JP3006798B2 - Dictionary creation support device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a natural language processing apparatus for processing human words by a computer, and more particularly to a dictionary used when creating a dictionary for machine processing on a computer. The present invention relates to a creation support device.

[Conventional technology]

Conventionally, when a dictionary of a certain word is created, a different notation different from the notation of the word has been considered and input by the dictionary creator one by one. For example, when creating a dictionary of the word "handle", the only difference is the "handle"
Must be entered in the dictionary, but it is up to the dictionary creator to notice whether or not such a notation is entered. Did not exist.

[Problems to be solved by the invention]

The conventional dictionary creation method described above has a drawback that it is difficult to guarantee the quality of the dictionary in terms of word notation.

When a computer processes a natural language, the computer searches a dictionary of words to be processed using the notation as a clue. That is, word notation, or dictionary headings, plays an important role in the first step of the process for this reason.

If the quality of the dictionary is not guaranteed in terms of notation, the first step will be to stumbling. Since the dictionary search by the computer is performed strictly according to the notation, the dictionary cannot be looked up unless the fluctuation of the notation such as a different notation is registered in the dictionary.

For example, if only "Handle" is registered as a dictionary heading and "Handle" is not registered,
The notation "handle" cannot be processed.

Assuming this, a human dictionary creator must remember in advance various differences in notation, but this is a heavy burden on the dictionary creator.

Different notations such as "handling" and "handling" are not often noticed by human beings, so it is easy to forget to register different notations when creating a dictionary. But,
Then, the quality of the dictionary could not be guaranteed, and this was an obstacle to improving the quality of natural language processing by machines.

[Means for solving the problem]

The device of the present invention is a dictionary creation support device used when creating a dictionary of a natural language such as Japanese or English on a computer, and stores knowledge about rules for synthesizing different notations conceivable from word notations. Heterogeneous knowledge base, and to the notation of the word that the dictionary creator is going to register, by applying the rules stored in the hemimorphic knowledge base, a notation inference unit that estimates the notation of the word, A different notation display device for displaying the estimated different notation.

〔Example〕

 Next, the present invention will be described with reference to the drawings.

 FIG. 1 is a block diagram showing one embodiment of the present invention.

In FIG. 1, a notation input device 1 is a character input device such as a keyboard, a different notation knowledge base 2 is a knowledge base storing knowledge of possible different notations for a word notation, and a different notation inference. The unit 3 receives the notation of the word input from the notation input device 1, searches the different notation knowledge base 2 by using the input as an input, synthesizes a possible different notation with the input notation, and converts the different notation into Output to the different notation display device 4 which is a display device such as a display.

 Next, the operation of the present apparatus will be described in detail using an example.

The present invention can be applied to any language such as Japanese and English.

FIG. 2 is an example of knowledge about a different notation stored in the different notation knowledge base 2.

The left column in FIG. 2 describes the conditions of the notation that may have different notations. For example, the condition on the first line indicates that when any of the word expressions includes any of va, vi, ve, ve, and vo, the notation may have a different notation. Examples that meet this condition are "violin", "viola", "Christmas eve", "Beethoven" and "voice".

In the right column of FIG. 2, rules for synthesizing the different notation by changing the original notation when there is a different notation are described. For example, for the notation that satisfies the conditions on the first line of FIG. It is described that it is a different notation.

“Violin” for “Violin”, “Viola” for “Viola”, “Christmas Eve” for “Christmas Eve”, “Beethoven” for “Beethoven”, “Voice” for “Voice” ] Means that each can be a different notation.

Other examples of knowledge about such notation are:
There is an example of the presence or absence of a ending long syllabary such as "computer" for "computer". When a foreign word is written in katakana, there may be both a notation with a prolonged sign and a notation with no long letter when the sound ends with a letter A. Examples of this are "meter" and "meter", "data" and "data", and the like.

In addition, it is possible to predict the presence of a different notation for a kanji notation. For example, when the notation is “Kanji + Hiragana + Kanji + Hiragana” such as “handle”, the first hiragana sandwiched between the kanji and the kanji tends not to be written. Speaking of the example of "handle", there may be a different notation of "handle".

In this way, knowledge for estimating a possible different notation for the notation from the features of the notation is stored in the different notation knowledge base 2.

FIG. 3 is a flowchart showing the operation of the variant notation inference unit 3 in FIG. Next, the operation of the present invention will be specifically described with reference to this flowchart.

For example, consider a case where the notation of the word “handle” is input through the notation input device 1. Different notation inference part 3
Searches the different notation knowledge base 2 to see if there is a different notation for the notation "handle".

As shown in FIG. 2, "kanji + hiragana + kanji + hiragana"
It can be understood that the notation having the configuration of can have a different notation. Under this condition, it can be seen from the different notation knowledge base 2 that the possible different notation is a notation in which the first hiragana sandwiched between the kanji characters is removed.

Therefore, the different notation inference unit 3 operates the characters in the notation “handle”, removes the hiragana “ri” in the notation, and synthesizes the character string “handle”. This string may be a variant of the first notation "handle". The different notation synthesized by the different notation inference unit 3 is sent to the different notation display device 4 and displayed.

The dictionary creator is trying to register the word "handle" in the dictionary. At this time, if the dictionary creation support device of the present invention is used, there is a possibility that the word "handle" is different from the notation "handle". You will be alerted to the presence of such variants. Then, when the indicated "handle" is actually recognized as a different notation of the original notation "handle", it is registered as a different notation in the dictionary.

〔The invention's effect〕

As described above, in the present invention, when a dictionary of words is created, the different notation is estimated and displayed from the word notation. It is possible to greatly reduce forgetting to input a different notation due to the presence or absence of a prolonged symbol such as "computer".

In this way, by alerting the dictionary entry person about the presence of the different notation, the quality of the dictionary can be improved in terms of notation. As a result, there is an effect that the quality of the whole natural language processing by the machine can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a diagram showing an example of knowledge between different notations stored in a different notation knowledge base 2 shown in FIG. FIG. 4 is a flowchart showing the operation of the variant notation inference unit 3 shown in FIG. 1 ... notation input device, 2 ... different notation knowledge base, 3 ...
Different notation inference unit, 4... Different notation display device.

Claims (1)

(57) [Claims] 1. A dictionary creation support device used when creating a dictionary of a natural language such as Japanese or English on a computer, wherein knowledge about rules for synthesizing different notations considered from word notations is stored. A variant notation knowledge base, and a variant notation inference unit for applying the rules stored in the variant notation knowledge base to the notation of the word that the dictionary creator intends to register, and estimating the variant notation of the word; And a different notation display device for displaying the different notation.