JPH0630100B2 - Kana-Kanji conversion method - Google Patents

Description

TECHNICAL FIELD The present invention relates to a kana-kanji conversion system, and more particularly to a kana-kanji conversion system using a natural language processing technique such as a Japanese word processor.

2. Description of the Related Art Conventionally, in a Japanese processing device such as a Japanese word processor, a kana / kanji mixed with kana and kanji is converted to a kana string input to an input device operated by an operator. In this Kana-Kanji conversion method,
In particular, the problem of processing compound words formed by connecting kanji words and kanji words is becoming more serious.

In particular, as disclosed in Japanese Examined Patent Publication No. Sho 58-4378, there is a means for searching a word dictionary by a combination and determining whether or not words can be combined, and the result of the judgment determines whether words can be combined. There is a method of confirming or unconfirming, and a method of judging whether or not it can be a compound word by testing the strength of connection between words as disclosed in Japanese Patent Laid-Open No. 56-38665.

However, in these methods, the degree of freedom in connecting words is large, especially in homonyms, so it is erroneous that a strict criterion is not set,
It will be a source of erroneous analysis. In practice, there is a limit to how strict the criterion is, and there is a drawback that the rate of erroneous analysis is high and it is not suitable for practical use.

An object of the present invention is to solve the above-mentioned drawbacks of the prior art and to provide a kana-kanji conversion method capable of easily and surely identifying homonyms.

Structure In order to achieve the above object, the present invention extracts a conversion candidate word in units of words by searching a word dictionary storage device for a kana character string input by an input device to obtain kana and kana mixed with kana and kana. In the Kana-Kanji conversion method for converting to
The feature is that the combination of adjacent words is extracted as a compound word, and its reading is tested to identify homonyms.

Hereinafter, specific description will be given based on examples of the present invention.

FIG. 1 is an overall block diagram by function when the present invention is applied to a Japanese word processor. In FIG. 1, a keyboard 1, an input character buffer 2, a kana-kanji conversion control unit 3, a word dictionary storage device 4, a compound word dictionary storage device 5, a part-of-speech utilization table unit 6, a connection weight table unit 7, and an evaluation are shown in FIG. It comprises a container 8, an output character buffer 9, a cathode ray tube (CRT) 10, a document file section 11 and a printer 12.

The keyboard 1 is an example of an input device, and has function keys such as phonetic character keys for inputting phonetic characters such as characters such as hiragana, katakana, alphabet, numbers, and symbols, and conversion keys for instructing kana-kanji conversion. Have. The input character string buffer 2 temporarily stores the input kana character strings input one after another from the keyboard 1 and partially erases the input kana character strings that have undergone the kana-kanji conversion processing.

The kana-kanji conversion control unit 3 fetches data from the input character string buffer 2 and a memory for storing a program of a kana-kanji processing procedure, and based on the data, a word and compound word dictionary storage device 3, which will be described later, A buffer used when retrieving 4 or fetching data, fetching data at the positions of columns (rows) and rows (hangs) from the part-of-speech utilization table 6, or the corresponding connection weight from the connection weight table 7 And a memory for storing the evaluation value calculated from the evaluator 8, the first candidate word (the candidate word having the largest evaluation among the candidate words) and the second candidate word, The work area and other functions are controlled as described later.

The word dictionary storage device 4 is a word (for example, an independent word, an adjunct, an affix, a classifier, etc.) which is a notation character necessary for notifying an input kana character that is a phonetic character. Also, all inflected forms of verbs and adjectives. Is treated as one word.), Its notation, its part of speech, its frequency rank, its output rank learning, the address of the compound word dictionary storage device 5 or the item of the pointer serving as the sequence number.

The compound word dictionary storage device 5 is a compound word that represents one concept by combining a plurality of words stored in the word dictionary storage device 4, and is a dictionary describing the relationship between adjacent words. For example, it is expressed that a compound word is composed of a combination of "super-high speed", "high speed-flying", and "flying-aircraft" which is a combination of two words "supersonic airplane".

Although there are various ways of expressing this dictionary storage device,
The address or sequential number of each word in the word dictionary storage device 4 is stored as a pair so as to form a compound word. Secondly, the compound word dictionary storage device 5 is stored in the pointer column of the word dictionary storage device 4. The corresponding beginning address of the corresponding word is recorded, and the address or sequential number of the subsequent word forming the compound word in the word dictionary storage device 4 is recorded in the compound word dictionary storage device 5. If there are a plurality of subsequent words such as "super high-speed airplane", record them continuously. In that case, the boundary between the preceding word and the succeeding word corresponding to the other preceding word is identified by alternately changing the leading 1 bit between 0 and 1. The address and the sequential number can be generically called address information.

The part-of-speech utilization table unit 6 is a row (column) and a column of a connection weight table of a connection weight table unit 7, which will be described later, with the part-of-speech of the extracted word searched in the word dictionary storage device 4 and the part-of-speech of the word connected before and after the word. It stores an index table for determining the location of (uke). In addition, the synonym system records only the line numbers and column numbers of "bare" and "uke", but the grammatical system also records the endings.

The connection weight table 7 has rows and columns of parts of speech arranged therein, and numerical values indicating the degree of connection between words are arranged in a matrix at the intersections thereof. The row number specified in the part-of-speech utilization table section 6 indicates the connection strength of the word having the part-of-speech at the intersection with the column number. The numerical value among them is 0; connection is not possible. 1; Connected, but very rare.
2; Generally connected. 3; Particularly strong connection. It is set to 4 ranks.

The evaluator 8 can find the word being handled at that position depending on parameters such as the reading length and frequency rank of the word extracted from the word dictionary storage device 4 and the connection weight between words extracted from the connection weight table unit 7. Evaluate the likelihood. As an example of the expression for the evaluation calculation, the word length x 3 + frequency rank + (connection weight) ² = evaluation value is used. The output character string buffer 9 temporarily stores the words evaluated by the evaluator 8 in the order of evaluation values. The words that have been accumulated or fixed are sequentially stored in the fixed order.

The CRT 10 is an example of a display device, and displays an example of the word with the highest evaluation value in the undetermined word string temporarily stored in the output character string buffer 9 to allow the operator to confirm whether or not it is confirmed. Display device for.

The document file unit 11 is a storage device that stores the word strings confirmed and corrected by the CRT 10 in a document format.

The printer 12 is a device for printing out the contents of the document file section 11.

FIG. 2 is a diagram schematically showing a part of the contents of the word dictionary storage device 4, FIG. 3 shows two kinds of embodiments of the contents of the composite dictionary storage device 5, and FIG. According to the first method, the address of the compound word dictionary storage device 5 corresponding to the pointer of the corresponding word is recorded on the left side, and the address of the word dictionary storage device 4 of the corresponding word is recorded in the middle. And
On the right side, the word dictionary storage device 4 of the subsequent words of the corresponding word
The address inside is recorded. FIG. 3 (b) shows the above-mentioned second method, in which a boundary identification flag of 1 bit is recorded on the left side, and the address of the succeeding word in the word dictionary storage device 4 is recorded on the right side.

FIG. 4 is a diagram schematically showing a part of the contents of the part-of-speech utilization table section 6, in which the column numbers and the row numbers are recorded in the “uke” column and the “hang” column.

FIG. 5 is a schematic diagram conceptually showing the connection weight table of the connection weight table unit 7, in which the row indicates the number of the part-of-speech utilization table (row number), and the column indicates the receiving number (column) of the part-of-speech utilization table. Number)
Numerical values of connection weights of four ranks are stored in the matrix part of these numbers.

FIG. 6 is a flowchart showing an example of this embodiment.
Next, an example of creating a kana-kanji conversion sentence of "There are many part-time farmers recently" is explained. It is assumed that the analysis up to "a lot of these days" is progressing, and the "no" at the end of the sentence is cut out as the case particle "no". At this point in time, the line (barrier) number of the connection weight table part 7 is specified from "bare" in the case particle "no" column of the table of the part-of-speech utilization table part 6.

Next, "Kenkyo no Ka ..." is successively input from the keyboard 1 and is temporarily stored in the input character buffer 2. At this time, every time one character is input from the keyboard 1 (S
20), it is judged whether it is a symbol or a character (S21), and if it is a character,
It is judged whether or not n characters have been accumulated (S22). In the case of a symbol, it proceeds to the next process, but in the case of a character, it does not proceed to the next process until a predetermined n characters are buffered in the input character buffer 2. The kana-kanji conversion control unit 3 makes these determinations. When n characters (for example, n = 6) are accumulated in the input character buffer, such as "Kenkyo no ...", "Ken", "Ken", "Kenkyo", and "Kenkyono" A kana character string for word dictionary search, which is a set of character strings of different types, is created by the kana-kanji conversion control unit 3 (S23). The kana-kanji conversion control unit 3 searches the word dictionary storage device 4 according to the reading of those kana character strings for search.
(S24), "hair", "ki", "ticket", "prefecture" ... "side job",
Candidate words are words with written characters such as "school". At that time, the data such as the part of speech, the frequency rank, the output rank, and the pointer of each word shown in FIG. 2 is taken out.

Next, the kana-kanji conversion control unit 3 determines whether or not there is a pointer corresponding to the address of the compound word dictionary storage device 5 (S25). For example, the pointer of the word "inspection" is 65535, which is equivalent to FFFF in hexadecimal notation.
In this case, since there is no address of the compound word dictionary storage device 5, there is no need to search the compound word dictionary storage device 5 and the process proceeds to the evaluation process.

The pointer of the word of the notation "part-time job" is number 7533, which is different from number 65535, so there is a compound word, and the kana-kanji conversion control unit 3 searches for that address in the compound word dictionary storage device 5 (S26). Address 75 of the compound word storage device 5
In the memory location No. 33, as shown in FIG. 3 (a), a compound word by address combination is stored.

One is 11379-17634, and when the word dictionary storage device 4 is searched by this address, it becomes "part-time worker", and its reading becomes "kengyosha". When the reading and the kana character string "kengyonou" in the input character buffer 2 are collated by the kana-kanji conversion control unit 3, there is a clear discrepancy.

Therefore, the next address combination 11379-2311
When the word dictionary storage device 4 is searched based on 5, it becomes "part-time-farmer", and its reading becomes "kengyonoka". When the reading and the kana character string “kengyo no uka” in the input character buffer are collated by the kana-kanji conversion control unit 3, they clearly match.

Since these match, 64 (general noun) which is the part of speech of the word "farmer", 4 which is the frequency rank, output rank 1 and pointer 65535 are taken out from the word dictionary storage device by the kana-kanji conversion control unit 3. Buffered (S28).
The word “farmer” extracted as the compound word candidate is evaluated by the evaluator 8 as a new word, “part-time farmer” as follows. The reading length is 8 which is the sum of both the previous word that is "side job" and the subsequent word that is "farmer", and the frequency rank is the lower of both words, the frequency rank of "side job" is 3
Is. When testing the connection weight with the word "no" before the compound word "part-time farmer". Kana-Kanji conversion controller 3 is "no"
Since it is known that the part-of-speech of the word "side job" that is connected to the part-of-speech is a sahen noun, the column number of the connection weight table is extracted according to the table in the part-of-speech utilization table unit 6 and the connection weight table unit 7 To the connection weight table section 7 in the same way, since the row (barred) number of the case particle "no" has already been transferred to the connection weight table section 7, the connection weight of the connection weight table (in the case of this embodiment) , 2) is taken out by the kana-kanji conversion controller 3.

The evaluator 8 calculated 31 as a result of calculating the evaluation value of “part-time farmer” using the above-mentioned numerical values according to the evaluation formula. The other words, such as the notations "side job" and "school check", were evaluated and calculated in the same way as above, resulting in 22, 20 (S2
9). Therefore, the output character buffer 9 is accumulated in the order of “part-time farmer”, “part-time job”, and “school inspection”.

Here, the word with the highest evaluation value is temporarily determined as the word that displays "part-time farmer", and the part of speech that specifies the line (hang) number uses the general noun of the subsequent word "farmer" and the part-of-speech utilization table. Set the row (take) number from the table of part 6. here,
The kana-kanji conversion controller 3 evaluates the highest priority word, that is,
The maximum evaluation value of the candidate words is integrated, and when the integrated value exceeds a certain threshold, a trigger signal is issued (S40). When this trigger signal is input to the output character buffer 9, the undetermined word string up to the word of the integrated evaluation value is confirmed (S41), and the document file unit 11 stores it in a predetermined storage position.

If the trigger signal is not generated, the process proceeds to the analysis of the next kana character string while it is still undetermined.

The kana character string in the input character buffer 2 is deleted by deleting "Kengyonoka", and in this case, since there is a punctuation mark, it proceeds to create the searched character string even if there are no n characters (S23). ). Similar to the above, the character strings for word dictionary search of "ga", "gai", "gaima", "gaima" are created in the Kana-Kanji conversion controller 3, and according to the flowchart of FIG. Similarly to the above, the kana-kanji conversion controller 3 searches the word dictionary storage device 4, and searches for the case particle "ga", "general",
Words such as "harm" and "the" are given as candidate words.

In the same procedure as above, the case particle "ga" is taken out,
Next, the auxiliary verb "i" is taken out, then the polite auxiliary verb "masu" is taken out and analyzed, and a trigger signal is issued by the punctuation mark, and those conversion words are confirmed, "Recently, many part-time farmers The kana-kanji converted sentences of “gaisare.” Are accumulated in the document file 11. When printing it out, it may be transferred to the printer 12.

Effect According to the present invention, even if there is a homonym, the homonym can be surely discriminated even if it is easy by searching whether the succeeding word forms a compound word, and the discrimination rate is improved. .

[Brief description of drawings]

FIG. 1 is an overall functional block diagram of an embodiment in which the present invention is applied to a Japanese word processor, and FIG. 2 is a diagram schematically showing a part of the contents of the word dictionary storage device of FIG. FIG. 4 is a diagram of each system schematically showing a part of the contents of the compound word dictionary storage device of FIG. 1, and FIG. 4 is a diagram showing a part of the contents of the part-of-speech utilization table part of FIG. Schematic diagram, FIG. 5 is a schematic diagram schematically showing a part of the contents of the connection weight table portion of FIG. 1, and FIG. 6 is a flowchart showing the processing steps of FIG. Explanation of code of main part 3 ... Kana-Kanji conversion control unit 4 ... Word dictionary storage device 5 ... Compound word dictionary storage device 7 ... Connection weight table unit 8 ... Evaluator 9 ... Output character buffer

Claims (1)

[Claims] 1. An input device for inputting a kana character string, a word dictionary storage device storing information on words for writing, and a combination of words stored in the word dictionary storage device. The compound word dictionary storage device and the kana character string input by the input device are searched based on the word dictionary storage device and the compound word dictionary storage device to convert the kana character string into a kanji kana mixed character string. A kana-kanji conversion control unit, the word dictionary storage device, for each word stored therein, discrimination information for discriminating whether or not it is necessary to search the compound word dictionary storage device; Search information for searching the compound word dictionary storage device when it is necessary to search the word dictionary storage device is stored in association with each word, and the compound word dictionary storage device includes the single word Among the combinations, the address information of the word dictionary storage device that stores information of at least the word located after is stored, and the kana-kanji conversion control unit is a kana character string input by the input device. The word dictionary storage device is searched based on to extract a word that is a homonym, and it is determined whether or not it is necessary to search the compound word dictionary storage device for the extracted word based on the determination information. When it is necessary to search the compound word dictionary storage device, the compound word dictionary storage device is searched according to the search information corresponding to the word, and at least after the combination of words stored in the compound word dictionary storage device. The information of the positioned word is obtained from the word dictionary storage device according to the address information held in the compound word dictionary storage device, and the compound word is extracted. Mito kanji conversion method by assaying the match between readings of the input kana character string and performing discrimination of the homonym.