JPH0782492B2 - Kana-Kanji conversion method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial Application Field> The present invention relates to kana-kanji conversion used in a Japanese word processor or the like, and more particularly to improvement of a homophone identification method.

<Prior Art> In the same language selection of a Japanese word processor, the co-occurrence relation between adjacent clauses has been described as shown in Japanese Patent Laid-Open Nos. 59-2125 and 59-109939. There was a kana-kanji conversion method using conversion tables.

For example, by registering “tax + pay” or “country + govern” in advance in the co-occurrence relation dictionary such as “pay tax” or “control the country”, the first candidate for kana-kanji conversion It is used for the decision.

<Problems to be Solved by the Invention> However, since the co-occurrence relation dictionary is searched without distinguishing the co-occurrence relation between adjacent clauses and non-adjacent clauses, there is a problem that the search efficiency is low.

The present invention has been made in view of the above problems, and in a conventional co-occurrence relation dictionary, in a conversion table that stores the co-occurrence relation between adjacent clauses, effective co-occurrence relations are distinguished only between adjacent clauses. By doing so, the purpose is to eliminate the application between non-adjacent bunsetsu and improve the retrieval efficiency.

<Means for Solving the Problem> In order to solve the above-mentioned object, it is characterized by adding information that distinguishes it from other co-occurrence relation data that is effective only between adjacent clauses.

<Operation> By adding clause information to the co-occurrence relation dictionary, the co-occurrence relation data to be searched is limited, for example, the co-occurrence relation is effective only between adjacent clauses.

<Embodiment> FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention. In this figure, 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. 2 is CRT or LCD
And the like, and a reference numeral 3 denotes an input unit such as a keyboard. Reference numeral 4 is an independent word dictionary that stores independent words, and 5 is a co-occurrence relation dictionary that stores co-occurrence relations between phrases. A buffer memory 6 is composed of 61 character buffers, 62 learning buffers and 63 same language buffers. The character buffer 61 is a buffer for storing the characters input from the input unit 3. The learning buffer 62 is a buffer used for the learning process of words, and the same language buffer 63 is a buffer for storing candidates for the same language at the time of kana-kanji conversion.

The kana string input from the input unit 3 is stored in the character buffer 61. Next, when the user gives a conversion instruction, or when the conversion timing comes, the control unit 1 causes the independent word dictionary 4
Also, the kana-kanji conversion is performed by referring to the co-occurrence relation dictionary 5 and other tables, and the conversion result is stored in the buffer 63. The first candidate is displayed on the display unit 2. Candidates other than the first candidate can be displayed on the display unit 2 and selected by operating the input unit 3. When the same language is selected by operating the input unit 3, the result is stored in the learning buffer 62.

FIG. 2 is a flow chart showing the processing operation of kana-kanji conversion after the kana is input.

The following description will be made in accordance with the flow of the processing, using FIG. 1 and FIG.

At 201, the kana string of "Keep the tune to the limit" is taken into the character buffer 61, and at 202, the preceding clause is extracted from the inputted kana string. In this case, "zeikinwo", "zeikin", and "zei" are extracted as preceding clauses. In 203, the phrase starting with the kana sequence following the preceding phrase extracted in 202 is extracted. When the extraction of the preceding and succeeding clauses is completed, next, in step 204, the preceding and succeeding clauses are processed to determine the preceding and succeeding clauses. In this case, the preceding phrase is "Zeiki no" and the following phrase is "Kijitsuni". In 205, it is determined whether the sentence is finished. If it is not finished, the process goes to 203, and if it is finished, the same language is selected in 206. By repeating the process of LOOP1, the process of dividing the input kana sequence into clauses ends. The candidates divided into clauses store the result in the same language buffer 63 for each clause.
The same language buffer 63 has an individual buffer for each clause. After dividing the input kana string into phrases, in the same language selection processing of 206, the same language candidates divided for each phrase are narrowed down. Here, the same language is selected for the three clauses of "Zeiki wo", "Before the start", and "Summer", and the above processing is continued until the end of the sentence (207).

FIG. 3 is an operation flow chart showing the flow of processing for selecting the same language.

In FIG. 3, the example process 301 refers to the co-occurrence relation dictionary 5 to check the likelihood between adjacent clauses. The co-occurrence relation dictionary 5 is a dictionary showing the relation between words, and stores idioms and examples. This is used to narrow down the same language between two clauses in the same language buffer.

FIG. 4 is a diagram showing an example of the example processing, showing a state in which the same language of “Zein” is stored in the preceding phrase same language buffer and the same language of “Osamu” is stored in the subsequent phrase same language buffer. Shows. FIG. 6 shows an example of the co-occurrence relation dictionary used in the example processing.

In FIG. 4, with respect to the input kana sequence “Zeiki wo wo suru”, the same language of ‘zei kin wo’ and ‘sou wa suru” is stored in the same language buffers of the preceding clause and the succeeding clause, respectively. At this time, in the compound word example process, the candidate “tax / pay” is determined by referring to the co-occurrence relation dictionary in FIG. The compound word example process is a process of combining all the words of the same language buffer of the preceding phrase and the words of the same language buffer of the subsequent phrase and checking whether the combination is described in the cooccurrence relation dictionary.

After the same language selection process of 206 is completed, the end of the sentence is checked in 207, and if the sentence is completed, the kana-kanji conversion process is terminated,
If there are still sentences, repeat the Kana-Kanji conversion process.
Start from 1.

Next, a method of eliminating the application of the example processing to non-adjacent phrases by adding information to the co-occurrence relation data that is valid only between adjacent phrases will be described.

This is realized by adding inter-phrase information to the co-occurrence relation dictionary shown in FIG. 6, as indicated by 901 in FIG. In the inter-phrase information in FIG. 5, the co-occurrence relation data with "*" indicates that the data is valid only between adjacent phrases. In other words, other inter-clause information that is not used when investigating the co-occurrence relationship between non-adjacent clauses, such as information about particles following the preceding clause, and information about intervening clauses when co-occurring between non-adjacent clauses. You can also register and use it.

In this way, by adding the distinction information to the co-occurrence relation data and limiting the co-occurrence relation data to be applied, the accuracy of the example processing can be improved.

<Effects of the invention> By preliminarily adding information that distinguishes the co-occurrence relation data that is valid only between adjacent clauses, it is possible to properly select applicable co-occurrence data, improve the accuracy of example processing, and increase the processing speed. Becomes possible.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention,
FIG. 2 is an operation flow chart of the kana-kanji conversion processing after the kana of the embodiment is input, FIG. 3 is an operation flow chart of the same language selection processing of the embodiment, and FIG. FIG. 5 is a schematic diagram showing the structure of the language buffer, FIG. 5 is a schematic diagram of a co-occurrence relation dictionary to which inter-section information is added, and FIG. 6 is a schematic diagram of a co-occurrence relation dictionary. 1 ... control unit, 2 ... display unit, 3 ... input unit, 4 ... independent word dictionary, 5 ... co-occurrence relation dictionary, 6 ... buffer, 61 ...
… Character buffer, 62 …… Learning buffer, 63 …… Same language buffer.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-60-124774 (JP, A) JP-A-61-82273 (JP, A) JP-A-59-2125 (JP, A)

Claims (1)

[Claims] 1. An input unit for inputting characters, an input character string storage unit for storing an input character string, conversion tables for converting the input character string into a Kanji character string, and a character string. In a Japanese processor having a display means for displaying, a co-occurrence relation data that is valid only between adjacent clauses is added with information that distinguishes it from other data in a conversion table that stores the co-occurrence relations between adjacent clauses. A kana-kanji conversion method characterized by eliminating application between non-adjacent bunsetsu by doing so.