JPH0619970A - Key word extracting system - Google Patents

Abstract

PURPOSE:To fairly and efficiently extract key words. CONSTITUTION:An input means 1 converts a type part except graphics and charts, etc., to a readable form by utilizing an OCR or the like. A data extraction part 12 reads data from the input means 1 and outputs a key word candidate from a dictionary provided beforehand and the predetermined kinds of stop words. A data maintenance part 13 retreives a data holding part 5 relating to the extracted key word candidate, newly stores the key word candidate in the data holding part 5 when it is not present on the data holding part 5 and increases the appearance number of time of counter of the key word candidate by '1' when it is present. An output part 14 outputs the key word candidate for which the value of the appearence number of times of counter of the extracted key word is more than a predetermined value as the key word.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a keyword extraction method by full text search in an information retrieval system.

[0002]

2. Description of the Related Art A conventional keyword extraction method in an information retrieval system has been individually performed by a document creator or a third party expert.

[0003]

The above-described conventional keyword extraction method relies on the sensible judgments of the creator of the document and the experts of the third party, and the important matters that the person wants to state most Although I can suppress it, there was a problem that I could not pay attention to other related matters.

If the target document is lacking in specialized knowledge, it is difficult to completely understand the contents, and some keywords to be extracted may be overlooked.
There was the problem of spending time.

[0005]

A first invention is an OCR in a keyword extraction method in an information retrieval system.
Input means for converting the type part except for figures and tables into a readable format by using the above, and reading keyword data from the input means and outputting keyword candidates from a dictionary and a stop word of a predetermined type. And a data holding unit that is provided in advance for the extracted keyword candidate, and if the data holding unit does not exist, the keyword candidate is newly stored in the data holding unit, and if there is, the keyword candidate. 1 occurrence counter
It is characterized by further comprising: a data maintenance unit for increasing the number; and an output unit for outputting, as a keyword, the keyword candidate in which the value of the appearance frequency counter of the extracted keyword candidate is a predetermined value or more.

[0006]

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

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 2 is a diagram showing a flow of operation of the data holding unit 5 in this embodiment.

The input means 1 has a function of using OCR or the like to convert a type part except for figures and tables into a readable format and a function of updating the stop word accumulated in the input data decoding section 3. Have.

The data extraction unit 12 reads data from the input unit 1 by the read instruction unit 2 and recognizes words by the dictionary function in the input data decoding unit 3 until detecting stop words such as particles, articles, and punctuation marks. Continue reading.

The data maintenance unit 13 compares the extracted word with the data stored in the data holding unit 5 by the comparison operation unit 4, newly stores it in the data holding unit 5 if it does not exist, and appears it if it exists. Increment the frequency counter by 1.

The output unit 14 classifies (weights) the data extracted by the registration unit 9 according to the frequency of appearance,
A keyword is selected and output to the output means 10. The control unit 6 executes the program stored in the memory 8 via the memory control unit 7, and controls the entire system.

Next, the operation of this embodiment will be described with reference to FIG.

When the read instruction unit 2 reads out one character from the beginning of the sentence, "Maigo no", the input data decoding unit 3 detects the word "Mago" and the stop word "No". The comparison operation unit 4 confirms whether or not the word “maigo” is already stored in the data holding unit 5. If it is already registered, the counter is incremented by one, and if it is not registered, it is stored in the data holding unit 5. In this example, if the word "maigo" is not stored yet, it is stored in the data holding unit 5 and the value of the counter is set to 1 (step 1).

When the above operation is repeated and the end of the document is reached (step 4), a word having a weight of a predetermined value or more is designated by the registration unit 9 as a keyword and output to the output means 10.

[0015]

As described above, according to the present invention, keywords can be automatically extracted on the basis of a predetermined standard, so that keywords can be extracted fairly efficiently.

[Brief description of drawings]

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

FIG. 2 is a diagram showing a flow of operations of a data holding unit in the present embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input means 2 Read instruction section 3 Input data decoding section 4 Comparison calculation section 5 Data holding section 6 Control section 7 Memory control section 8 Memory 9 Registration section 10 Output means 12 Data extraction section 13 Data maintenance section 14 Output section

Claims (1)

[Claims] 1. A keyword extraction method in an information retrieval system, comprising input means for converting a type part except for figures and tables into a readable format by using OCR and the like, and reading data from the input means and preliminarily provided. A data extraction unit that outputs a keyword candidate from a dictionary and a predetermined type of stopword, and a data holding unit that is provided in advance with respect to the extracted keyword candidate are searched, and if the data holding unit does not exist, the keyword candidate is found. A data maintenance unit that increases the appearance count counter of the keyword candidate by 1 if it is newly stored in the data holding unit and a value of the appearance count counter of the extracted keyword candidate is a predetermined value or more. A method for extracting a keyword, characterized by having an output means for outputting a keyword candidate as a keyword .