JPS5965343A - Kana (japanese syllabary)/kanji (chinese character) converter - Google Patents

Abstract

PURPOSE:To decrease the processing time for KANA-KANJI conversion without deterioration accuracy, by adding the connection checking information between words during retrieval of a compound word, and excluding a word string in which monosyllabic words continue out of the result of retrieval during retrieval of compound words for retrieval of clauses. CONSTITUTION:For instance, ''MIGOTONA-'' is fed as a syllable codes string. Then a clause checking part 7 actuates a word connection deciding part 9, and a syllable code number checking part 8 checks the number of syllable codes with each word to be decided for its connection before the connection is decided. With ''MIGOTONA-'', (MI)1, (MI)2 are both read as (MI), and (GO)1, (GO)2 are both read as (GO), and the number of codes for (MI)1, (MI)2, (GO)1, (GO)2 are all 1, but these characters do not satisfy the constitution for ''MIGOTONA'' as a part of speech. Therefore, the part 9 decides that the connection is possible in case the connection is possible in terms of parts of speech with front and back wards and all words are not monosyllabic. Otherwise, the part 9 decides that the connection is impossible. The retrieved word information is shown in figures 1-3. Based on this information, only ''MIGOTONA'' (KANJI) means a retrieved clause.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a kana-kanji conversion device that converts a code string by manual kana input using kana keys, Roman character input using alphabet keys, or voice input into a character string containing kanji and kana. It is.

Construction of conventional examples and their problems In recent years, various methods have been devised for kana-kanji conversion devices, such as solid input, 5-syllable input, etc., and their development has been remarkable.

A conventional kana-kanji conversion device will be described below with reference to the drawings.

FIG. 1 is a block diagram of a conventional kana-kanji conversion device.

In FIG. 1, 1 is an input syllable code string storage unit that stores input characters or sounds as a syllable code string; 2
3 is a word dictionary section that has part-of-speech information, syllable code strings and the number of codes for each word, and 3 is a word dictionary section 2 that stores words that match the syllable code string from a specified position in the input syllable code string storage section 1 4 is a word connection determination unit that determines the possibility of connection between the searched word and the preceding word on the input syllable code string based on word part-of-speech information; 5 is word connection determination unit 4 A word string information storage section 6 stores words connected to each other that are determined to be acceptable as word string information; Certify,
a clause recognition unit that sets the recognition result in the word string information storage unit 5;
Reference numeral 7 denotes a clause retrieval unit that controls the operations of the above-mentioned units and performs clause retrieval.

The operation of the kana-kanji conversion device configured as described above will be explained below.

As an input example, it is assumed that a syllable code string "Migoto na..." is stored in the input syllable code string storage section 1, and a bunsetsu search is started from the first position of this syllable code string.

First, the word search unit 3 is activated by the phrase search unit 7 and searches for a word that matches a code string starting from the beginning of the input syllable code string. The search results for “amazing adjective verb”
Suppose we have obtained the following expressions: ``masterly,'' ``mi noun fruit,'' and ``mi prefix ending.'' The information is, in order, syllable code, part of speech information, and kanji notation.

Next, the word connection determination unit 4 is activated by the phrase search unit 7 to determine whether the word currently searched for is a word in the preceding position on the input syllable code string stored in the word string information storage unit 5. , determine whether to connect or not. However, at this point, since the word search is from the beginning of the input syllable code string, there are no prefix words, so it is determined whether the above three words can be the beginning of a clause. . The word connection determining unit 4 determines from the part-of-speech information of the three words that any of them can be the beginning of a clause.

Therefore, based on the judgment of the word connection judgment section 4, the phrase search section 7
Each of the three words mentioned above is stored in the word string information storage section 5 as word string information.

Then, the phrase search section 7 activates the phrase recognition section 6 to perform phrase recognition for the three stored word string information categories.

The phrase recognition unit 6 performs phrase recognition by sequentially performing two-part-of-speech processing and adjunct word processing on the three word strings. As a result, out of the three word string information above, "Sweet"
is recognized as a phrase "stunning" based on the success of adjunctive word processing, and the other two are not recognized as phrases, and recognition information is set in the word string information storage unit 5 for each. As a result, the contents of the word string information storage section 5 are approximately as shown in FIG. 2, 761 to 761 to 763.

Thereafter, the phrase search section 7 attempts to convert the input syllable code string into a compound word formed by combining a plurality of words.

First, at this point, the phrase search section 7 checks whether there is word string information in the word string information storage section 6 that has a part of speech that can constitute a compound word, and if so, proceeds with the search for the compound word. , if it does not exist, end the clause search. In the above example, the ``real noun'' and the 1-final prefix shown in Figure 2 cannot constitute a compound word, so the compound word search proceeds.

Here, the phrase search section 7 activates the word search section 3 to search for a word that matches the code string starting from the position of "go" on the input syllable code string. As a result, suppose that ``go prefix error'' and ``go suffix after'' were obtained.

The phrase search unit 7 searches the obtained two words and the word string information in the word string information storage unit 5, ga C2 in FIG.
The word connection determination section 4 is activated to determine whether or not the connection with the word string No. 63 is possible. The word connection determination unit 4 is /162 in FIG.
From the word string /163, whether or not it can be connected to each of the above two words is sequentially determined based on the part-of-speech information of each word. For example, it is possible to change from ``real noun'' to ``post suffix'';
It is not possible to change from "unprefixed word" to "postfixed word". Based on the results of the word connection determination section 4, the phrase search section 7 adds connectable word combinations to the word string information storage section 5. Examples of the added word string information are shown in FIG. 2, 164-6. Furthermore, the phrase search unit 7 activates the phrase recognition unit 6 for the word string information of each added word.

In this way, the phrase search section 7 searches for phrases while sequentially activating the word search section 3, word connection determination section 4, and phrase recognition section 6, and when a word is not found or the searched word is The clause search is terminated when none of the searched word strings can be connected to an already detected word string, or when the searched single nrr has a part of speech that does not constitute a compound word.

In this way, for the example input syllable code string 1 - ``Beautiful'', the phrase search results include ``Beautiful,'' ``Actual Goto Name,'' and ``Migoto Name.''

However, in the above compound word search, since the word-to-word connection test is based only on the part-of-speech information of each word, the search results may include items such as ``Uncorrected House Name'' as explained above. Something useless is appearing. Furthermore, even if it does not appear in the final search result, there is a strong possibility that it will appear as a useless word string during the search.
Both of L and the like cause a significant increase in conversion time.

OBJECTS OF THE INVENTION In view of the problems of the above-mentioned conventional examples, it is an object of the present invention to suppress the appearance of unnecessary word strings and, as a result, reduce the time required for kana-kanji conversion.

Structure of the Invention The present invention uses the number of syllables of a word in addition to the conventional part-of-speech information of the word to check the connection between words during a compound word search. The above objective is achieved by disabling connection between the two.

DESCRIPTION OF EMBODIMENTS An embodiment of the present invention will be described below with reference to FIG.

In FIG. 3, 1 is an input syllable code string storage section, 2 is a word dictionary section, 3 is a word search section, 5 is a word string information storage section, and 6
7 is a phrase recognition section, and 7 is a phrase search section, which is the same as the configuration shown in FIG. 1.

The configuration differs from that shown in FIG. 1 in that the word connection determination unit 4 in FIG. The point is that the word connection determining unit 9 determines the connectability of words based on word part-of-speech information and the number of syllable codes.

Below, we will explain the operation of the kana-kanji conversion device configured as described above, as well as the input example ``Splendid...
・'' will be explained as an input syllable code string.

Note that FIG. 4 is a diagram showing an example of word string information, and shows Jr;
The operations until word string information 1 to 3 are stored are the same as in the conventional example. It is also assumed that the search results for words that match the code string starting from the position of ``go'' on the input syllable code string are ``go prefix error'' and ``go suffix after''.

Now, the phrase search unit 7 starts up the word connection determination unit 9. Prior to connection determination, the word connection determining section 9 causes the syllable code number checking section 8 to check the number of syllable codes for each word to be determined for connection. Regarding this example, “actual” and “not”
The number of syllable codes is 1 for each of , ``mistake'', and ``later''.

Therefore, the jl'i word connection determination section 9 uses the syllable code inspection section 8.
Based on the result, it is determined that the words before and after are connectable in terms of parts of speech and that they are not monosyllabic words, that they are connectable; otherwise, that they are not connectable. In this example 2,
Since both combinations are one-syllable words, it is determined that they cannot be connected.

Therefore, when the phrase search section 7 receives that a word string to be added to the word string information storage section 6 is not found, it uses this as a trigger to end the phrase search. At this time, the only word string information retrieved is FIG. 4/I61-3, and the only phrase retrieved is ``stunning''.

It goes without saying that the input to the kana-kanji conversion apparatus of the above embodiment may be either solid input or segmented input.

Effects of the Invention As described above, according to the present invention, by adding connection check information between words during a compound word search, a word string consisting of successive one-syllable words can be extracted from the search results during a compound word search in a bunsetsu search. Since they are eliminated beforehand, wasteful processing time can be suppressed without reducing the accuracy of kana-kanji conversion, and the conversion speed can be improved. This can be done by reducing the buffer size required for the word string information storage unit during bunsetsu search, and its value is considerable.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a conventional kana-kanji conversion device, and FIG. 2 is a diagram showing the storage state of word string information in a word string information storage unit.
FIG. 3 is a block diagram of a kana-kanji conversion device according to an embodiment of the present invention, and FIG. 4 is a diagram showing the storage state of word string information in the word string information storage section. 1... Input syllable code string storage section, 2... Single flff dictionary section, 3... Word search section, 4...
. . . word connection determination section, 5 . . . word string information storage section, 6 . . . phrase recognition section, 7 . . . phrase search section, 8 . Syllable code number test section, 9...
...Word connection determination section. Name of agent: Patent attorney Toshio Nakao and 1 other person 2nd
figure

Claims (1)

[Claims] an input syllable code string storage unit that encodes and stores input characters and sounds; a word dictionary unit that stores at least part-of-speech information and syllable codes for each word; and a syllable code that matches the input syllable code string. a word search unit that searches the first part of the word dictionary for words with It is equipped with a word connection determination unit that makes a determination based on the number of syllable codes, and a word string information storage unit that stores word string information consisting of words that are determined to be connectable to each other. If the number of syllable codes of the words searched in consecutive positions above are both 1, by disabling the connection,
A kana-kanji conversion device characterized by eliminating consecutive syllabic words.