JPH0772896B2 - Japanese input method - Google Patents

Description

Detailed Description of the Invention

[0001]

The present invention relates to a Japanese input method,
Particularly, the present invention relates to a Japanese input method for outputting a more appropriate kanji / kana mixed sentence from a sentence in which kanji and kana that do not have reading information input.

[0002]

2. Description of the Related Art The conventional Japanese input method has a kana-kanji conversion function that converts the input Japanese reading information into a kanji-kana mixed sentence, and performs kana-kanji conversion from the beginning of the input reading information. The operator selects homonyms and the like, and creates an appropriate kanji-kana mixed sentence desired by the operator.

[0003]

As described above, in the conventional Japanese input method, kana-kanji conversion is performed by inputting kanji reading information at the time of input, and input without giving kanji reading information. For example, in the case of pen input or character input by an image reader or the like, there is a problem that an output of an appropriate kanji / kana mixed sentence cannot be expected.

Particularly, when a handwritten input is made, a part of the kanji in the idiom is forgotten to be remembered, and this problem occurs when the kana is input.

An object of the present invention is to provide a Japanese input method capable of converting and outputting an appropriate kanji / kana mixed sentence even when inputting a kanji / kana mixed sentence to which no kanji reading information is given.

[0006]

The Japanese input method of the first invention is a Japanese-Kan dictionary storing means for storing a kana heading and corresponding kanji and its word information, a kanji heading and a kanji word of this kanji, and Idiom dictionary storage means for storing the word information, adjunct word information storage means for storing adjunct word headings and their connection information, and input means for inputting a sentence in which kanji and kana without reading information are mixed. A character recognition means for generating a character code of a character constituting them from the text information supplied from the input means, and a kanji code of a kanji corresponding to the kana code supplied from the character recognition means and its word information And a kanji character corresponding to the kanji search means for searching the Japanese and Chinese dictionary storage means and the kanji code supplied from the character recognition means and the kanji character search means. And its word information are searched from the phrase dictionary storage unit, and when the phrase supplied from the phrase search unit is a Kanji phrase input without the reading information, it is searched by the Kanji search unit. When it is not likely to extract a kanji containing a kanji, an output kanji candidate extracting means for extracting a kanji containing a kanji input without the reading information, and the adjunct word from the word information of the kanji provided by the output kanji candidate extracting means. A phrase candidate creating unit that refers to the information storage unit to search for an adjunct word of the phrase and creates a phrase candidate composed of the phrase and the adjunct word; and an output phrase from the phrase candidate supplied from the phrase candidate creating unit. And an output phrase determining means for determining
The Japanese input method according to the invention is characterized in that, in the first invention, the input means inputs a character shape.

[0007]

Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the Japanese input method of the present invention.

The Japanese input method of FIG. 1 includes an input unit 1, a character recognition unit 2, a Kanji search unit 3, a idiom search unit 4, an output Kanji candidate extraction unit 5, a phrase candidate creation unit 6, and Output clause determining unit 7, output unit 8, Japanese-Korean dictionary 11, and idiom dictionary 12
And an accessory word information table 13.

The Japanese-Kanji dictionary 11 includes kana headings, corresponding kanji and their word information (eg, part-of-speech information).
And are stored, and the storage state is shown in FIG. Both kana and kanji are stored as character codes, and word information is stored as a corresponding code. The same applies hereinafter.

The kanji word dictionary 12 stores kanji headings, kanji words of the kanji (including the kanji itself) and word information thereof, and the storage state is shown in FIG.

The adjunct word information table 13 stores adjunct word headings and connection information thereof, and the storage state is shown in FIG.

The input unit 1 is for inputting a character shape, and does not need to input kanji reading information. A pen input device, an image reader or the like can be used for the input unit 1.

The character recognition unit 2 recognizes an input character from the character shape input from the input unit 1 and outputs a corresponding character code. The character recognition mechanism used in OCR or the like can be used.

The Kanji search unit 3 searches the Kana dictionary 11 for a Kanji group and corresponding word information corresponding to the inputted Kana code.

The idiom search unit 4 obtains the idiom group corresponding to the input Kanji code (including the Kanji code retrieved by the Kanji retrieval unit 3) from the inputted Kanji code and these word information from the Jukugo dictionary 12. Search for.

The output Kanji candidate extraction unit 5 inputs the idiom group searched by the idiom search unit 4, and when the idiom group is a Kanji idiom entered without reading information, the Kanji retrieval unit 3 searches the Kanji. A idiom including "" is searched, and if not, a idiom containing a kanji input without reading information is searched and extracted as an output kanji candidate.

The phrase candidate creating unit 6 creates a phrase candidate by searching the adjunct word information table 13 for an adjunct word associated with the output kanji candidate extracted from the extracted word information of the output kanji candidate.

The output phrase determining unit 7 determines an output phrase from the phrase candidates by a phrase determining process such as the two-phrase longest matching method which has been often performed conventionally.

The output unit 8 outputs the determined output phrase by printing or displaying.

The operation of this embodiment will be described below with reference to an example.

In the present embodiment, a case will be described in which only a part of the desired idiom is input in kanji and the rest is input in hiragana, and the kanji-kana mixed sentence is output as a more suitable kanji-kana mixed sentence. .

First, the pattern of "Kanji + Kana" will be described.

A case will be described in which a kanji / kana mixed sentence “meet” is input from the input unit 1 such as an image reader and a more suitable kanji / kana mixed sentence is output.

When the text "meeturu" mixed with kanji and kana is input through the input unit 1, the character recognition unit 2 recognizes the character from the character shape, and the character code "89EF (kai), 82ED" is input.
(Wa), 82B7 (su), 82E9 (ru) "is output.

Among the obtained character codes, the kanji dictionary 11 for each of the character code parts excluding the kanji code part in the kanji search part 3 is
Is obtained, and for example, for "wa", a kanji group such as "ring (noun), story (noun), ..." and its word information are obtained.

Further, the idiom search unit 4 searches the idiom dictionary 12 using the character code "89EF" of the input Chinese character "kai" as a search key, and searches for "kai" (noun, awa line 5). Verbs), companies (nouns), conversations (nouns, sa verbs), and the like, and the word information thereof.

Next, in the output kanji candidate extraction part 5, since the idiom of the kanji "kai" entered without reading information is input in this case, from the idiom group obtained by the idiom search part 4, By extracting the idiom including the kanji obtained by the kanji search unit 3 as an output kanji candidate, "conversation (noun, sa verb)" can be obtained.

Then, in the phrase candidate creating unit 6, based on the word information (noun, sa verb) of "conversation", the hiragana part excluding the "conversation" part of the input kanji-kana mixed sentence. By searching the adjunct information table 13 for “do”, the adjuncts “do” and “su” connectable to the sa verb are obtained. Therefore, the phrase "suru" and "su" are connected to "conversation" to create a phrase candidate, and "conversation, conversation, conversation" is acquired.

Next, in the output phrase determining unit 7, a more appropriate candidate as a phrase is determined from the created phrase candidates by the conventional two-phrase longest matching method or the like.

Next, the phrase determined by the output phrase determining unit 7 is output to the display, the printer or the like at the output unit 8, and the process is terminated.

Next, the pattern of "kana + kanji" will be described.

When a sentence "Kai Talk" mixed with kanji and kana is input through the input unit 1, the character recognition unit 2 causes the character codes "82A9 (or), 82A2 (I), 9862.
(Story), 82B7 (su), 82E9 (ru) "is obtained.

From the obtained character codes, the Kanji search section 3 searches the Japanese-Kanji dictionary 11 for "Kai" for the character code sections excluding the Kanji code section, and for "Kai", "Kai (noun)," A group of kanji such as "kai (noun), ..." and its word information are obtained.

Next, the phrase search unit 4 searches the phrase dictionary 12 using the character code "89F1" of the Chinese character "times" obtained as described above as a search key to search for "times" (noun, noun,
Sa line 5 dan verb), answer (noun), retrospective (noun), ...
A phrase group such as "..." and its word information are obtained.

Next, in the output kanji candidate extraction part 5, in this case, it is not the case that the idiom of the kanji "talk" input without reading information is input, so that the idiom group obtained by the idiom search part 4 Is searched for an idiom that includes the kanji "story" entered without reading information. However, it turns out that there is no corresponding idiom.

Then, by using the character code “89EF” of the kanji “kai” obtained next by the kanji search part 3 as a search key, the phrase search part 4 searches the phrase dictionary 12 to find the word “kai”. Meetings (nouns, a-wa lines 5 verbs), companies (nouns), conversations (nouns, sa verbs), ... "
Get a group of idioms such as and the word information.

Next, in the output kanji candidate extraction section 5, in this case as well, it is not the case that the idiom of the kanji "talk" input without reading information is input, so that the idiom group obtained by the idiom search section 4 By extracting the idiom that includes the kanji "story" input without reading information as an output kanji candidate,
"Conversation (noun, sa verb)" is obtained.

Then, in the phrase candidate creating section 6, based on the word information (noun, sa verb) of "conversation", the hiragana part "do" except the part corresponding to "conversation" of the input kanji-kana mixed sentence. To the adjunct information table 13
By searching for the adjuncts “suru” and “su” that can be connected to the sa verb. Therefore, the phrase "suru" and "su" are connected to "conversation" to create a phrase candidate, and "conversation, conversation, conversation" is acquired.

Next, the output phrase determining unit 7 determines a more appropriate phrase from the created phrase candidates.

Next, the output section 8 outputs the clause determined by the output clause determining unit 7 to a display or a printer.

As described above, according to the present embodiment, a sentence in which kanji and kana for which reading information is not input is mixed can be output by utilizing a Japanese-Kanji dictionary, a idiom dictionary or the like to output a more suitable kanji and kana mixture. .

[0043]

As described above, according to the Japanese input method of the present invention, a sentence containing a mixture of kanji and kana for which reading information is not input is searched for in a kanji dictionary, a idiom dictionary, or the like, so that the kanji is more suitable. It has the effect of being able to output sentences with mixed kana.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a Japanese input method of the present invention.

FIG. 2 is a data storage diagram showing a data storage state of a Japanese-Chinese dictionary.

FIG. 3 is a data storage diagram showing a data storage state of a phrase dictionary.

FIG. 4 is a data storage diagram showing a data storage state of an adjunct word information table.

[Explanation of symbols]

 1 Input Part 2 Character Recognition Part 3 Kanji Search Part 4 Jukugo Search Part 5 Output Kanji Candidate Extraction Part 6 Phrase Candidate Creation Part 7 Output Phrase Decision Part 8 Output Part 11 Japanese-Kanji Dictionary 12 Jukugo Dictionary 13 Adjunct Information Table

Front Page Continuation (72) Inventor Hiroshi Tsuruta 2-15-3 Kume, Naha City, Okinawa Prefecture Okinawa Electric Software Co., Ltd. (56) Reference JP-A-1-321552 (JP, A) JP-A-3- 100758 (JP, A) JP-A-4-177465 (JP, A)

Claims (2)

[Claims] 1. A Japanese-Kanji dictionary storing means for storing a kana heading and corresponding kanji and its word information, a kanji dictionary storing means for storing a kanji heading, a kanji of this kanji and its word information, and an annex. These are composed of auxiliary word information storage means for storing a headline and its connection information, input means for inputting a text in which kanji and kana that do not have reading information are input, and text information supplied from the input means. A character recognition means for generating a character code of a character, and a kanji search means for searching the kanji code corresponding to the kana code and its word information corresponding to the kana code supplied from the character recognition means from the Japanese-kanji dictionary storage means, From the kanji code supplied from the character recognition means and the kanji search means, the corresponding idiom and its word information are detected from the idiom dictionary storage means. When the phrase search means for searching the phrase and the phrase supplied from the phrase search means are the kanji of the kanji input without the reading information, the phrase containing the kanji searched by the kanji search means is not extracted. An output kanji candidate extraction means for extracting a kanji word containing a kanji input without the reading information, and an adjunct of the kanji word by referring to the attached word information storage means from the word information of the kanji word supplied from the output kanji candidate extraction means. A phrase candidate creating means for searching a word to create a phrase candidate consisting of the phrase and the adjunct word; and an output phrase determining means for determining an output phrase from the phrase candidate supplied from the phrase candidate creating means. Japanese input method characterized by. 2. The Japanese input method according to claim 1, wherein the input means inputs a character shape.