JPH0329054A - Kana/kanji converter - Google Patents

Abstract

PURPOSE:To register only words having grammatical meanings into a word dictionary by extracting the combination of a noun or a word equivalent to a noun and an affix out of the morpheme analysis results and registering the combination as a new word when a word is obtained in a correction process. CONSTITUTION:A combination extracting means 7 is provided in addition to a conventional morpheme analyzing means 4 and a correction means to extract the combination of a word equivalent to a noun and an affix which are easily combined together as a new word out of the morpheme analysis results. Furthermore a word registering means 8 registers the extracted combination into a word dictionary as a new word only when the combination is obtained via the correction process. Thus it is possible to prevent such a case where the capacity of the word dictionary is increased due to the inadvertent registration of words and to expand the word dictionary with the significant words only.

Description

[Detailed Description of the Invention] (Industrial Application Field) The present invention relates to a kana-kanji conversion device, and particularly to a kana-kanji conversion device for efficiently registering compound words that are not registered in a dictionary during kana-kanji conversion. be. (Prior Art) With the improvement of computer-based document processing technology, document processing devices such as Japanese word processors have come into general use. The concept of using document processing devices is that documents can be created efficiently by using various functions. As an input function for this type of Japanese processing device, kana-kanji conversion is performed by inputting a desired character string containing kanji and kana using its pronunciation information, and then performing morphological analysis processing such as word dictionary search and grammar test processing. Based on the analysis results (a grammatically correctly arranged word string), a string of characters containing kanji and kana is found. Character strings containing many types of kanji and kana can be entered in the form of kana and romaji using standard high-speed character input methods such as keyboards.
It has been widely used since ancient times. However, Kana-Kanji conversion has the disadvantage that it cannot be analyzed unless the word is registered in the dictionary. For this reason, users had to manually register unregistered words in advance, which placed a heavy burden on users. As a conventional technique to overcome this problem, JP-A-63
-25766r Japanese Input Device" has been proposed. This allows the user to input a kana character string, separate words, and select homonyms without being aware of whether the word is registered or not. The system identifies unregistered words and registers them in a single-open dictionary. For example, assume that the word "Nichiden" is not registered. The user first inputs ``Nichiden J'' and tries to convert it into kana-kanji.However, because ``Nichiden'' is not registered, the analysis fails. Therefore, the user divides the input kana character string into ``Nichi 1'' and ``Den'', and then performs the kana-kanji conversion again to create ``日、r-den'' (the coined word component of Chinese is also considered as a word). ) and
As a result, it is possible to obtain "Nichiden" by combining these. At the same time, the above "Japanese input device" is able to confirm that the input kana character string "Nichiden" is the reading, and finally The resulting kanji-kana mixed character string ``Nichiden j'' can be registered in the word dictionary, so the user can immediately obtain ``Nichiden'' from the next time onwards. (Problem N to be solved by the invention) As described above, in the conventional technology, the dictionary automatically retrieves unregistered words in which the input kana character string is the reading and the obtained character string mixed with kanji and kana is the notation. You can register. However, in the conventional method, all word candidates determined as Sueno #jB are registered in the word dictionary without exception, so simple combinations of letters that have no grammatical meaning are automatically registered, resulting in the dictionary capacity being reduced. This will lead to an increase in (Means for Solving the Problem) As mentioned above, the conventional technology has the drawback that grammatically meaningless character strings are also registered as words. However, ``nouns or words equivalent to nouns (nominal expressions such as sa-verbal nouns and adjective verbs, hereinafter referred to as ``noun equivalents'') and affixes (
By effectively utilizing the characteristic of Japanese language that "prefixes or suffixes can be combined to form new words," it will be possible to register only grammatically meaningful words in a word dictionary. The present invention focuses on this point, and according to the present invention,
an input means for inputting a kana character string; a word dictionary having at least word pronunciation, notation, and grammatical information; morphological analysis means for correctly performing morphological analysis, storage means for temporarily storing the morphological analysis results by the morphological analysis means, and output means for outputting a character string containing kanji and kana based on the morphological analysis results. and a kana-kanji conversion device comprising a correction means for correcting the morphological analysis result by selecting a homophone B or changing a word break position when the character string containing kanji and kana is not desired. combination extracting means for extracting a combination of a noun or a word corresponding to the noun and an affix from among the combinations; and when at least one word of the combination extracted by the combination extracting means is obtained by a correction process by the correction means, the combination and word registration means for registering the word as a new word in the AAA dictionary. (Operation) In addition to the conventional morphological analysis means and correction means, the present invention provides a combination extraction means for extracting combinations of noun equivalents and affixes that are likely to be combined to form new words from the morphological analysis results. We have provided a word registration means that registers the combination as a new word in the word dictionary only when the combination is obtained as a result of the correction process. This prevents the capacity of the word dictionary from increasing due to careless registration of words, and allows the dictionary to be expanded with only meaningful words. (Example) The present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing one embodiment of the present invention. In the figure, 1 is a control unit that controls the overall operation, and 2 is an input means for inputting a kana character string to be used for kana-kanji conversion. However, in this implementation, instructions such as homophone selection instructions, delimiter position specifications, correction processes such as correction of kana character strings, and confirmation of the obtained analysis results are also performed from the input means 2. 3 is a word dictionary that includes at least the pronunciation and notation of words and grammatical information; 4 is a word dictionary 3;
This is a morphological analysis tool that performs morphological analysis by searching and grammatical analysis. It is assumed that the morphological analysis means 4 also performs processes such as homophone selection and word break position change under the control of the control section 1. 5 is an analysis result storage means for temporarily storing the morphological analysis results obtained by the morphological analysis means 4;
6 is an output means for generating and outputting a kanji-kana character string based on the morphological analysis results in the analysis result storage means 5; 7 is a combination of noun equivalent words and affixes from the morphological analysis results in the analysis result storage means 5; It is a combination extraction method that extracts
Reference numeral 8 denotes word registration means for registering the combinations extracted by the combination extraction means 7 in the word dictionary 3 as new words. The present embodiment will be described below by taking as an example a procedure for registering the word "under the scorching sun" in the word dictionary 3 by inputting the sentence "under the scorching sun". Note that FIG. 2 is a flowchart showing the flow of the operation of the present invention. In addition, the word dictionary 3 contains the noun “
It is assumed that although the suffixes ``hoten J'', ``ka'', ``lower J'', and the particle ``de'' are registered, the noun ``hotenka'' is not registered.

First, the user inputs the kana character string ``Entenkade'' using the input means 2 (step 201). The input kana character string is passed to the morphological analysis means 4 according to instructions from the control unit 1. Morphological analysis means 4 performs morphological analysis processing such as searching the word dictionary 3 and grammar test on the ephemeral character string "endenkade" (step 202>. The obtained analysis results are stored in the analysis result memory It is temporarily stored in the Pi 5 (step 203).The analysis result storage means 5 includes, for example, a third
As shown in Figure (a), the pronunciation, spelling, and grammatical information of the words that make up the analysis results are stored. Next, output hand P! 16, under the control of the control unit 1, combines the notations of the analysis results in the analysis result memory P15 to output a character string containing kanji and kana (step 204>. In this embodiment, a character string containing kanji and kana is output). The column "Entenka de" is output. The user checks whether the output "Entenka de" is the desired one (step 205), and in this case, it is not the desired one, so the user inputs it. Correction work is performed by giving an instruction from means 2 (step 206) In this embodiment, correction can be performed by selecting ``shita'', which is a tg between 2, instead of the word ``ka''.This result , the contents of the analysis result storage means 5 are changed as shown in FIG.
, the user's desired character string ``under the scorching sun'' is output.At this point, the control unit l also remembers that the correction work has been performed on the suffix ``under''. The kanji-kana character string "under the scorching sun" is confirmed by a confirmation instruction from the user (for example, by pressing the "non-conversion key J" in a normal word processor).Here, the control unit 1 sends the analysis result to the combination extraction means 7. A combination of a noun equivalent word and an affix is extracted from the analysis results in the storage means 5 (step 207).In this example, the noun "Hoten" and the suffix "shita" are extracted.
The combinations of are extracted. Here, iI1 Gobe 1 is
It is determined whether or not this combination has been obtained through correction work (step 208), and if so, this combination is registered in the word dictionary 3 as a new word. In the example of this explanation, since "lower" of the combination "hoten" and "lower" is obtained by modification, the control unit 1 issues an instruction to the word registration means 8 to register the word. The word registration means 8 inputs the word whose reading is "entenka" and the notation is "hotenka" into the candy dictionary 3 based on the combination of "hoten" and "lower j" extracted by the combination extraction means 7. register. At this time, the grammatical information is simply "noun"
In addition to ways to enjoy
In the case of , the adjective verb type suffix (``target, r wind'')
In this case, there are methods such as using it as an adjective verb, and providing information indicating that it is an automatically generated new word. Through these series of steps, "scorching sun" is registered in the word dictionary 13. In this embodiment, the control unit 1 and the combination extracting means 7 include a microprocessor and a memory element (
(RAM). In addition, the input means 2, the word dictionary 3, the morphological analysis means 4, the analysis result storage means 5, and the output means 6 are those realized in a conventional Japanese word processor, etc., and the word registration means 8 is used for registration. Except for the fact that the candidates to be used are provided by the combination extracting means 7, this is realized in a conventional Japanese word processor, so a detailed explanation will be omitted.

Furthermore, in the procedure shown in Figure 2, the combinations of noun equivalents and affixes are extracted and then the presence or absence of correction work is checked, but it is also possible to extract combinations only when there is correction work. (Effects of the Invention) As explained above, according to the present invention, words are automatically registered based on the kana character string input by the user, so that the user can efficiently obtain desired words from now on. be able to.
In addition, when the desired character string containing kanji and kana is obtained through modification operations such as correction and delimiter specification, the word is automatically registered only in the case of a combination of a noun equivalent and an affix, so only meaningful words are registered. can6

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a flowchart showing an example of the processing flow in this embodiment, and FIG. 3 (a> and (b) is the implementation of the analysis result storage means 5. It is a conceptual diagram for explaining an example, 1... Control unit, 2... Input means, 3... Word dictionary, 4... Morphological analysis means, 5
... Analysis result storage means, 6... Output means, 7...
Combination extraction means, 8... Single crystal registration means.

Claims (1)

[Claims] an input means for inputting a kana character string; a word dictionary having at least word pronunciation, notation, and grammatical information; morphological analysis means for correctly performing morphological analysis, storage means for temporarily storing the morphological analysis results by the morphological analysis means, and output means for outputting a character string containing kanji and kana based on the morphological analysis results. In the kana-kanji conversion device, the kana-kanji conversion device is provided with a correction means for correcting the morphological analysis result by selecting a homophone or changing the word break position when the character string containing kanji and kana is not desired. a combination extraction means for extracting a combination of a noun or a word corresponding to a noun and an affix from among the nouns, and at least one word of the combination extracted by the combination extraction means is obtained by a correction process by the correction means; and word registration means for registering the combination as a new word in the word dictionary.