JPH03116375A - Information retriever - Google Patents

Abstract

PURPOSE:To reduce the capacity of a memory and to enable the flexible matching of a keyword by dividing the keyword into simple words which constitute the keyword, and holding all the simple words as the keyword when the keyword, which is a composite word, is applied. CONSTITUTION:When the keyword is decided as the composite word by a composite word deciding means 7, all the simple words of the composite word divided by a composite word dividing means 8 are respectively applied as the keyword by a keyword applying means 2 and held into a keyword holding means 3. Thus, since a user designates a suitable retrieving condition, the flexible matching of the keyword can be executed and desired information retrieval can be executed. Then, it is enough to register only the number of simple words constituting the composite word, and the capacity of the memory can be reduced.

Description

[Detailed Description of the Invention] Industrial Field of Application The present invention stores information such as documents, images, and even audio in association with free keywords.
The present invention relates to an information retrieval device that retrieves information corresponding to a keyword input.

Conventional technology When searching for desired information from a large amount of information, keywords are assigned to each piece of information and stored, and when searching, input a conditional expression that includes the keyword, and then search for information that has the keyword that matches it. Output is -
It is within the membrane. This method can be divided into two types depending on how keywords are given. One is called the controlled keyword method, which limits the keywords that can be used. The other method is called the free keyword method, which does not limit the keywords that can be used. Both have their advantages
Each method has its own shortcomings, but the free keyword method has the problem of matching the input keyword with the keyword attached to the information.

For example, consider a case where the keyword "published patent information" is attached to information and the user inputs the keyword "published patent". In this case, it cannot be determined that there is a match simply by checking whether the notations of the two keywords match. In order to solve this problem, conventional methods have been used, for example, [Automatic extraction of keywords in newspaper article databases] (Information Management Vol. 1.321h4.
There is a method shown in July 1989).

In this method, first, the keywords to be registered are simple words (
Create new words by combining multiple adjacent simple words, and register all of them. For example, if a certain keyword consists of four simple words A, B, C, and D, the words to be registered are ABCD (that is, the keyword itself).

ABC, BCD, AB, BC, CD, A, B.

There will be 10 pieces of C and D. Specifically, when using "published patent information" as a keyword, "published patent information""publishedpatent"
Register six words: "patent information,""publication,""patent," and "information." By doing this, even if the user specifies "published patents" as a search condition, matching can be achieved.

Problem to be Solved by the Invention However, in this method, when registering a compound word consisting of n simple words as a keyword, n(n+1)72
It is necessary to register words, and the memory area required to hold the keywords becomes large.

Further, for example, when the keyword "noise prevention ordinance" is registered, even if the keyword "noise ordinance" is searched, it is determined that there is no match, and the desired information cannot be searched and obtained. This is because only combinations of adjacent simple words that make up compound words are registered (registered are "noise prevention ordinance" and "noise prevention").
The six words are ``prevention ordinance'', ``noise'', ``prevention'', and ``ordinance'').

In order to solve these two problems, it is possible to register all combinations of simple words that make up a compound word.

For example, in the case of "Noise Prevention Ordinance," in addition to the six words listed above, the word "Noise Ordinance" is also registered.

However, with this, when registering a compound word consisting of n simple words, it is necessary to register (2°-1) words, and a much larger memory area is required to hold the keyword than in the conventional example above. It's not realistic.

Means for solving the problem: An information storage means that stores the information to be searched; a word dictionary that outputs the grammatical attributes of the word using the notation of the word as a key; compound word determining means for determining whether a compound word is a simple word; compound word dividing means for dividing a compound word into individual simple words; keyword assigning means for assigning a keyword to the information stored in the information storage means; Keyword holding means for storing sets of keywords and corresponding information; search condition input means for inputting search conditions such as combinations of keywords and operators; and information for searching for information having keywords matching the search conditions. When a keyword is determined to be a compound word by the compound word determining means, the keyword assigning means assigns each simple word of the compound word divided by the compound word dividing means as a keyword, and the keyword retaining means I made it to be stored in .

When assigning a keyword that is an action compound word, by dividing the keyword into its component simple words and storing all the simple words as keywords, you can register only the number of simple words that make up the compound word. It takes up less space and allows for flexible keyword matching depending on the search conditions.

Example An embodiment of the present invention will be described based on the drawings.

First, an information storage means 1 is provided that stores information to be searched, such as documents, images, and audio. As this information storage means 1, a hard disk or an optical disk is generally used to store a large amount of information.

Next, a keyword assigning means 2 is provided for assigning a keyword expressing the content of the information to the information stored in the information storing means 1.

Generally, a plurality of keywords are assigned to one piece of information, and the keywords are usually assigned by a human using a keyboard. However, if the information stored in the information storage means 1 is a document, it is also possible to mechanically (automatically) assign keywords. In this case,
For example, a sentence is divided into words, words called unused words that are difficult to be used as keywords are removed, and the remaining words are used as keywords.

Further, a keyword storage means 3 is provided that stores a set of a given keyword and information to which the keyword corresponds. Here, it is common for a single keyword to correspond to a plurality of pieces of information.

Furthermore, the keywords are arranged according to their character codes, so that when a certain character string is given, it can be determined quickly whether or not that character string is held as a keyword.

On the other hand, search condition input means 4 is provided for specifying the content of information to be searched for by a combination of a keyword and an operator. The following operators are available:

A and B: Information with both the keywords A and 8 AorB: Information with at least one of the keywords A and B not A: Information operators without the keyword A are: They can also be used in combination, such as ((A and B) or C).

Next, an information retrieval means 5 is provided which interprets the search conditions input by the search condition input means 4, searches the keyword holding means 3, and extracts information matching the conditions as a search result.

Further, a word dictionary 6 is provided that collects words that can be used as keywords. This word dictionary 6 uses the notation of a word as a key to extract the grammatical attributes of the word, in this case the part of speech.As illustrated in FIG. 2, a pair of notation and part of speech is stored for each word. ing. There are three types of parts of speech: nouns, prefixes, and suffixes. Prefixes are never used alone; they are always followed by a noun or a prefix. Suffixes are also never used alone; they always follow a noun or suffix. Words are arranged according to their written character codes, and you can check whether a word is registered in the word dictionary or not.
If it is registered, its part of speech can be determined quickly.

Further, a compound word determining means 7 is provided which determines whether a certain keyword assigned by the keyword assigning means 2 is a compound word or a simple word consisting of only one word. This determination is performed in the following steps.

■ If the keyword consists of characters other than kanji (katakana, alphabets, etc.), it is not a compound word.

■ If the keyword is 2 characters or less, it is not a compound word.

■ If the entire keyword is included in the word dictionary 6, it is not a compound word.

■ If none of the above apply, it is a compound word.

Compound word dividing means 8 is provided for dividing the compound word into individual simple words when the compound word is determined to be a compound word by the compound word determining means 7. A simple word refers to a word level word recorded in the word dictionary 6. As a restriction on dividing compound words, ■ The beginning of a compound word must be a noun or a prefix.

■ Compound words end with a noun or a suffix.

■ A suffix does not immediately follow a prefix.

There is a rule. If there are multiple division patterns, the division pattern with the least number of constituent words is adopted.

In such a system configuration, the process of assigning keywords, which is a feature of this embodiment, will be explained with reference to the flowchart of FIG. 3. Note that in this embodiment, the keyword is input by a human using a keyboard. First, if the keyword consists of characters other than kanji (katakana, alphabets, etc.), register it as is. Furthermore, if the keyword is two characters or less, it is registered as is, and if the entire keyword is included in the word dictionary 6, it is registered as is. If none of the above applies, the keyword is divided into individual simple words by the compound word dividing means 8, and only the individual simple words are registered. For example, if the keyword "Noise Prevention Ordinance" is input and the word dictionary 6 is configured as shown in FIG. In other words, when assigning a keyword that is a compound word, you only need to register the number of simple words that make up the keyword, and the memory area required to hold the keyword is small. Since the keywords are stored in the form of simple words, flexible matching is possible depending on the search conditions.

Next, the process for inputting search conditions will be explained. In this case as well, the search conditions are entered by a human using the keyboard. The processing shown in the flowchart of FIG. 4 is performed for each keyword included in the search conditions. First, if the keyword consists of characters other than kanji (katakana, alphabet, etc.), do nothing (leave it as is). Also, nothing is done if the keyword is two characters or less, and nothing is done if the entire keyword is included in the word dictionary 6. If none of the above applies, the keyword is divided into individual simple words by the compound word dividing means 8, and the individual simple words are replaced with a form connected with the operator and. In other words, even if a compound word is included in the keywords that appear in the search conditions, the simple word can be used with operator a.
By simply converting to a form connected by nd, there is no need to perform special processing for compound word searches, and fast and flexible matching can be performed.

For example, if the search condition (``Noise Ordinance J or r double glazing'') is input and the word dictionary 6 is configured as shown in Figure 2, the search condition is ((``Noise J and r Ordinance J)''. orr double window window'') and passed to the information retrieval means 5.

As a result, information with both the keywords "noise" and "ordinance" and information with the keyword "double glazing J" are obtained.

In this way, according to this embodiment, as described above,
Even if the information has been given the keyword "Noise Prevention Ordinance" by a human, it can be searched using the keyword "Noise Ordinance."

Effects of the Invention The present invention is configured as described above, and when a keyword is determined to be a compound word by the compound word determining means, all the simple words of the compound word divided by the compound word dividing means are individually divided by the keyword assigning means. Since the keywords are assigned as keywords and stored in the keyword storage means, users can specify search conditions as appropriate to perform flexible keyword matching, making it possible to search for desired information. It is only necessary to register the number of simple words that make up , and the memory capacity is small.

[Brief explanation of drawings]

The drawings show one embodiment of the present invention, and FIG. 1 is a block diagram;
FIG. 2 is a block diagram of a word dictionary, FIG. 3 is a flowchart showing keyword assignment processing, and FIG. 4 is a flowchart showing search condition input processing. l...information storage means, 2...keyword assigning means,
3... Keyword holding means, 4... Search condition input means, 5... Information retrieval means, 6... Word dictionary, 7...
・Compound word determination means, 8...Compound word division means Applicant Rico Co., Ltd.

Claims (1)

[Claims] An information storage means that stores information to be searched, a word dictionary that outputs the grammatical attributes of the word using the notation of the word as a key, and this word dictionary is referenced to determine whether a keyword is a compound word or a simple word. a compound word determining means, a compound word dividing means for dividing a compound word into individual simple words, a keyword assigning means for assigning a keyword to the information stored in the information storage means, and a set of information corresponding to the keyword. It consists of a keyword storage means for storing keywords, a search condition input means for inputting search conditions such as a combination of keywords and operators, and an information search means for searching for information having keywords that match the search conditions. When the compound word is determined to be a compound word by the compound word determining means, all the simple words of the compound word divided by the compound word dividing means are assigned as keywords by the keyword assigning means and stored in the keyword holding means. An information retrieval device characterized by: