JPS6162175A - Analyzer for japanese language - Google Patents

Abstract

PURPOSE:To analyze morphological elements correctly without processing the morphological elements repeatedly by retaining the results of morphological analysis of Japanese languages in the Japanese syllabary/Chinese characters converter. CONSTITUTION:When a series of characters are sent to a controller 12 by manipulating the keys on a keyboard 10, the data for said series of characters are transmitted from the controller 12 through a discriminator 14 to a morphological element analyzer 16 and the discriminator 14 stores the manipulation of conversion keys. The analyzer 16 analyzes the morphological elements with reference to the dictionary in a word dictionary ROM 18 and performs Japanese syllabary/Chinese character conversion. The data showing the dictionary addresses in ROM 18 registering a string of characters divided into words as the morphological analysis of converted Japanese syllabary/Chinese character and each word for the string of words are sent from the analyzer 16 to the discriminator 14 and to a display device 22. On the other hand, morphologically analyzed data of string of words are divided into words at an input letter buffer memory 20 and stores the string of the converted Japanese syllabary/Chinese character and the dictionary addressed of string of words.

Description

DETAILED DESCRIPTION OF THE INVENTION TECHNICAL FIELD The present invention relates to a Japanese language analysis device, and more specifically, to a single-language processing device that performs morphological analysis of Japanese sentences that can be used for syntactic analysis such as machine translation.

In conventional machine translation, etc., morphological analysis of a Japanese sentence is generally performed prior to syntactic analysis of the Japanese sentence. In conventional translation systems, morphological analysis of Japanese sentences requires, for example,
This method had the following drawbacks because it was applied to nine solid kanji/kana mixed sentences that had been input in some way.

In other words, since the input Japanese text is a solid text of kanji, kana, etc., the algorithm for morphologically analyzing it is extremely complex and large. Furthermore, since the sentences containing kanji and kana are written in solid letters, ambiguity arises in word segmentation, resulting in multiple analysis results and the morphological analysis process has to be performed multiple times.

Purpose The present invention aims to eliminate the drawbacks of the prior art and provide a Japanese language analysis device that can accurately perform morphological analysis without performing morphological analysis processing multiple times. .

Configuration The configuration of the present invention will be explained below based on the implementation 'I=lJ. In the present invention, by retaining the results of the morphological analysis of Japanese sentences performed in the kana-kanji conversion process, this can be used in the next stage of syntactic analysis such as machine translation.

FIG. 1 shows the configuration of an embodiment of the Japanese language analysis device according to the present invention, and in the same figure, keys as input means are 1 and 1o are numeric keys indicating numbers, characters such as hiragana, etc. It consists of character keys, conversion keys for instructing whether or not to execute kana-kanji conversion processing, function keys including non-conversion keys, etc. The user operates the keys on the key gate 10. Syntax analysis 1 - Japanese sentence data and kana-kanji conversion A command for processing is input to the control device 12.

The control device 12 sends the Japanese sentence data and commands input from the keyboard IO to the determination device 14, and receives the morphological analysis results of the Japanese sentence data by the morphological analysis device 16 via the determination device 14, and displays them on the display device 22. Forward.

Further, the determination device 14 receives Japanese sentence data inputted from the control device 12 and the key gate 10, transfers the Japanese sentence data to the morphological analysis device 16, and sends the Japanese sentence data to the key gate 10.
It stores whether or not the conversion key and non-conversion key have been operated, and also transfers the morphological analysis result of the Japanese sentence data by the morphological analysis device 16 to the input sentence panzer memory 20.

Furthermore, the word dictionary ROM l8 contains characters indicating words.
Information such as the meaning is classified and stored in correspondence with the data in column '. , are stored in the word dictionary ROM 1g. Performs kana-kanji conversion processing for Japanese sentence data by referring to dictionary information. Here, if the determination device 14 remembers that the conversion key in the key gate number was operated, the kana-kanji conversion process including morphological analysis is executed, but it is not remembered that the non-conversion key was operated. If so, only morphological analysis is performed. Note that in this specification, "kana-kanji conversion processing" includes both morphological analysis and kana-kanji conversion processing.

The input sentence buffer memory 2o uses the morphological analysis results of the Japanese sentence data input to the key gate 1° to the judgment device 14.
Uke is memorized and accumulated through .

Next, the specific operation of the Japanese language analysis device having the above configuration will be explained below. First, is the user the key number? It is assumed that, for example, the character string "Anata wa" is input into the control device 12 by operating the keys on the - code 10, and the non-conversion key is operated.

As a result, the character string data is transferred from the control device 12 to the determination device 1.
4 to the morphological analysis device 1G, and the determination device 14 stores that the no-conversion key has been operated.

On the other hand, the morphological analysis device 16 performs a morphological analysis process on the character string data while referring to the dictionary information stored in the word dictionary W ROM l 8, and determines whether or not the kana-kanji conversion stored in the determination device 14 has been performed. In this case, the determination device 14 receives data indicating that the no-conversion key is being operated, that is, data indicating that kana-kanji conversion is not to be performed.
Therefore, only the morphological analysis results are output to the determination device 14. Here, the morphological analysis result refers to the character string divided into words and the word dictionary ROM 1 of each divided word.
This means data indicating the address registered in 8.

The determination device 14 transfers the morphological analysis result received from the morphological analysis device 16 to the input sentence panzer memory 20, and the input sentence buffer memory 20 contains word-divided character strings and the word-divided character strings as shown in FIG. Correspondingly, the dictionary addresses of the divided words registered in the word dictionary ROM 18 are stored.

Next, it is assumed that a second user inputs the character string "you" into the control device 12 by operating keys on the keyboard 10, and then operates a conversion key.

As a result, the character string data is transferred from the control device 12 to the determination device 1.
4 to the morphological analysis device 16, and the determination device 1
4 stores that the conversion key has been operated.

On the other hand, the morphological analysis device 16 performs a morphological analysis process on the character string data while referring to the dictionary information stored in the word dictionary ROM l 8, and also determines whether or not kana-kanji conversion stored in the determination device 14 has been performed. The kana-kanji conversion is performed by referring to the data indicating that the conversion key is being operated (in this case, data indicating that the conversion key is being operated, that is, data indicating that kana-kanji conversion is to be performed). Then, it shows the morphological analysis result after kana-kanji conversion, that is, a character string divided into words such as "For you" and the dictionary address in the word dictionary ROM 18 where each word of this character string is registered. The data is sent from the morphological analysis device 16 to the shading device 14.

The determination device 14 inputs the morphologically analyzed character string data to the display device 22 via the control device 12, and when the morphologically analyzed character string data is confirmed by the user (the user inputs the character string data to the keyboard 10). When the next character string data is input by key operation), the input sentence buffer memory 20 stores a character string divided into words and converted into kana-kanji as shown in FIG. 2(B). Data indicating the dictionary address of this character string is transferred, and these data are stored and accumulated in the input sentence buffer memory 20.

As explained above, according to this embodiment, shape B chord analysis can be performed accurately without performing morphological analysis multiple times as in the conventional case. -Next, the configuration of another embodiment of the Japanese language analysis device according to the present invention is shown in FIG. This embodiment differs in structure from the embodiment shown in FIG. This is due to the addition of a character string combining device 30 that processes the words divided into words to form one word, and the other configurations are basically unchanged.
Functional blocks that are the same as those in the embodiment shown in FIG.

First, let's say that the user operates the keys on a keyboard.
input the character string data "to you" into the control device 12,
Next, it is assumed that the non-conversion key is operated. In this case, the morphological analysis process is performed in the same way as the implementation sequence shown in FIG.
Word dictionary ROM 18 for the character string “you” and this character string
The data of the dictionary address in is stored in the input sentence buffer memory IJ 20. Then, the character string "For you" is displayed on the display device 22. However, when this process is executed, the data of the auxiliary word "ni" of "you" is stored in the internal stack of the character string combination device 30.

Then the key from the user? When the character string "kansuru" is input by key operation of the -do 1G and the conversion key is operated, the character string "kansuru" is transferred to the character string combination device 30 via the control device 12 and the determination device 14. String joiner 3
At 0, the character string data "ni" and "kansuru" are combined to form the character string "nikansuru". This character string "nikansuru" is passed to the character string combiner 30 and the morphological analyzer 16
The text is sent to , where morphological analysis and kana-kanji conversion are performed. Here, it is assumed that the character string "regarding" is registered in the word dictionary ROM 18 as a case particle equivalent word.

Therefore, as a result of the kana-kanji conversion process, the morphological analysis device 1
Data indicating the dictionary address of the 6-yoshi character string "regarding" and the case particle equivalent word is sent to the character string combination device 30.

The character string combining device 30 searches for the character string ``Kitsu'' using the data indicating the character ``ni'' in the internal stack as an index, and divides the string into words such as ``nimu kinki''(``mu'' is a punctuation mark). ), and the data indicating this character string "related to" is sent from the character string combination device 30 to the determination device 14, which determines the character string "related to"
The data of the part "related" is extracted from the data, and the data indicating this character string "related" is sent to the display device 22 via the control device 12.

Next, as in the case of the embodiment shown in FIG. Based on 12 control commands, the character string “
Input the data indicating "related to" in the sentence buffer memo IJ.
20, as shown in FIG. 4, the character string "regarding" is stored in correspondence with the adjunct part of the previously input character string "you", and the storage area for this character string "regarding" is stored. Store the dictionary address of the character string ``related'' in the storage area.

As explained above, according to this embodiment, in addition to the effects of the embodiment shown in FIG. The effect is that morphological analysis processing can be performed even for words that do not exist.

Effects According to the present invention, morphological analysis can be performed accurately without performing morphological analysis processing multiple times.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the Japanese language analysis device according to the present invention, FIG. 2 is an explanatory diagram showing an example of the stored contents of an input sentence panzer memory, and FIG. 3 is a block diagram showing the configuration of an embodiment of the Japanese language analysis device according to the invention. A block diagram showing the configuration of another embodiment of the Japanese language analysis device, FIG. 4 shows another 91J of the storage contents of the input sentence buffer memory.
FIG. Explanation of symbols of main parts 10...Keyboard 12...Control device 14...Determination device 16...Morphological analysis device 18...Word dictionary ROM 20...Input sentence panzer memory 22... Display device 30...Character string combination device Fig. 1 Fig. 2 (A) (B)

Claims (1)

[Scope of Claims] Input means for inputting Japanese sentence data to be parsed; and first storage means in which information such as the meaning of words is classified and stored in correspondence with character string data indicating words. a second storage means for storing and accumulating the morphological analysis results of the Japanese sentence data input by the input means; and a second storage means for capturing the Japanese sentence data input from the input means and storing them in the first storage means. morphological analysis means that performs morphological analysis and kana-kanji conversion processing of Japanese sentence data by referring to dictionary information stored in the input means; A Japanese language analysis device characterized by comprising a control means for transferring the morphological analysis results to the second storage means and for outputting the results to an output means such as a display device.