JPS62144269A - Information retrieving device - Google Patents

Abstract

PURPOSE:To reduce the capacity of a memory storing each word by converting an inputted character string into an inscription character string having a high using frequency and using this character string to retrieve said memory. CONSTITUTION:When a character string is inputted, an inscription converting part 2 uses the inputted character string to retrieve a memory 3 via a memory retrieving part 4. If the inputted character string is not detected, the part 2 uses a word reading table 1 which contains the correspondence between the component KANJI (Chinese character) included in the words containing KANJI and the ways of reading to those KANJI to convert the input character string into an inscription character string having high using frequency. Then the memory 3 is retrieved by said converted character string. Thus it is required for the memory 3 just to store only the inscription of a high using frequency for each word. As a result, the capacity of the memory 3 is reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to an information retrieval device in a Japanese language processing system that handles sentences containing kanji and kana.

Conventional technology In the past, in many Japanese language processing systems, Japanese writing was limited to Kana characters or Romaji, but recently, it has become possible to process sentences mixed with Kanji and Kana, which is the normal Japanese writing system. It has become a strong demand. In this case, it is necessary to solve the following problems regarding the dictionary, which is an important component of the search method or system.

In Japanese, it is generally possible to use both kanji and kana to represent a single word, but it is extremely difficult for a system to determine the same representation for each word; I have to enter and acknowledge the multiple notations used in the sentence. Therefore, for words that have multiple representations, it is necessary to construct a dictionary that can be accessed from any representation.

An example of the above-mentioned conventional information retrieval device will be described below with reference to the drawings.

FIG. 6 shows a dictionary of a conventional information retrieval device.

In the dictionary record in Figure 6, "@" or "="
The part before this is the key, that is, the notation of the word. "@"
The part after ``='' contains word-attached information, and the part after ``='' contains word-attached information and is pointer information to the code.

Regarding the dictionary of the information retrieval device configured as above,
The operation will be explained below.

For words that have multiple representations, in order to create a dictionary that can be accessed from any representation, it is necessary to create multiple records with each representation as a key. In this case, word-attached information such as grammar, meaning, translation, etc. is usually written in one of these records, and pointer information to the record in which the word-attached information is written is written in the remaining records. For example, as shown in Figure 5, a total of 1 for 4 words.
If four notations are allowed, records like the one shown in the dictionary in Figure 6 must be created and stored (for example, ``
Nikkei Electronics” No. 324, pages 199-215).

Problems to be Solved by the Invention However, the above configuration has a problem in that as the number of notations increases, the dictionary capacity increases due to the records having pointer information.

The present invention solves the above problems, and for words that have multiple spellings, the grammar and meaning can be determined from any spelling.

The present invention provides an information retrieval device that is capable of retrieving word-attached information such as translations and has a small dictionary capacity.

Means for Solving the Problems In order to solve the above problems, the information retrieval device of the present invention includes:
A memory that stores each word in its frequently used notation, a word reading table that associates constituent kanji in words with their kana notation for words that include kanji, and t!
This system includes a notation conversion unit that converts a notation into a frequently used notation using a reading table, and a memory search unit that searches the memory for frequently used character strings.

Effects With the above-described configuration, the present invention converts a character string input in any notation into a frequently used notation by using a word pronunciation table, and makes it possible to search the memory using that character string. Since the memory only needs to store the expressions of each word that are used frequently, the memory capacity can be reduced.

Embodiment Hereinafter, an information retrieval device according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 shows a configuration diagram of an information retrieval device according to a first embodiment of the present invention. In Figure 1, 1 is a word reading table that associates constituent kanji in words containing kanji with their pronunciations, and 2 is a word reading table that uses word reading table I to convert the notation to a frequently used notation. A notation conversion unit 3 is a memory that stores each word in frequently used notation, and 4 is a memory search unit that searches the memory 3.

The operation of the information retrieval device configured as described above will be described below with reference to FIGS. 1, 2, 3, and 4.

FIGS. 2 and 3 show word continuation tables, and will be used to explain the operation of the notation conversion unit that converts notations into frequently used notations. FIG. 3 shows the word pronunciation table shown in FIG. 2 by extracting the words surrounding the word "shoji", which will be taken as an example below.

Here, it is assumed that the headwords in the word reading table are arranged in a predetermined code order. Now, if the character string ``chenji'' is input, the notation conversion unit first searches for the first character ``hindrance'', and then searches for the character ``hindrance''.

If the character ``Shind'' is found, the next step is to search for the second character ``shi'', but the second character ``shi'' in the input string has a low frequency of use. Since it is not a high character, it is not written as a headword in the word reading table, and the character string "disability" cannot be found. Therefore, the notation conversion unit uses headwords, kana notation, and positions corresponding to the kanji readings to identify the most frequently used character strings of the word ``handicap'' from among headwords that have the character ``handicapped'' as the first character. I'm going to find another notation string. The headwords, kana notation, and kanji reading positions are described below. The kanji reading correspondence position indicates which character in the kana character string each character in the frequently used headword character string corresponds to. In the case of the headword "shoji", the first character of the kana character string corresponding to the first character "shoji" is "shi", which is the first character of the kana character string "shoji".

Next, the first character of the kana character string corresponding to the second character ``ko'' of the headword is ``shi'', which is the 4th character of the kana character string ``shoji''.
It's literal. Therefore, we need to determine which character in the character string in kana notation each character of the headword corresponds to, in this example, "1" and "
4", is the kanji reading corresponding position. From this kanji reading correspondence position, the kana notation of each character of the headword can also be obtained. For example, in the case of the current example, the first character of kana corresponding to the second character of the headword is the fourth character of the kana character string, so the first character of the kana character string corresponding to the first character of the headword is the first character of the kana character string. It can be obtained by taking the third letter from the first letter. Next, in the kana notation corresponding to the second character of the headword, since the headword consists of two characters, the first position in the kana character string of the kana notation corresponding to the third character of the headword is written. Not yet,
In such a case, it can be obtained by extracting the character string after the first character of the kana corresponding to the second character of the headword, in this case the character string after the 4th character of the kana character string. .

By the way, it is natural that the first character of the headword corresponds to the first character of the kana character string, so only the characters after the second character of the headword are stored in the kanji reading corresponding position. , in this example, it can be seen that only "4" is required. Next, a method for obtaining the frequently used character string "shoji" from "shoji" using the corresponding position of the kanji reading will be described. That is, among words whose first character in the headword is ``hindrance'', words whose second character in kana notation is ``ji'' are searched from the positions corresponding to the kana notation and the kanji reading.

For words where the first letter of the headword is "disability", the first character is "disability".
The word ``shogai'' was found, and its kana notation is ``shogai''.
Since the kanji reading corresponding position is "4", the kana notation for the second character of the headword is "kai", which means that it does not correspond to what you are currently searching for. Next, we find the word ``shoji'' and similarly look up the kana notation for the second character, which turns out to be ``shi'', which matches the word we are currently searching for, so we can take it up as a candidate.

Next, a case will be described in which the character string "Shouko" is input. In this case, like last time, we first search for the character string ``Shoko'' that was input as it was.

However, since the character string "Youko" is not used frequently, it is not listed in the headword of the word reading table and cannot be found. Therefore, the notation conversion unit uses headwords, kana notation, and positions corresponding to kanji readings,
I'm going to find a frequently used alternate notation for the character string "Shouko". Their usage is as described above. However, in this case, the procedure is to match with poor spellings, and only those that match are compared with headwords. In this way, the word ``shoji'' can be taken up as a candidate.

As described above, according to this embodiment, by providing a word pronunciation table, for the word "shoji" which has multiple spellings, the word "shoji" that has multiple spellings can be changed to the less frequently used spellings in memory - "shoji", "shiyoushi", and "shoji". ” “Shoji” heading and “Shoji”
Since the area occupied by the pointer to can be removed, the contents of the memory become as shown in FIG. 4, and the memory capacity can be reduced.

Note that the feature of the present invention is that it uses a word reading table to obtain a character string with a different notation, so it can be applied to all devices that have a built-in word reading table and a notation conversion unit that uses the word reading table to change the notation. It is something. Further, the information retrieval device taken up in the first embodiment is one example, and if we consider a memory that has a word pronunciation table as the header section of the memory, in the header section of the memory,
It is also possible to perform a search by performing matching while obtaining a character string with a representation different from that of the input character string.

Effects of the Invention As described above, the present invention has a memory that stores each word in its frequently used notation, and a memory that stores the constituent kanji and their pronunciation in kana for words that include kanji. By providing an associated word pronunciation table, a spelling conversion unit that uses the word pronunciation table to convert spellings into frequently used spellings, and a memory search unit that searches memory for frequently used character strings, it is possible to By using the word pronunciation table, a character string input in such a notation is converted into a character string with a frequently used notation, and the memory uses each word so that the memory can be searched using that character string. Since only frequently occurring expressions need to be memorized, memory capacity can be reduced.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram of an information retrieval device according to a first embodiment of the present invention, FIGS. 2 and 3 are content configuration diagrams showing an example of the word reading table in FIG. 1, and FIG. Fig. 5 is a correspondence diagram showing an example of correspondence between Japanese words and their notation, and Fig. 6 is a content diagram showing an example of records in a dictionary according to the conventional method. FIG. 1...word reading table, 2...notation conversion section, 3...memory, 4...memory search section. Name of agent: Patent attorney Toshio Nakao and one other person Figure 1 Information Figure 2 Figure 3 Figure 4 Figure 5

Claims (1)

[Claims] A memory that stores each word in its frequently used notation, a word reading table that associates the constituent kanji in the word with its reading in kana for words that include kanji, and the word reading table. An information retrieval device comprising: a notation conversion section that converts a notation into a frequently used notation using a notation; and a memory search section that searches the memory for frequently used character strings.