the present invention, in a process of converting a homonym or a polysemous word included in a Japanese sentence sentence file for reading into a word that is easy to hear and editing, A method for morphological analysis of original Japanese sentences using a Japanese word dictionary and a grammar dictionary to obtain word recognition and word linguistic information, and to add polysemous words to the word information of the Japanese word dictionary in advance. When a plurality of word candidates (semantic polysemy) satisfying a grammatical connection condition with a polysemous word are present, a means for setting an information flag shown, a means for extracting a polysemous word by the information flag, and the word file are attached to the text file. A means for narrowing down polysemy by co-occurrence (field dependency) of the text field attribute and the meaning attribute of polysemy (meaning of a word is shown in general terms), and polysemous word for each polysemous word in advance Word headlines The present invention is characterized by having a conversion word dictionary containing a conversion word to be replaced for a part of speech and a means for searching the conversion word dictionary to convert and edit a polysemous word, and also to previously prepare word information of the Japanese word dictionary. A means for setting an information flag indicating a homonym of a verb, a means for extracting the homonym with the information flag, and a semantic attribute of a noun (hereinafter referred to as a case element word) having a case structure relationship existing in front of the homonym. And a conversion word dictionary containing conversion words to be replaced in advance for each combination of the homophone heading and the meaning attribute of the case element word, and a means for searching the conversion word dictionary and converting and editing the homophone. It is characterized by including.

[0009]

In the present invention, in a process of converting a homonym or a polysemous word included in a Japanese sentence sentence file for reading into a word that is easy to hear and editing it, the homonym or polysemous word is shown in the word information of the Japanese word dictionary. Preset information flag,
It is equipped with a conversion word dictionary that contains the conversion words to be replaced, which enables the processing of extracting homonyms and polysemous words that should be noted and the processing of replacing easily recognizable words by the semantic attributes of case element words in the front. This enables the editor to reduce the amount of text editing processing.

[0010]

Embodiments of the present invention will now be described in detail with reference to the drawings. 1 to 8 are diagrams showing an embodiment of the present invention. 4 to 6 are views showing an embodiment of claim 1 of the present invention in the block configuration of FIG. 1, and FIGS. 7 to 8 are claims 2 of the present invention in the block configuration of FIG.
It is a figure which shows the Example of.

FIG. 1 is a processing block configuration example, FIG. 2 is a Japanese word dictionary configuration example, FIG. 3 is a conversion word dictionary configuration example, and FIG.
7 and FIG. 7, and FIG. 5, FIG. 6, and FIG. 8 are diagrams for explaining an example of the word conversion editing process.

In FIG. 1, a processing device 120 is a processing device including a CPU and a memory and has the following functional units.
That is, the Japanese sentence sentence of the Japanese sentence original sentence file 10 for reading out is extracted from the morphological analysis processing unit 40 that authenticates the word and acquires the language information of the word using the Japanese word dictionary 20 and the grammar dictionary 30; Of a plurality of information flags, a polysemous word extraction processing unit 50 for extracting a word for which the “isomorphic polysemous flag” is turned on; a polysemous word determination process for narrowing down polysemous words when there is a word candidate (semantic polysemous) that satisfies a grammatical connection condition Part 60; Homophone extraction processing unit 70 that extracts words for which the “homophone flag” is ON from the information flags extracted from the word information;
Case element semantic attribute extraction processing unit 80 for extracting the semantic attribute of the case element word existing in front of the homophone; the polysemous word headline and the conversion word for polysemous word replacement, and the homonym word headline and the semantic attribute of the case element word. A conversion word dictionary 90 in which conversion words for homophone substitution are previously recorded for each combination of; a word conversion edit processing unit 100 for searching the conversion word dictionary to convert and edit a polysemous word or a homophone; edited text file 110 Consists of.

In this processing device 120, morphological analysis processing is performed from the beginning for an arbitrary sentence of a Japanese original sentence for reading, and the part of speech, reading, homophone flag, isomorphic flag, meaning as word recognition and word language information. Authenticate with attributes. Next, a polysemous word (a word for which the “isomorphic polysemous flag” is ON) is extracted from the recognized word string (polysemous word extraction processing unit 50). When there are multiple word candidates (semantic ambiguity) that satisfy the grammatical connection condition, the word sense is narrowed down by the co-occurrence (field dependency) between the sentence field attribute attached to the text file and the meaning attribute of the polysemous word. Is determined (the polysemous word determination processing unit 60). Furthermore, the converted word dictionary 90 is searched with the word headwords of the polysemous words and the certified part of speech, the converted words are extracted, replaced with the character strings of the original sentence file, and written in the edited sentence file 110 (the word conversion edit processing unit 10).
0).

Further, a homonym of which the part of speech is a verb (a word whose "homophone flag" is ON) is extracted from the recognized word string (homophone extraction processing section 70), and the case element existing in front of the homophone is extracted. The semantic attribute of the word is extracted (case element semantic attribute extraction processing unit 80).

The converted word dictionary 90 containing the converted words to be replaced in advance for each combination of the word heading and the meaning attribute of the case element word is searched and matched with the homologous word heading and the extracted meaning attribute of the case element. The conversion word of the pattern is extracted, replaced with the character string of the original text file, and written in the edited text file 110 (word conversion editing processing unit 10
0).

FIG. 2 is a diagram showing a configuration example of the Japanese word dictionary 20 which is a component of the block diagram of FIG. In “Chinese”, there are word candidates having two ambiguous meanings such as a toki and a proper noun, so the isomorphic ambiguous flag is given. In addition, as a semantic attribute (indicating the meaning of a word in general terms) to determine the meaning of a word candidate, "time"
There is a "country, region". Also, the verb homophone “Take”
The same-sound flag is attached to each of “take”, “capture”, and “shoot”.

FIG. 3 is a diagram showing an example of the structure of the converted word dictionary 90 which is a component of the block diagram of FIG. here,
130 is the semantic attribute of the case element word preceding the homophone, 140
Is a conversion word. Since there are two ambiguous word candidates for "Japanese and Chinese", that is, a tongue and a proper noun, each converted word is "daytime" when it is a tongue, and "Japan and China" when it is a proper noun. ] And is stored. Further, in the case of a homonym of a verb, the converted word differs depending on the semantic attribute of the case element word. For example, in the case of "collect", when the semantic attribute of the case element word is "system", it is "adopted", and when it is "biological", it is "collected".

FIG. 4 is a diagram showing a schematic flow of a process of converting a polysemous word included in the Japanese sentence original sentence file 10 in the processing block configuration example shown in FIG. 1. The operation will be described according to the schematic flow. ..

The processing target sentence to which the word conversion editing process is applied is read from the Japanese sentence original sentence file 10 (step 100). In the morphological analysis processing unit 40 for all the read sentences, the Japanese word dictionary 20 and the grammar dictionary 30 are read. Using the morphological analysis process, the word is identified and the part of speech, reading, homophone flag, homomorphic flag, and semantic attribute are identified as the language information of the word (step 110). , A polysemous word is extracted from the recognized word string by an information flag indicating a polysemous word, and if it is not a polysemous word, the process branches to step 180 (step 120). It is determined whether there are a plurality of word candidates (semantic ambiguity) to be satisfied, and if there is no semantic ambiguity, the process branches to step 160 (step 1 0) In the polysemous word determination processing unit 60, in the case of a polysemous word having semantic polysemy (when the determination in step 130 is YES), the co-occurrence of the sentence field attribute and the meaning attribute of the polysemous word (field dependency degree) ) Is confirmed (step 140) The word candidate with the higher co-occurrence (field dependency) is recognized (step 150) The converted word dictionary 90 is searched with the recognized polysemous word heading and certified part of speech to convert the word Is extracted and the character string in the original sentence file is replaced (step 160) It is written in the edited sentence file 110 (step 170) It is determined whether or not the processing has been completed for all words, and if the determination is YES, the processing is terminated. .. If the determination is NO, the next word is read (step 190) and the process proceeds to step 120 to continue the process (step 180). FIG. 5 shows the polysemous determination process in the schematic flow of the process for converting polysemous words in FIG. This is an example.

Here, 150 is a text field attribute, 160 is an original text character string, 170 is a polysemous word, 180 is a grammatical word connection condition, 190 is a field dependency determination condition, and 195 is a certified part of speech after polysemous judgment. ..

In the extracted polysemic word "Japanese-Chinese", it is determined whether or not there are a plurality of word candidates (semantic polysemous) that satisfy the grammatical connection condition. In "Japanese-Chinese opened", "Japanese (Chinese)" is defined as a grammatical connection condition with the verb "Opened" of the following word.
There is no semantic ambiguity because only is certified. However, in "Japan-China Suspicion", multiple word candidates "Chinese (Hijiki)" and "Chinese (Proper noun)" are identified based on the grammatical connection conditions with the suffix "Ma" of the subsequent words. As a result, semantic ambiguity occurs. If the semantic polysemy remains, it may hinder the conversion word search process. Therefore, the coexistence with the semantic attribute of the polysemous word (field dependency) is checked using the sentence field attribute 150 to narrow down the polysemy. Here, since the text field attribute is "international politics", "Japanese and Chinese (proper noun)" whose semantic attribute is "country, region" is recognized (part of speech 195 after recognition).

FIG. 6 is an embodiment of the schematic flow of the processing for converting the polysemous words in FIG. Here, 200 is a conversion word, 2
10 is a character string after conversion. For a polysemous word with a narrowed meaning of meaning, the converted word dictionary 90 is searched with the recognized word headings and parts of speech to extract the converted word 200, and the converted character string 210 is replaced with the character string of the original text file. To create.
In the embodiment, "daytime" is converted to "daytime (timer)", and "Japan and China" is converted to "daytime (proper noun)".

As described above, a conversion word dictionary is prepared in advance for each polysemous word of the polysemous words, which contains the conversion words for replacing the word headings of the polysemous words and the parts of speech. Since the converted word dictionary is searched with the word headings and the parts of speech of the polysemous words that have been identified by narrowing down the meaning of the polysemous word, conversion into a conversion word that is easy to hear is performed, so that the editor can reduce the amount of text editing processing.

FIG. 7 is a diagram showing a schematic flow of a process for converting a homonym of a verb included in the Japanese sentence original sentence file 10 in the processing block configuration example shown in FIG. I do.

Here, step 200 and step 21
0, step 280, and step 290 are respectively shown in FIG.
Step 100, Step 110, Step 180,
Equivalent to step 190.

In the homophone extraction processing unit 70, a homophone is extracted from the recognized word string by an information flag indicating a homophone of a verb, and if it is not a homophone, the process branches to step 280 (step 220). In the case element semantic attribute extraction processing unit 80, a word preceding the homophone is searched for in the extracted homophone, the case element word is extracted, and the semantic attribute of the case element word is extracted (step 230). The converted word dictionary 90 is searched using the headline and the extracted semantic attribute of the case element word (step 240). It is determined whether there is a matching pattern in the converted word dictionary. If there is no matching pattern, the process branches to step 280 (step 250). The converted word is extracted and the character string of the original sentence file is replaced (step 260). The edited sentence file 110 is written (step 270). 9 is an example of a schematic flow of a process of converting the same phoneme in FIG. 7.

Here, 220 is an original sentence character string, 230 is a homophone (verb), 240 is a case structure relationship with a case element word,
Reference numeral 250 is a converted word, and 260 is a character string after conversion. With respect to the extracted homophones “capture” and “take”, a word in front of the homonym is searched to extract a case element word, and a semantic attribute of the case element word is extracted. In the embodiment, “salmon (biological)” is extracted for “catching”, and “pathogen (biological)” and “means (system)” are extracted for “collecting”. Next, the converted word dictionary 90 is searched by the homonym word heading having the case element word and the meaning attribute of the extracted case element word, and it is determined whether there is a matching pattern in the converted word dictionary. If there is a matching pattern, the converted word 250 is extracted and replaced with the character string of the original text file to create the converted character string 260. In the embodiment, since "salmon (biological)" has a case structure relationship with "capture", "capture" is converted to "capture". In addition, for "take", the case element word is converted from "take" to "collect" for "pathogen (biological)", and the case element word is "for means (system)". "Take" is converted to "Take".

In this way, a conversion word dictionary containing conversion words to be replaced for each combination of the homophone heading and the meaning attribute of the case element word is prepared in advance, and when the homophone is extracted, the case element is extracted. The converted word dictionary is searched using the word heading of the homophone having the word and the meaning attribute of the extracted case element word, and converted into a converted word that is easy to hear.

That is, in the Japanese sentence reading word conversion / editing processing method of the present invention, in the process of converting and editing the homonyms and polysemous words contained in the Japanese sentence sentence file for reading into easily audible words, the means already described is used. ,
It is possible to perform processing for extracting homonyms and polysemous words that should be noted, and processing for replacement with a conversion word that is easy to hear based on the semantic attribute of the preceding case element word.

[0030]

As described above, according to the present invention,
In particular, when reading out Japanese sentences and delivering breaking news articles using voice data such as telephones, in order to eliminate misunderstanding and difficulty of listening to the read contents and create Japanese sentences with high intelligibility, Generally, when editing Japanese sentences, it is possible to reduce the editor's processing amount of sentence editing.

[Brief description of drawings]

FIG. 1 is an example of a processing block configuration of the present invention.

FIG. 2 is a configuration example of a Japanese word dictionary.

FIG. 3 is a configuration example of a conversion word dictionary.

FIG. 4 is a schematic process flow.

FIG. 5 is a diagram illustrating an example of a word conversion editing process.

FIG. 6 is a diagram illustrating an example of a word conversion editing process.

FIG. 7 is a schematic process flow.

FIG. 8 is a diagram illustrating an example of a word conversion editing process.

[Explanation of symbols]

 10 Japanese sentence original sentence file 20 Japanese word dictionary 30 Grammar dictionary 40 Morphological analysis processing unit 50 Polysemous word extraction processing unit 60 Polysemous word determination processing unit 70 Homophone extraction processing unit 80 Case element meaning attribute extraction processing unit 90 Converted word dictionary 100 Word conversion edit processing unit 110 Edited sentence file 120 Processing device 130 Case element Semantic attribute of word 140 Converted word 150 Text field attribute 160 Original sentence character string 170 Polysemous word 180 Sentence word connection condition 190 Field dependency judgment condition 195 Certification Part-of-speech 200 Converted word 210 Converted character string 220 Original sentence character string 230 Homophone 240 Case structure relation 250 Converted word 260 Converted character string

Claims (2)

[Claims] 1. In a process of converting a homonym or a polysemous word included in a Japanese sentence text file for reading into a word that is easy to hear and editing the original Japanese sentence for reading, a Japanese word dictionary and a grammar dictionary are used. Morphological analysis processing to obtain word recognition and word linguistic information, means for setting an information flag indicating a polysemous word in the word information of the Japanese word dictionary in advance, and polysemous word extraction by the information flag And a means for narrowing down the polysemous word by the co-occurrence of the sentence field attribute attached to the text file and the semantic attribute of the polysemous word when there are multiple word candidates that satisfy the grammatical connection condition for the polysemous word, A conversion word dictionary in which a polysemous word heading and a conversion word to be replaced with respect to a part of speech are recorded for each polysemy, and means for converting and editing a polysemous word by searching the conversion word dictionary Japanese sentence reading word conversion editing processing method characterized by. 2. A means for setting an information flag indicating a homonym of a verb in the word information of a Japanese word dictionary in advance, a means for extracting a homonym by the information flag, and a case structure relationship existing in front of the homonym. Means for extracting the semantic attributes of the nouns, a conversion word dictionary containing the conversion words to be replaced in advance for each combination of the homonym headings and the meaning attributes of the nouns having a case structure relationship, and the same word by searching the conversion word dictionary. 2. A Japanese sentence reading word conversion / editing method according to claim 1, further comprising means for converting and editing words.