JPS63311432A - Retrieving system for associative word - Google Patents

Abstract

PURPOSE:To obtain a proper retrieving key word by retrieving the associative words obtained from combinations of plural input words and furthermore repeating the retrieving actions with addition of those retrieved words. CONSTITUTION:Plural words are supplied from an input part 1 as key words. Many pairs of word strings containing plural words and the associative words obtained from said word strings are listed up in an associative word dictionary 3. A retrieving part 2 retrieves a word string which is equal to those words (word groups) supplied from the part 1 or included in the word groups out of those word strings stored in the dictionary 3 based on a rule designated previously. Furthermore the associative words of the relevant word string are added to the word group and then a word string which is equal to the word group added with said associative words or included in the word group is retrieved again. These actions are repeated until the end conditions designated previously are obtained. Thus the retrieved associative words are all delivered to an output part 4.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to an associative word search that searches for associative words from a thesaurus using one or more keywords in the field of information retrieval, in which keywords are assigned to documents and in document searches using keywords. Regarding the system.

[Conventional technology]

Conventionally, when assigning keywords to documents or searching documents using keywords, a thesaurus consisting of a large number of terms required for the information search and several relationships between these terms is used to search for a certain keyword. I was looking for suitable keywords related to.

The relationship between terms in a conventional thesaurus includes a broader term (BT), a narrower term (NT), a related term (BT), a preferred term, etc., as shown in Table 1 below.

Table 1 [Problems to be Solved by the Invention] When assigning keywords using the conventional thesaurus mentioned above or performing a literature search using keywords, the word is related to a word that has a related word in the thesaurus, so 2.
Words used in fields where two or more fields intersect may not be listed in the thesaurus as related words, making it difficult to find accurate keywords.

[Means for solving problems]

The associative word search system of the present invention stores a plurality of pairs of associative words, which are a word string consisting of a plurality of words and one or more words associated with the word string, and searches the word string from the associative word dictionary. , a new word group is created by adding the associated words of the searched word string to the original word group input from the input section, and a word string that is the same as this new word group or is included in this new word group. The method is characterized in that the search is repeated, such as searching for from the associative word dictionary.

〔Example〕

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

FIG. 1 is a block diagram showing the overall configuration of the present invention.

A plurality of words are input as keywords from the input unit 1. The associative word dictionary 3 stores a large number of pairs consisting of a word string consisting of a plurality of words and one or more words (associative words) associated with the word string. The search section 2 searches word strings in the word association dictionary 3 that are the same as a plurality of words (word group) input from the input section 1 or that are included in the word group according to prespecified rules. Search, add the associated word of the word string to the word group, repeat the search for a word string that is the same as the word group to which the associated word was added, or is included in the word group, and then meet the pre-specified termination condition. When this happens, end this repetition. All the associated words searched by the search unit 2 are output to the output unit 4.

FIG. 2 is a flowchart of an associated word search. When a word group is input from the input unit 1 (step 21) 1, the search unit 2
searches the word association dictionary 3 for a word string that is the same as or included in the inputted word group according to a predetermined rule (step 22). If an associated word is retrieved, a prespecified termination condition is not satisfied, otherwise the retrieved associated word is added to the word group (steps 23 to 25) and the associated word search is performed again. If no associated words are retrieved, or if the termination condition is met, all the retrieved associated words are output from the output unit 4 and the process ends (steps 23, 24, 26).

Table 2 shown below shows examples of associative crystals corresponding to word strings stored in the associative word dictionary 3.

Table 2 Next, an associative word search using the associative word dictionary 3 having the contents shown in Table 2 will be explained using a specific example.

A user of this system who assigns keywords to documents reads the abstract and text of a document with the title ``Circuit organization system from schematic functional diagrams'' and uses a thesaurus to identify 1 logic circuit I1% artificial intelligence I.

1. It is assumed that the word ``graphical recognition l9% computer-aided system l'' is selected as a keyword. Here, these four words (word group) are input from the input section 1 of this associative word search system.
is input (step 21).

The search unit 2 searches the word strings in the word association dictionary 3 for word strings that are the same as the word group or included in the word group according to prespecified rules (step 22). This search rule is, for example, to search in the order in which they are stored in the associative word dictionary 3, or in order of the number of words included in the word string. In this case, since the word strings of note logic circuit I and demon artificial intelligence l are included in the word group, the association m Azuma Design I has been retrieved (step 23). If the prespecified termination condition is not met at this point (step 24), the retrieved associated word 1 design l is newly added to the word group (step 25).

The termination condition is, for example, that the number of searched associated words has reached five.

When a new associated word is added to a word group, the search for associated words of this word group is repeated. If there is no associated word to be searched or if the termination condition is met, the searched associative word design is output from the output unit 4! 2% CAD# etc. are all output (step 26). The user selects an associated word that is considered appropriate as a keyword from among them.

In addition, users who search for literature using keywords can also use the thesaurus to select the keywords to use for their search.
This can be done using this associative word search system.

〔Effect of the invention〕

As explained above, the present invention allows a conventional thesaurus to be used by inputting multiple words, searching for associated words associated with the combination of words, and repeating the search by adding the searched associated words. It is possible to find words that represent important concepts in documents that may otherwise be overlooked, and is effective in assigning appropriate keywords or finding appropriate search keywords.

In addition, associative word dictionaries are made up of simple elements: a string of words and their associated words, so it is easy to update, delete, and add content, and it is possible to quickly respond to new words and concepts. can.

[Brief explanation of the drawing]

FIG. 1 and FIG. 2 are a block diagram and a flowchart, respectively, of an embodiment of the present invention. 1... Input section, 2... Search section, 3...
...Associative word dictionary, 4...Output section. 7 Piece Agent Patent Attorney Original Letter -〇-2-

Claims (1)

[Claims] An associative error dictionary that stores a plurality of pairs of associative words that are a word string consisting of a plurality of words and one or more words associated from the word string, and a word group that is a plurality of words input from the input section. Search the word association dictionary for a word string that is the same as or included in this word group, and add the word association of the searched word string to the original word group input from the input section to create a new word group. An associative word search system characterized by repeating a search such as searching the associative word dictionary for a word string that is the same as a new word group or included in this new word group.