JPH08235171A - Japanese syllabary and chinese character converting device - Google Patents

Abstract

PURPOSE: To perform conversion again according to correct paragraph sectioning and prevent misconversion due to wrong paragraph sectioning by putting a paragraph right before an input position back into HIRAGANA(cursive form of Japanese syllabary), merging it currently inputted HIRAGANA(square form of Japanese syllabary), and performing conversion even if paragraph sectioning is done by conversion indication at a position irrelative to original paragraph sectioning. CONSTITUTION: When an undetermined character string is present in a conversion character buffer 62, whether or not an input character string has the same reading with a suffix is retrieved from a suffix table 46. When the same reading is found, the KANJI(Chinese character) and KANA(Japanese character) mixed paragraph right before being inputted to the conversion character buffer 62 is converted into HIRAGANA, which is merged with the current input character string; and whether or not the character string is present in a suffix example dictionary 53 is retrieved. When the string is retrieved, it is stored as a 1st candidate in the conversion buffer 62. When not, on the other hand, a KANA-KANJI mixed paragraph positioned right before the input of the conversion buffer 62 is converted into HIRAGANA, and when the same reading is not found in the suffix table 45, retrieval from an independent word/proper noun dictionary is performed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a kana-kanji conversion process for converting a character input in hiragana into a kana-kanji mixture such as a Japanese word processor.

[0002]

2. Description of the Related Art As a conventional technique, Japanese Patent Publication No. 61-3268 discloses a technique for performing kana-kanji conversion by learning not only word units but also word combinations. In other words, if you have learned the conversion "directly below" for the reading "masuta", enter "you gotta",
When kana-kanji conversion is performed, the meaningless result of "top / underneath" is output because the conversion is done with "meta / ta".

Therefore, a learning buffer for registering the learning result of a compound word composed of a combination of a plurality of words is provided, and the learning result of each word "just below" is learned from the compound word "Ore". By prioritizing + “masuta”, the conversion result of “got” is output.

Japanese Unexamined Patent Publication (Kokai) No. 4-65751 discloses a technique of storing connection conditions for connecting a suffix to an independent word. This technique prevents conversion to a meaningless word when a suffix is added when converting Kana-Kanji. Semantic classification code (for example, "information", "animal", etc.) is added to an independent word, and the suffix is added only when it can be connected by collating with a table that stores which suffixes can be connected. It becomes possible to prevent meaningless suffix conversion by performing the conversion. In addition, JP-A-61
-239357 is provided with a means for inputting a prefix or suffix. For example, when inputting "reuse", "reuse" + "resume" + conversion is performed. That is, the prefix and suffix are input without conversion. According to this, erroneous conversion such as "for wife and child" can be prevented,
It is too inefficient to have input keys for all prefixes and suffixes. In Japanese Patent Laid-Open No. 63-266575, the suffix is treated as a kind of independent word, the suffix is searched as an independent word from the relationship with the previously determined clause, and the word length is also added for conversion. . Also in this invention, it is possible to prevent a meaningless word with a suffix added unnecessarily from being selected as a candidate. However, the possibility of adding an affix word to the word resulting from the conversion is not considered.

As described above, as a conventional technique, there has been a technique for deriving a correct conversion candidate from the relation between phrases.

[0006]

However, in the conventional kana-kanji conversion device, when the input character string is converted into a kana-kanji mixed sentence, if the result is a compound phrase, the first phrase and the subsequent phrase are converted. The bunsetsu was analyzed and the conversion candidates were derived using the following information.

(1) Prefix, independence word, proper noun part, meaning of suffix, co-occurrence relation in the independence part of bunsetsu (2) Case relation between independence part in bunsetsu and adjunct (3) bunsetsu And co-occurrence relationship between the self-reliance part in the above and the preceding sentence's self-reliance part / proper noun part (4) Case independence of consecutive candidates / proper noun part and adjunct part (5) Consecutive candidates' self-reliance part / Meaning of proper noun part / Co-occurrence relation (6) Information including word frequency Using such information, for example,
When the input is converted, "Sun" + "Climb""Climb"
"Climb" to "Climb" was selected.

Even when the converted data is still in an undetermined state, the previous clause information is stored, and [1] "the sun" + conversion → the sun [2] "rises" + Conversion → When the sun rose, it converted and output.

However, in the conventional technique, when "Sogo" + conversion + "kita" + conversion is input,
It will be converted to "general / enemy". This means that even if you remember the phrase information "Sogou" = "general", once it is converted, even if it is in the unconfirmed state, 1
This is because it is recognized as one bunsetsu, and there is a lack of a process of revising the above-mentioned (1) the prefix in the independent part of the bunsetsu, the independent word / proper noun part, the meaning of the suffix, and the cooccurrence relationship. Therefore, without being linked to the prefixes, independent words / proper noun words, meanings of suffixes, and co-occurrence relations between the prefixes of “general” and “target” in the independence part, simply entering “ki” and converting it into the most promising candidate. This is because the "enemy" of will be output first.

In the case of "Kiki" in this example, the candidate is "enemy".
There are 2 candidates for "target", but in case of "Ryo" (diagnosis fee)
Are "quantity", "dormitory", "fishing", "hunting", "good", "territory", etc., and the desired kanji cannot be called in a few conversions, resulting in a list of meaningless words. Requires the conversion key to be operated many times, and it is difficult to select a desired kanji from many candidate groups.

[0011] Similarly, "tentensosaku (abduction side)"
"Tenten" + conversion + "Hansoku" when converting
When conversion is input, the conversion result is often "turning / foul".

[0012] This means that even if the word "tenten-hansoku" = "inversion side" is registered in the dictionary, the user unconsciously converts "inversion" and "inversion" into two phrases. And even if it is in an undecided state, the converted bunsetsu will be recognized as one bunsetsu, and the (1) bunsetsu's prefix, independent word / proper noun part, and suffix This is because the process of reexamining the meaning and co-occurrence relationship is missing. For this reason, if you enter the word "Hansoku" without being linked to the prefix, independent word / proper noun word, meaning of the prefix, and co-occurrence in the "tenten" and "hansoku" independence parts, and perform conversion "Foul" of the most probable conversion candidates of
Is output. Similarly, "Ikki Tousen" is changed to "Ikki"
If you enter the words and "Tosen" separately, it will be "Ichiki / Win" and "Kankyukyudai" will be "Mitigation / Fifth". In such an example, in order to convert "rest", in the end, "kyu" and "dai" will be converted separately, and the number of candidates will be so large that input will be difficult. There was a problem.

Since the user has a recognition of a bunsetsu break which is not related to the bunsetsu break position determined by the system, there is a possibility that the user often presses the conversion key at a position that is not supposed to be one bunsetsu by the system. Therefore, the kanji desired by the user is not easily output, many meaningless word candidates are output, labor is required for selecting candidates, and operability is impaired.

Therefore, the present invention does not force the user to use the concept of phrase breaks in the system, but assumes that the user may misplace the phrase breaks, and the user gives a conversion instruction with the wrong phrase. Even when the bunsetsu is delimited, the system determines whether or not the bunsetsu delimiter can be used as the original bunsetsu delimiter, and promptly outputs the candidate desired by the user.

[0015]

According to the present invention, in order to solve the above problems, input means for inputting kana characters, an input buffer for storing kana characters input by the input means, and an input buffer for storing kana characters are stored in the input buffer. A kana-kanji conversion device for converting a kana-kana into kanji kana-kanji conversion means, and a kana-kanji conversion device including a conversion character buffer storing the conversion candidates converted by the kana-kanji conversion means. In response to the instruction, a returning unit that returns the conversion candidate stored in the conversion character buffer to a kana character, a combining unit that combines the returned kana character and the kana character stored in the input buffer, and the combining Kana characters combined by means as the conversion target character string,
The reference means for referencing the dictionary, and the reference means, the notation of the kanji that partially matches the conversion candidate stored in the conversion buffer is used as the conversion candidate.

[0016]

In the present invention, if there is an undetermined Kanji / Kana mixed sentence immediately before the input Kana character when there is a Kana / Kanji conversion instruction, the previous Kanji / Kana mixed sentence is read again and this time. Combined with the input Yomi and collated with the dictionary,
In the case of "same reading" or "preceding kanji string is the same as undetermined kanji" condition in the dictionary, the notation of that word is output as the first candidate for conversion, and if it does not match the condition, normal kana kanji Perform conversion processing.

[0017]

Embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the present invention. In FIG. 1, reference numeral 1 denotes a control unit, which includes a microprocessor, a program, a table, an external storage device and the like other than those specified below, and controls what is specified below. Reference numeral 2 is a display unit including a CRT, an LCD and the like. An input unit 3 includes a keyboard and the like.

Reference numeral 4 denotes a dictionary unit such as an independent word / proper noun / affix word / phonetic dictionary / user dictionary including a meaning code and a table unit. 5 is a compound example showing the co-occurrence relationship of words /
A dictionary unit such as a prefix example / suffix example dictionary.

Reference numeral 6 is a buffer memory. The buffer memory further includes 61 input character buffers, 62 converted character buffers, 63 learning buffers, 64 homophone buffers, 65 sentence buffers, 66 screen buffers, 6
It is composed of seven division position table buffers.

The input buffer 61 is a buffer for storing the characters input from the input means 3. The conversion character buffer 62 is a buffer for storing the first candidate data when the data of the input character buffer 61 is converted. The learning buffer 63 is a buffer for increasing the priority when there is a confirmed homophone.

In the 64 homophone word buffer, a homophone word other than the first candidate of the converted character buffer 62 is stored, and another word data internally created during the conversion process. The sentence buffer 65 stores fixed sentence data. The screen buffer of 66 has a form in which the input character buffer of 61 (before conversion) or the conversion character buffer of 62 (after conversion) and the sentence buffer of 65 are combined, and the buffer tables of 61 and 62 are also in the screen buffer of 66. Have In the division position table buffer 67, bunsetsu candidates are stored as much as possible (a plurality of types) as to how the bunsetsu candidates are arranged closer to the correct answer candidate.

The kana sequence input from the input unit 3 is first 6
1 is stored in the input buffer. At the next conversion timing, the control unit 1 refers to the dictionary unit 4, the example dictionary 5, and the like to carry out Kana-Kanji conversion processing, and stores the first priority candidate in the conversion character buffer 62. Further, different word data different from the word data other than the first priority candidate are stored in the homophone word buffer of 64, and the different division position information is stored in the division position table buffer of 67, and are displayed on the display unit 2 by the input unit 3 and the operation. It is displayed and can be selected. When a homonym is selected by operating the input unit 3, the result is 63
Information is stored in the learning buffer of. The confirmed text data is copied from the conversion character buffer 62 to the text buffer 65.

As an actual example, “sougo” is placed in the middle (/ position) of the text data “teacher or parent / preparatory meeting”.
The contents of each data buffer immediately after inputting and converting a character string are as shown in FIG. 13, and the contents of each data buffer immediately after inputting “Kai” and being converted are as shown in FIG. Has become.

The conversion character string 1 is stored in the conversion character buffer 62.
It has reading information for each character. In the 64 homophone buffers, the word information when a phrase starts from each character string is stored in the 64 homophone buffers.

Although there are 66 screen buffers, if there is no undetermined character string, the contents of the 65 text buffer are output to the screen buffer as they are. When there is an undetermined character string and before conversion, 61 input character string data is inserted into the cursor position of the 65 sentence buffer and output to the screen buffer. In that case, information on where in the screen buffer the undetermined character string is stored is stored. If there is an undetermined character string after conversion, 62 converted character string data is inserted at the cursor position of the 65 sentence buffer and output to the screen buffer. In that case, information on where in the screen buffer the undetermined character string is stored is stored. In the separation position buffer 67, information on how to collect the phrase candidates in the homophone word buffer to form one sentence or a plurality of kinds of separation positions is stored.

A specific example of the present application will be described with reference to a flowchart. FIG. 9 shows a flowchart. Step S1
Wait for key input with. When the conversion key is input there, the process proceeds from step S2 to step S3. When the cancel key is input, the process proceeds to step S6. If any other key is input, the procedure advances to B.

In the process of B, it is determined in step S9 whether or not the key is a cursor movement key. If it is the cursor movement key, the process proceeds to step S10 to perform the cursor movement process. If the input is not the cursor movement key, step S1
It judges whether it is a character input at 1. Since the processing when it is not a character input is not directly related to the present invention, it is omitted. If the input is a character, the character is stored in the input character buffer in step S12. When the conversion key is input in step S2, the process proceeds to step S3.

In step S3, it is confirmed whether or not there is an unfixed character string in the converted character buffer 62. If there is an undetermined character string, the suffix table of 46 is searched for whether the input character string has the same reading as the suffix. here,
If the suffix table of 46 has the same reading, the process proceeds to step S5, and the Kanji-kana mixed bunsetsu located immediately before the input in the conversion character buffer of 62 is changed to hiragana,
It is combined with the input character string this time, and the character string is searched for in the suffix example dictionary of 53. If the same reading is retrieved, it is stored in the conversion buffer 62 as the first candidate. If it is not searched, the kana-kanji mixed phrase located immediately before the input in the conversion character buffer 62 is changed to hiragana and the 45 prefix table is searched. If there is no same reading in the prefix table, the independent word / proper noun dictionary of 41 is similarly searched.

Here again, if the same reading is not in the table,
In step S4, based on the constituent elements of the clause in FIG. 3, the data in the inflection ending table in FIG. 5, the auxiliary word table in FIG. 6, and the affix table in FIG. 7 is judged from the connection matrix in FIG. To generate. Then, the generated clauses are combined to form a sentence. Information on which position to insert the break of the clause is stored in the position table buffer in the order of 67 from the combination considered to be the best.

If the 45 prefix table and 41 independent words / proper noun dictionaries have the same reading, the character string combined with the input character string this time is represented in the prefix example dictionary and independent word / proper noun dictionary. To see if If there is, it is stored in the conversion buffer 62 as the first candidate in step S5.

When the key input is the cancel key,
In step S7, it is determined whether or not there is an undetermined output candidate. If there is an output candidate, in step S8, processing for returning the phrase at the cursor position to the input character for one phrase before is performed.

Next, referring to FIGS. 10, 13 and 14, FIG.
A description will be given in accordance with an actual example of steps S3 to S5.

"Sogo" to convert to "comprehensive"
This is an explanation in the case of inputting + conversion + “ki” + conversion.
First, enter "Sougo" as the input character string. The input characters are stored in the 61 input character buffer. Next, by inputting the conversion key, it is confirmed whether or not the unconfirmed character string exists in the conversion character buffer 62. In this example, there is no undetermined character string yet, so as many clauses as possible are generated based on the character string stored in the 61-input character string buffer (that is, normal conversion).

Then, "ki" is input. Input character is 6
1 Stored in the input character string buffer. Here, by operating the conversion key, it is confirmed whether or not there are unconfirmed characters in the conversion character buffer 62. Since there is an undetermined character string here, the 46 suffix table is searched to see if the input character string [kita] has the same reading as the suffix. Since the 46 suffix table has the same reading, the kanji kana mixed phrase [comprehensive] located immediately before the input in the 62 conversion character buffer is changed to hiragana [sougo], and the input character string this time [come]
It is searched whether the character string [Sougouki] combined with is in the 53 suffix example dictionary.

The notation character [general] is selected from the 53 suffix example dictionary, and it is confirmed in the 62 conversion buffer whether or not the character is the same as the kanji [general] changed to reading. Since the same notation is used in this example, [comprehensive] is 6 as the first conversion candidate.
Store in the conversion buffer of No. 2. Then, the 65 sentence buffer and the 62 converted character buffer are combined to create a 66 screen buffer for display.

An example in which "Ikki Tousen" is input as "Ikki" + conversion + "Tosen" + conversion will be described with reference to FIGS. 11, 15 and 16. FIG. Similarly to the above example, the 46 suffix table is searched for whether the input character string "tousen" has the same reading as the suffix. Since there is no reading here, the kanji-kana mixed phrase located immediately before the input in the 62 conversion character buffer is changed to hiragana "Ikki" and the 45 prefix table is searched. Since there is no same reading here, the kanji-kana mixed bunsetsu located just before the input of the 62 conversion character buffer is changed to hiragana, and the character string "Ikkitousen" that combines the input character strings this time is 41 independent word dictionary. / Search the proper noun dictionary.

41. A notation character [Ikkitousen] is selected from the 41 independent word dictionary / proper noun dictionary, and it is judged whether the notation is the same as the Kanji [Ikki] converted into reading in the 62 conversion buffer. Since the same notation is used, [Ikkitousen] is 62 as the first candidate.
Store in the conversion buffer of. Similarly, an example of inputting "Vice counselor" as "Fuku" + conversion + "Counselor" + conversion Fig. 1
2, and FIG. 17 and FIG. 18 will be described. Similar to the above example,
Whether the input string "sanji" has the same pronunciation as the suffix, 4
6 The suffix table is searched, and since there is no same reading, 62
The Kanji-kana mixed phrase located immediately before the input in the conversion character buffer is changed to hiragana “fuku” and the 45-prefix table is searched.

Here, since the 45 prefix table has the same reading, it is searched whether or not the character string "Fukusanji" combined with the input character string this time is in the 53 suffix example dictionary. 52. Select the notation character “Deputy Counselor” from the 52 suffix example dictionary, and confirm that it has the same notation as the Kanji character “Sub” changed to the reading in the conversion buffer 62. Since it is the same notation,
The "counselor" is stored in the 62 conversion buffer as the first candidate.

[0039]

[Effects of the Invention] Even when the user breaks a phrase by a conversion instruction at a position unrelated to the original phrase break, the phrase immediately before the input position is returned to Hiragana once and the current phrase is input. By combining with Hiragana and performing the conversion, the conversion can be performed again according to the correct segment break, and the incorrect conversion due to the incorrect segment break is prevented.

[Brief description of drawings]

FIG. 1 is a block diagram of the present invention.

FIG. 2 is a diagram showing an example of a text input screen according to the present invention.

FIG. 3 is a diagram showing an example of constituent elements of a clause.

FIG. 4 is a diagram showing a connection matrix relating to connection of words.

FIG. 5 is a diagram showing a table of inflection endings of a word.

FIG. 6 is a diagram showing a table storing attached words.

FIG. 7 is a diagram showing a table storing affix words.

FIG. 8 is a diagram showing an example of a suffix example dictionary.

FIG. 9 is a diagram showing an outline of a processing flow of the present invention.

FIG. 10 is a diagram showing the flow of an embodiment of the present invention.

FIG. 11 is a diagram showing a flow of one embodiment in the present invention.

FIG. 12 is a diagram showing a flow of one embodiment in the present invention.

FIG. 13 is a diagram showing the contents of a buffer during execution of one embodiment of the present invention.

FIG. 14 is a diagram showing the contents of a buffer during execution of one embodiment of the present invention.

FIG. 15 is a diagram showing the contents of a buffer during execution of one embodiment of the present invention.

FIG. 16 is a diagram showing the contents of a buffer during execution of one embodiment of the present invention.

FIG. 17 is a diagram showing the contents of a buffer during execution of one embodiment of the present invention.

FIG. 18 is a diagram showing the contents of a buffer during execution of one embodiment of the present invention.

[Explanation of symbols]

 1 control unit 2 display unit 3 input unit 4 dictionary unit 41 independent word / proper noun dictionary 42 connection matrix table 43 inflectional end table 44 adjunct word table 45 prefix table 46 suffix table 5 co-occurrence relation dictionary unit 51 compound example dictionary 52 Prefix example dictionary 53 Suffix example dictionary 54 Semantic code example dictionary 6 Buffer 61 Input character buffer 62 Converted character buffer 63 Learning buffer 64 Homophone buffer 65 Text buffer 66 Screen buffer 67 Minute position table buffer

Claims (4)

[Claims] 1. Input means for inputting kana characters, an input buffer for storing kana characters input by the input means, kana-kanji conversion means for converting kana characters stored in the input buffer into kanji-kana kanji mixing, In a kana-kanji conversion device having a conversion character buffer for storing conversion candidates converted by the kana-kanji conversion means, conversions stored in the conversion character buffer in response to a kana-kanji conversion instruction from the input means. A dictionary for converting the candidates into kana characters, a combining means for combining the returned kana characters with the kana characters stored in the input buffer, and a kana character combined by the combining means as a conversion target character string; The reference means for referring, and the reference means performs forward matching with the conversion candidates stored in the conversion buffer, and if they match, the A kana-kanji conversion device characterized by using kanji notations as conversion candidates. 2. The kana-kanji conversion device according to claim 1, wherein the reference means refers to a suffix example dictionary. 3. The kana-kanji conversion device according to claim 1, wherein the reference means refers to an independent word / proper noun dictionary. 4. The kana-kanji conversion device according to claim 1, wherein the reference means refers to a prefix example dictionary.