JP2585815B2 - Kana-kanji conversion device and kana-kanji conversion method - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a kana-kanji conversion apparatus and a kana-kanji conversion method for converting a kana character string into a sentence mixed with kanji kana in a document processing apparatus capable of processing Japanese. .

[Prior Art] Conventionally, a device for performing kana-kanji conversion using syntactic and semantic analysis is described in the Transactions of the Information Processing Society of Japan (vol.22, No.1), "Kana-Kanji Conversion System for Solid Writing and Its Homophone Processing" Are known.

In this method, kana-kanji conversion is performed by performing homophone selection processing including morphological analysis processing and syntactic and semantic analysis processing based on syntactic and semantic analysis information given in advance. The syntactic and semantic analysis information is information composed of a case-related relation to a verb and a semantic concept of the case.

However, this assumes that syntactic and semantic analysis information has been provided for all declinable words.

Japanese Patent Application Laid-Open No. 63-65566 describes a method in which this is improved and new syntactic and semantic analysis information for a selected word is created and registered based on the result corrected by the user. .

[Problem to be Solved by the Invention] In the above-mentioned conventional technology, in the former, when a verb without syntactic and semantic analysis information is output as a result of morphological analysis processing,
When syntactic and semantic analysis processing cannot be performed and there is a homonym having syntactic and semantic analysis information, one of the homonyms is always output regardless of the user's intention.

In the latter case, new syntactic and semantic analysis information is created based on the user's correction result, so that it is correctly converted from the next conversion, but a buffer for holding the new syntactic and semantic analysis information is required. It has become.

In general, a dictionary for syntactic and semantic analysis has a larger capacity than a dictionary for morphological analysis, and thus the latter method requires a limited number of words or a large-capacity buffer.

An object of the present invention is to output conversion information and operation of the user by giving priority to other homophones when syntactic and semantic analysis information is not present in the latest word selected by the user. Another object of the present invention is to provide a kana-kanji conversion device that improves the kanji.

[Means for Solving the Problems] In order to achieve the above-mentioned object, a plurality of homophones are added to the idiom in the sentence candidate mixed with kanji kana obtained by performing morphological analysis processing on the input kana character string. If there is a word and there is a verb without syntactic and semantic analysis information among those verbs, the verbal without the above syntactic and semantic analysis information is learned latest among the plurality of homophones. If it is, the verb that does not have the syntactic and semantic analysis information may be output.

For this purpose, the present invention provides a morphological analysis unit that performs a morphological analysis process on an input kana character string to obtain a sentence candidate mixed with kanji kana, and a homophone of a kanji kana mixed sentence candidate obtained by the morphological analysis unit. Is selected using syntactic and semantic analysis information for the idiom in the sentence mixed with the kanji kana, and if the homonym selected by the syntactic and semantic analyzing means is different from a desired result, Correction means for correcting the same homonym to another homonym, a term learning means for holding the homonym corrected by the above-mentioned correction means in time series, and by referring to the homonym held by the above-mentioned word learning means And a determination unit for determining whether to activate the syntactic and semantic analysis unit.

In the present kana-kanji conversion device, the determination means determines that a plurality of homonyms exist in the idiom in the target kanji-kana-mixed sentence, and that the phonogram having no corresponding syntactic and semantic analysis information in the homonym is used. If there is, a verb without the syntactic and semantic analysis information is retained by the term learning means, and
If the homonym of the above-mentioned verb is the latest one in time series, the verb without the above-mentioned syntactic and semantic analysis information is output without activating the above-mentioned syntactic and semantic analysis means. Activate the syntactic and semantic analysis means.

The kana-kanji conversion device can be applied to an information processing device.

[Operation] The homonyms modified by the user are stored in a time-series FIFO manner by the term learning means. On the other hand, when performing the kana-kanji conversion, the above-mentioned determination means determines whether there is a plurality of homophones in the target kanji mixed sentence in the kana-kana sentence and the corresponding homophone does not have the corresponding syntactic and semantic analysis information. If a word is present, referencing the homophone held by the term learning means, a verb without the syntactic and semantic analysis information is held by the term learning means, and
If the held order is the latest, the parsing is output without activating the syntactic and semantic analysis means. Otherwise, the syntactic analysis means is activated.

By doing so, it is possible to output at least the word candidate selected at least immediately before by the user next time.

If the determination means determines that the syntactic and semantic analysis means is to be activated, the homonym selected by the syntactic and semantic analysis means is output.

Embodiment An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a document processing apparatus in which a kana-kanji conversion apparatus according to an embodiment of the present invention is provided in an information processing apparatus.

In FIG. 1, reference numeral 201 denotes a control device such as a microprocessor for controlling the entire operation. Reference numeral 202 denotes an input device for giving instructions for inputting, converting, and correcting kana characters.
A display device 208 displays a conversion result and a correction result. Also, 200 is the kana-kanji conversion device of the present embodiment, 209
Is a storage device.

FIG. 2 is a block diagram showing the configuration of the kana-kanji conversion device 200 of FIG.

In FIG. 2, reference numeral 203 denotes morphological analysis means for converting a kana character string input from the input device 202 into a sentence candidate mixed with kanji and kana.

Reference numeral 205 denotes a syntactic and semantic analysis unit that selects and determines a homophone of a candidate sentence mixed with kanji kana converted by the morphological analysis unit 203. The homophone selected and determined by the syntactic and semantic analysis unit 205 is output to the display device 208 as a conversion result.

Reference numeral 204 denotes a determination unit that determines whether to start the syntactic and semantic analysis unit 205. If the determining unit 204 determines that the syntactic and semantic analyzing unit 205 is not activated, the determining unit 2
04 is converted to the display device 208 as the conversion result.
Is output to Details of the determination means 204 will be described later.

Reference numeral 206 denotes a correction unit that corrects a conversion result according to a user's instruction. This is because the user outputs a corrected image to the input device 20 with respect to the conversion result output to the display device 208.
When input is made from step 2, the correction result is output to the display device 208 in response to the input.

Reference numeral 207 denotes a term learning unit that registers the homophone corrected by the correction unit 206 in a FIFO manner.

The morphological analysis means 203, the determination means 204, the syntactic and semantic analysis means 205, the correction means 206, and the term learning means 207 are executed by causing the control device 201 or a different microprocessor to execute a program corresponding to each function. , The function can be achieved. The program can be stored in the storage device 209 in advance.

 The feature of the present embodiment lies in the operation of the determination means 204.

In general, in Japanese verbs, there are many cases where a plurality of homonyms exist, such as "rehabilitate", "compose", and "proofread". The syntactic and semantic analysis processing for eliminating polysemy of these homophones realizes high-accuracy kana-kanji conversion by creating conversion results prior to verbal learning. However, if syntactic and semantic analysis information is held for all of these words, a large-capacity storage device is required.

Therefore, a method of giving syntactic and semantic analysis information only to frequently used words is conceivable. However, in this method, for example, if the syntactic and semantic analysis information is added to “compose” and not to “correct”, simply input “kosei” and perform kana-kanji conversion. When doing so, since syntactic and semantic analysis processing cannot be applied,
By the term learning process, a result desired by the user is obtained from the next time. However, if the user inputs "make this part", and the conversion result is created by the syntactic and semantic analysis processing, it will be "compose the sentence", and "compose" with the syntactic and semantic analysis information will always be output. However, no matter how many times the user corrects "calibrate", a correct conversion result cannot be obtained, causing a decrease in operability.

Therefore, in the present embodiment, by referring to the term learning information, it is determined whether or not to activate the syntactic and semantic analysis means 205, and by performing the kana-kanji conversion, it is possible to determine whether or not the vocabulary for which the user has learned the term immediately before. , And prioritize the result of the syntactic and semantic analysis processing.

In the following, the present embodiment will be described with reference to FIG.

FIG. 3 is a flowchart showing the operation of the determining means 204.

When a sentence candidate with a kanji kana mixed with a sentence type that can be analyzed syntactically and semantically by the morphological analysis unit 203 is passed to the determination unit 204,
The determining means 204 determines in step 101 whether or not a homonym exists in the idiom in the sentence candidate mixed with the kanji kana, and if there is no homonym, in step 102, the syntactic and semantic analysis Check whether there is any information, and proceed to step 106.

If a homophone exists, in step 103,
First, the order of term learning in the term learning means 207 is investigated for each homonym.

If there is no homonym that has been learned, the process proceeds to step 108, where it is checked whether or not there is syntactic and semantic analysis information for all homonyms.
Proceed to.

If it is determined in step 109 that there is no homonym having syntactic and semantic analysis information, in step 111, homonyms are selected in the learning order or the dictionary registration order. If there is at least one homophone having syntactic and semantic analysis information, the step
At 110, the synonyms of the declinable word are subjected to syntactic and semantic analysis without special priority, and the homonyms are selected according to the result.

If it is determined in step 104 that there is at least one homonym that has been learned, the process proceeds to step 105, in which there is syntactic and semantic analysis information for the homonym having the latest order of term learning in the term learning means 207. Check if you want to.

As a result, if it is determined in step 106 that there is syntactic and semantic analysis information, the process proceeds to step 110, where syntactic and semantic analysis is performed, and a homophone is selected based on the result.

If it is determined in step 106 that syntactic and semantic analysis information does not exist in the latest homonym, the syntactic and semantic analysis means 205 is not activated, and in step 107, the latest homonym is selected. , The conversion result.

By doing as above, even if at least the synonym candidate of the verb selected by the user immediately before has no syntactic and semantic analysis information, the user can obtain a result desired by the user in the next input. In other cases, by performing syntactic and semantic analysis processing, a more correct result can be obtained for the entire sentence.

[Effects of the Invention] As described above, according to the present invention, it is possible to configure a kana-kanji conversion device that does not reduce the operability of the user while reducing the dictionary capacity for syntactic and semantic analysis, and efficiently creates a document. There is an effect that can be.

[Brief description of the drawings]

FIG. 1 is a block diagram of a document processing apparatus in which an information processing apparatus is provided with a kana-kanji conversion apparatus according to one embodiment of the present invention, and FIG. 2 is a block diagram showing a configuration of the kana-kanji conversion apparatus of FIG. FIG. 3 is a flowchart showing the operation of the determination means. 200 …… Kana-Kanji conversion device, 201 …… Control device, 202 ……
Input device, 208 display device, 209 storage device, 203
... morphological analysis means, 204 ... determination means, 205 ... semantic analysis means, 206 ... correction means, 207 ... term learning means.

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Kozo Mineki 1-1-1, Higashitaga-cho, Hitachi City, Ibaraki Prefecture Inside the Hitachi, Ltd. Taga Plant

Claims (4)

(57) [Claims] A morphological analysis means for performing a morphological analysis process on an input kana character string to obtain a sentence candidate mixed with kanji kana; and holding syntactic and semantic analysis information for one or more declinable words. Syntactic and semantic analyzing means for selecting the same phonetic word of the kanji and kana mixed sentence candidate obtained by using the syntactic and semantic analysis information held for the words in the kanji and kana mixed sentence, and the syntactic and semantic analyzing means If the homonym selected by the above is different from the desired result, a correcting means for correcting the homonym to another homonym, and term learning for holding the homonym corrected by the correcting means in time series Means for determining whether or not to activate the syntactic and semantic analysis means by referring to the homophone words held by the term learning means, wherein the determination means comprises When there are a plurality of homophones in the vocal words in the candidate sentence, and there is a vocal language in which the syntactic and semantic analysis information corresponding to the If the semantics whose semantic analysis information is not held are held by the term learning means, and are the latest in the chronological order among the homophones of the semantics, the syntactic / semantic analysis means is not activated. A kana-kanji conversion device characterized in that the syntactic and semantic analysis information is not stored, and if not, the syntactic and semantic analysis means is activated. 2. An information processing apparatus comprising: an input device capable of inputting a kana character string; a display device capable of displaying kana and kanji; and a kana-kanji conversion device according to claim 1. apparatus. 3. An input means for inputting a kana character string,
A display unit for displaying at least a processing result, a morphological analysis unit for performing a morphological analysis process on the kana character string input by the input unit to obtain a sentence candidate mixed with kanji kana, and a syntactic and semantic analysis for one or more verbs Hold information,
Syntactic and semantic analysis means for selecting homophones of kanji and kana mixed sentence candidates obtained by the morphological analysis means using syntactic and semantic analysis information held for declinable words in the kanji kana mixed sentence, When the homonym selected by the syntactic and semantic analyzing means is different from a desired result, a correcting means for correcting the homonym to another homonym, and a homologous word corrected by the correcting means in time series. Terminology learning means to be held, and judgment means for judging whether to activate the syntactic and semantic analysis means by referring to the homophone words held by the word learning means, wherein the judgment means When there are a plurality of homophones in the idiom in the sentence candidate mixed with kanji kana, and there is an utterance that does not hold the syntactic and semantic analysis information corresponding to the syntactic and semantic analysis information among the homonyms, , The syntax above If the vocabulary for which taste analysis information is not retained is retained by the term learning means, and is the latest one in time series among the homonyms of the vocabulary, the syntactic and semantic analysis means is not activated. A document processing apparatus which outputs a verb in which the syntactic and semantic analysis information is not stored, and activates the syntactic and semantic analyzing means if not. 4. A method in which syntactic and semantic analysis information for one or more verbs is held in advance, a morphological analysis process is performed on an input kana character string, a kanji kana mixed sentence candidate is obtained, and the obtained kanji kana is obtained. A homonym of a mixed sentence candidate is selected using syntactic and semantic analysis information stored in advance for a word in the kanji kana mixed sentence, and if the selected homonym is different from a desired result, the homonym is selected. In a kana-kanji conversion method that modifies a word to another homonym and learns the corrected homonym, words in the kana-kana kana-mixed sentence candidate obtained by performing morphological analysis on the input kana character string In the case where there are a plurality of homophones and there is a verb in which syntactic and semantic analysis information is not stored in advance among those verbs, the verb in which the syntactic and semantic analysis information is not stored in advance is Most recently learned homonyms , Outputting a verb without the syntactic and semantic analysis information.