JPH05135110A - Key word development device for information retrieval - Google Patents

Abstract

PURPOSE:To obtain the key word development device for information retrieval which obtains information relative to all key words together with the degree of relation when plural words are inputted as the key words as to a key word development device for information retrieval which inputs words as key words and retrieves and outputs the relative information. CONSTITUTION:This key word development device consists of an input part 1 for inputting plural words, a display part 2 which displays various information, a thesaurus 30 which contains information relative to the input words, and a key word development part 40 which receives the words inputted at the input part 1 as the key words and retrieves the thesaurus 30 by using the key words to obtain the relative information, displays the result at a display part 2, and outputs the result as a key word for information retrieval.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a keyword expanding device for information retrieval in which a word is input as a keyword and relevant information is retrieved and output.

With the spread of databases in recent years, there is a demand for an information search system that enables any user to easily search without having to acquire advanced search skills. Therefore, a thesaurus (means for providing various information related to input keywords (words)) that enables flexible search in response to a user's search request (keyword input) is provided.

[0003]

2. Description of the Related Art FIG. 8 is a conceptual diagram showing the construction of a conventional keyword expanding device for information retrieval. In the figure, 10 is a console including an input unit 1 and a display unit 2. The input unit 1 inputs a word that is a keyword for information search, and the search result is displayed on the display unit 2. Reference numeral 20 is a thesaurus that stores, for each word, information related to the word input from the input unit 1. Reference numeral 21 in the figure is an information unit configured for each word. A required number of these information units 21 are provided. The contents read from the thesaurus 20 are displayed on the display unit 2 in the console 10.

When the operator inputs the word A from the input unit 1 in the apparatus configured as described above, the word A
Is given to the thesaurus 20 as a keyword. The thesaurus 20 retrieves the information unit 21 relating to the word A and displays it on the display unit 2.

FIG. 9 is a diagram showing a configuration example of the information unit 21. Synonym 21a and antonym 21 of the input word
b, an antonym 21c and other information 21d.

[0006]

In the conventional device, when a certain word is input, the related information corresponding to it can be taken out as shown in FIG. However, when a plurality of words such as the word A, the word B, and the word C are input, it is not possible to extract the related information having the input words as keywords.

That is, in the case of the conventional device, as shown in FIG. 10, only the related information for each input word is output. In many cases, the actually required information is a hatched area in which the related information of the word A, the related information of the word B, and the related information of the word C are overlapped with each other as shown in FIG.

Moreover, in the case of the conventional device, it is difficult to extract the strength of the relation between the related words with respect to the input keyword (hereinafter referred to as the degree of association).

The present invention has been made in view of such a problem, and when a plurality of words are input as keywords, information retrieval that can obtain related information related to all of these keywords together with the degree of association It is intended to provide a keyword expansion device in.

[0010]

FIG. 1 is a block diagram showing the principle of the present invention. The same parts as those in FIG. 8 are designated by the same reference numerals. In the figure, 1 is an input unit for inputting a plurality of words, and 2 is a display unit for displaying various information. The input unit 1 and the display unit 2 form a console 10.
However, the input unit 1 and the display unit 2 do not necessarily have to be provided in the form of a console.

Reference numeral 30 denotes a thesaurus that stores information related to an input word, and 40 receives a word input from the input unit 1 as a keyword, searches the thesaurus 30 by using this keyword, and retrieves related information. It is a keyword expansion unit for displaying the result on the display unit 2. It is assumed that the thesaurus 30 is configured to input a plurality of keywords and output related information (related words) common to these keywords. In addition, it is assumed that the degree of association of the output related word can be calculated. The configuration form does not matter.

[0012]

The thesaurus 30 receives a plurality of input keywords, outputs the related words and the degree of association common to these keywords, and gives them to the keyword expanding section 40. The keyword expanding unit 40 outputs the related word and the degree of association received from the thesaurus 30 to the display unit 2 and displays it on the display unit 2. In this way, according to the present invention, when a plurality of words are input as keywords, it is possible to provide a keyword expansion device for information retrieval that can obtain related information related to all of these keywords together with the degree of association. ..

[0013]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 2 is a diagram showing an example of the structure of a thesaurus. The embodiment shown in the figure shows an example in which the thesaurus is realized by a neural network. In the figure, 31
Is a neural network composed of a plurality of units 32, and 33 is an output unit for extracting the output of each unit.

Each unit 32 corresponds to a word and the combination is arranged to indicate the degree of association between these words. That is, the link connecting the units 32 has a weight corresponding to the degree of association. When a keyword is input to the thesaurus, the word unit 32 corresponding to this keyword is initialized. After initialization, the unit weight propagation is calculated and the fired unit is output as a related word of the input keyword. Also, each word unit that has fired has a value, and this value is output as the degree-of-relationship data indicating the strength of the relationship with the keyword.

FIG. 3 is a block diagram showing the configuration of an embodiment of the present invention. The same parts as those in FIG. 1 are designated by the same reference numerals. In the figure, 42 is a database (DB) that stores information such as patent information, 41 is a keyword 42 and a related word output from the keyword expansion unit 40 as input, searches the database 42, and displays the search result on the display unit. Two
It is a search unit to be displayed on. Other configurations are the same as those in FIG. The operation of the apparatus configured as described above will be described below.

When a user inputs a search command from the input unit 1, the input unit 1 analyzes the search command and identifies a keyword. The plurality of keywords identified by the input unit 1 are sent to the keyword expansion unit 40. If the database 42 is searched using only the keyword specified by the user, there is a high possibility that the search may be omitted. Therefore, the keyword expanding unit 40 inputs the related words related to the input keyword to the thesaurus 30 and extracts the related words corresponding to the plurality of keywords input from the thesaurus 30.

The keyword expanding unit 40 gives the retrieved related word and the original keyword to the searching unit 41. The search unit 41 searches the database 42 using these related words and the original keyword as new keywords. Then, the retrieved result is sent from the retrieval unit 41 to the display unit 2 and displayed in an appropriate display format. As described above, by searching the database 42 for the original keyword and the related word related thereto, it is possible to obtain the complete information.

For example, assuming that the database 42 is patent information, and when "semiconductor" and "integrated circuit" are input from the input unit 1, the thesaurus 30 causes the related words related to these input words, such as "transistor" and "FET". , Related words such as “thin film manufacturing equipment” and “diode” are output.
If these related words are added to the original keywords “semiconductor” and “integrated circuit” and the database 42 is searched, complete patent information can be obtained.

The keyword expansion unit 40 can also display the related words output from the thesaurus 30 and the degrees of association on the display unit 2. By looking at the display on the display unit 2, the user can easily proceed with the search. For example, the user can use the related words of the previous keyword when inputting the keyword. By checking the degree of association, the user can know the degree of association of the related word.

According to this embodiment, by using the neural network as the thesaurus, it is possible to quickly and easily obtain the related words related to the plurality of input words. Further, it is possible to provide the user with the degree-of-association data indicating the degree of association, which was difficult to obtain by the conventional technique.

Next, a concrete embodiment will be described.
Now, it is assumed that the keywords "Mikan" and "Winter" as shown in FIG. The command input from the input unit 1 at this time is as shown in FIG. Figure 5
Shows a display example of the display unit 2. The input unit 1 analyzes this command line and extracts a keyword as shown in FIG. The extracted keywords are notified to the keyword expansion unit 40. The keyword expansion unit 40 inputs 2
Each keyword is given to the thesaurus 30. FIG. 6 is a diagram showing how the thesaurus 30 is calculated at this time. 32
Is a unit of neural network. First, "Mikan", which is the word unit 32 for the input keyword.
And "winter" is initialized. After that, when weight propagation is calculated, strong propagation occurs from strong coupling.

In the figure, thick solid lines indicate strong propagation, and thin solid lines indicate weak propagation. For example, "mandarin orange" and "apple" have a strong relationship in the sense of fruit, and "mandarin orange" and "Shizuoka" also have a strong relationship in the sense of their origin. Also, there is a strong relationship with "kotatsu," which is often eaten in the kotatsu. On the other hand, "Mikan"
Has little to do with "cat". The same applies to "winter".

The fired word units (“apple”, “kotatsu”, “Shizuoka”, “cat”) and the values of these units 32 are used as the output of the thesaurus.

The keyword expansion unit 40 displays on the display unit 2 the related words that have been fired by performing neural net-like calculation from the input keyword. FIG. 7 is a diagram showing related words obtained from the thesaurus 30. The number next to the word is a numerical value indicating the degree of association (relevance) with the keyword of the related word. These numerical values are obtained by taking out the value of each word unit 32 from the output unit 33 (see FIG. 2) after the calculation by the neural network.

In FIG. 7, for example, "Mikan" and "Winter"
With respect to the input word, the related word "Kotatsu" having a high degree of relationship and the related words having a relatively weak relationship such as "apple", "Shizuoka", and "cat" are obtained. Since "Kotatsu" is related to both "Mikan" and "Winter", it is highly related, but "Apple" and "Shizuoka" are related only to "Mikan", and "Cat" is only "Winter". The degree of relevance is low because it is related to.

In the above-described embodiment, the case where the neural network is used as the thesaurus has been taken as an example, but the present invention is not limited to this, and common related information is provided to a plurality of input keywords. Any device may be used as long as it has an algorithm capable of calculating with.

[0028]

As described above in detail, according to the present invention, by using a thesaurus that can output common related information common to a plurality of input keywords, a plurality of words can be used as keywords. When input as, the related information related to all of these keywords can be obtained together with the relevance. Further, by using these related information (related words) together with the original keyword in the database search, it is possible to obtain complete information.

[Brief description of drawings]

FIG. 1 is a principle block diagram of the present invention.

FIG. 2 is a diagram showing a configuration example of a thesaurus.

FIG. 3 is a configuration block diagram showing an embodiment of the present invention.

FIG. 4 is a diagram showing an example of input keywords.

FIG. 5 is a diagram showing an example of a display screen.

FIG. 6 is a diagram showing how a thesaurus is calculated.

FIG. 7 is a diagram showing related terms obtained from a thesaurus.

FIG. 8 is a conceptual diagram of a conventional device.

FIG. 9 is a diagram showing a configuration example of an information unit.

FIG. 10 is an operation explanatory diagram of a conventional device.

FIG. 11 is a diagram showing an overlap of related information.

[Explanation of symbols]

 1 input section 2 display section 10 console 30 thesaurus 40 keyword expansion section

Claims (4)

[Claims] 1. An input unit (1) for inputting a plurality of words, a display unit (2) for displaying various information, and a thesaurus (30) for storing information related to the input words.
And receiving the word input from the input unit (1) as a keyword and using the keyword, the thesaurus (3
0) to retrieve related information, display the result on the display unit (2), and output the result as a keyword for information search.
0) and a keyword expansion device for information retrieval composed of 2. A search unit (41) for searching the database (42) based on the keyword + related word output from the keyword expansion unit (40) and displaying the search result on the display unit (2). The keyword expanding device for information retrieval according to claim 1, wherein the keyword expanding device is provided. 3. The thesaurus (30) is realized by using a neural network.
Alternatively, the keyword expansion device in the information search described in 2. 4. The thesaurus (30) inputs a keyword given from a keyword expansion unit (40),
4. The keyword expansion device for information retrieval according to claim 3, wherein the related word related to the keyword and the degree of association indicating the strength of the association are output.