JPH09128399A - Method and device for extracting keyword data of patent database - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately grasp the gist of an invention and to improve the keyword retrieval efficiency by mechanically extracting keyword data as it is from the range, etc., of a patent demand. SOLUTION: The extraction device 1 is provided with a keyword data storage means 2 where keyword data is previously set and stored, a control means 4 where keyword data set in the keyword data storage means 2 is compared with document information stored in an external storage means 3 and processed and a storage means 5 where a result processed in the control means 4 is temporarily stored. The control means is connected to an input means 6 and an output means 7. First keyword data consisting of KANJI(Chinese character), KATAKANA(square form of Japanese syllabary), signs and Roman latters, etc., is extracted from the document information of patent application such as the range, etc., of the patent demand which is stored in the external storage means 3 and second keyword data consisting of the character strings of HIRAKANA(cursive form of Japanese syllabary) is extracted. Then, third keyword data consisting of specified KANJI is obtained in a non-extraction state.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a keyword data of a patent database for automatically extracting keyword data used in a keyword search from document information of a patent database storing a large number of patent information. The present invention relates to an extraction method and a keyword data extraction device.

[0002]

2. Description of the Related Art Conventionally, as a patent database search system for searching patent information by machine search, there is, for example, "PATOLIS" of Japan Patent Information Organization. In this search system, a patent database is constructed by storing the patent information generated for each patent application as a database file (text file) in the storage means of the computer at any time.

The patent information stored in this patent database includes bibliographic items of a patent application (for example, application number, publication number, publication number, registration number, application date, publication date,
The date of publication, the date of registration, the name of the invention, the inventor, the applicant, the patent classification, etc.) and various keywords (for example, free keywords, fixed keywords, F-terms) that have been added by analyzing the technical content.

A user who uses this search system, for example, searches for a prior art, objection material, invalid material, etc., and a patent having the same keyword as a specific patent application or utility model registration application (subject application). We may search for applications. In this case, the user selects a specific keyword from the keywords given to the target application and inputs it to the terminal device as a search key. The computer of the search system searches for a patent application (corresponding patent) having the same keyword as the search key in the database file based on the input search key, and displays the bibliographical information of the patent, etc. on the terminal display and printer. Is output to.

[0005]

However, this search system has a problem that it is not possible to accurately search the keywords described in the claims, for example, as they are. . That is,
For example, when assigning keywords such as free keywords to a patent application, the keyword assigner analyzes and assigns the entire technical content of each patent application (claims, specifications, etc.), so when assigning keywords There is a case in which a human being is easily involved in a human factor, and for example, a keyword in the claims is regarded as a superordinate concept.

As a result, the words described in the claims are not added as keywords in the state as they are. In particular, as the technical contents of patent applications have become complicated and subdivided in recent years, it is possible to efficiently identify specific keywords only from the description of the claims of the patent application and the title and abstract of the invention (purpose, structure, etc.). Although there is an increasing demand for machine search, the fact is that the above search system cannot meet such a request.

The present invention has been made in view of such circumstances, and an object thereof is to be able to automatically extract keyword data as it is from the scope of claims and the like, thereby improving keyword search efficiency. An object of the present invention is to provide a keyword data extraction method and a keyword data extraction device for a patent database.

[0008]

In order to achieve the above object, a keyword data extracting method for a patent database according to claim 1 is a keyword data extracting keyword data from document information of a patent application stored in an external storage means. In the extraction method, kanji, katakana,
A step of extracting the first keyword data consisting of a code, a roman character, etc., a step of extracting the second keyword data consisting of a hiragana character string from the document information, and a third keyword data consisting of a specific kanji character are set to a non-extracted state. And a step of obtaining.

According to this extraction method, for example, from the document information in the claims, first keyword data consisting of kanji, katakana, code, romaji, etc. is extracted, and secondly consisting of hiragana character strings. Extract keyword data. Then, for example, from the extracted first and second keyword data, the third keyword data consisting of specific kanji that cannot be a keyword of patent information such as "invention", "above", "composition" is deleted. As a result, the keyword data including the phrase, the hiragana character string, the code, etc. described in the claims is extracted as it is,
This is searched by keyword.

According to another aspect of the present invention, there is provided a keyword data extracting device for a patent database, wherein plural types of keyword data are set in the keyword data extracting device for extracting keyword data from document information of a patent application stored in an external storage means. Based on the keyword data storage means to be stored and the keyword data in the keyword data storage means, the first keyword data consisting of kanji, katakana, code, romaji, etc. and the hiragana character string from the document information in the external storage means are stored. And a control means for extracting the second keyword data and for putting the third keyword data consisting of a specific kanji character into the non-extracted state.

According to this keyword data extraction device,
First, a plurality of types of keyword data, such as second keyword data composed of a hiragana character string and third keyword data that cannot be a keyword of patent information, are set and stored in the keyword data storage means. The control means
From the document information stored in the external storage means, kanji,
First keyword data consisting of katakana, codes, etc. and keyword data corresponding to the second keyword data stored in the keyword data storage means are extracted, and for example, the extracted keyword data corresponds to the third keyword data from the inside. Delete keyword data. Thereby, for example, the keyword data in the claims is extracted as it is, and the keyword data is searched.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. FIG. 1 shows a block diagram of a keyword data extraction device according to the present invention. FIG.
1, the keyword data extraction device 1 (hereinafter, simply referred to as the extraction device 1) includes a keyword data storage unit 2 in which keyword data is preset and stored, and keyword data and external storage unit set in the keyword data storage unit 2. 3 has a control means 4 for comparing the document information stored in the storage means 3 and the like, a storage means 5 for temporarily storing a result processed by the control means 4, and the like. An input means 6 and an output means 7 are connected to the control means 4.

The extraction device 1 is composed of a personal computer and has a patent database retrieval device 10 (hereinafter simply referred to as retrieval device 10) as shown in FIG.
Is constructed. That is, the search device 10 includes a personal computer 11 (hereinafter referred to as a personal computer 11), a keyboard 12 as an input unit, a display (CRT) 13 and a printer 14 as an output unit, and a CD-ROM drive as an external storage unit. It has a CD-ROM 15 (external storage device) and the like mounted on the device.

The personal computer 11 is a central processing unit (CP
U) 16, random access memory (RAM) 17, read only memory (ROM) 18, etc.
Various calculations and controls are performed according to the search program stored in 8. The external storage device is the CD-ROM 1
Not limited to 5, a magnetic disk of a patent database or the like is used, and a large number of document information such as the above-mentioned bibliographic items of patent applications and keywords are stored and stored.

Next, the operation of the extraction device 1 will be described with reference to the flowchart of FIG. This flowchart is executed by the search program stored in the memory 18 of the personal computer 11. First, two kinds of keyword data (hereinafter, this keyword data is referred to as second and third keyword data) are stored in the memory 18 of the personal computer 11.

As shown in FIGS. 4 and 5, this keyword data is stored in the table 18 and the table B on the memory 18.
The second keyword data of table A is composed of a hiragana character string that has not been converted to kanji or the corresponding kanji does not exist, and the third keyword data of table B cannot be used as patent information. It is composed of.

Then, the program starts with the table A and the table B stored (S100).
Then, in a part of the text file stored in the CD-ROM 15, that is, in the document information of the first patent application, for example, the title of the invention / invention, the abstract (purpose, configuration)
Then, the claims (the utility model registration request in the case of the utility model registration application) are read as file data in the state shown in FIG. 6 (S101).

When this file data is read, the CPU 16 of the personal computer 11 determines whether or not there is the first keyword data in the file data (S102).
Is done. The first keyword data consists of Kanji, Katakana, codes such as numbers and symbols, Roman letters, etc. If there is the first keyword data, it becomes "YES" at step 102, and these keyword data are as shown in FIG. It is extracted (S103) and temporarily stored in the memory 17.

Next, it is judged whether or not there is the second keyword data in the file data (S104). This judgment is made by comparing the data with the table A of the second keyword data stored in advance in the memory 18, and if there is the second keyword data, it becomes “YES” in step 104, and FIG. A character string composed of a specific hiragana as shown (in the embodiment, a "screw" with a dotted underline) is extracted (S105) and temporarily stored in the memory 17.

If there is no first keyword data in step 102, "NO" in step 102,
Jump to step 104. Also, step 104
If there is no second keyword data in step 104,
Becomes "NO", and the process jumps to step 106 described later.

When the first and second keyword data are extracted and temporarily stored in the memory 17, it is judged whether or not there is the third keyword data (S106). This determination is performed by comparing the table B stored in advance in the memory 18 with the extracted keyword data stored in the memory 17, as in step 104. If there is the third keyword data, Step 10
When the answer is 6, YES, the third keyword data is erased from the memory 17 (S107). This step 10
7. For example, in FIG. 8, the third keyword data such as underlined “patent claim” or “previous” is deleted by 7, and the memory 17 stores the keyword data as shown in FIG. become.

When the third keyword data is erased, it is judged whether or not there is a dull keyword data (S108).
If "YES" in this determination, all the same keyword data located in the rear of the sloppy keyword data are erased (S109). In the embodiment of FIG. 9, the underlined "energy absorbing plate 1" and "column bracket" are eliminated. As a result, the keyword data is extracted from the file data of the first patent application (name of invention, abstract and claims) as shown in FIG. 10, and the program ends (S11).
0).

Then, with respect to the file data relating to the second patent application, steps 101 to 109 are similarly repeated, and the keyword data is extracted and stored in the memory 17.
The keyword search processing is performed by the search device 10 based on each of these keyword data. As this search processing, for example, Japanese Patent Application Laid-Open No. 5-1351
As disclosed in Japanese Patent Publication No. 09, the method is performed by calculating the similarity between the keyword of the target patent and the relevant patent. At the time of this search, the extracted keyword data is mechanically extracted with no human element involved, so that the keywords such as the claims are searched as they are.

In the above flow chart, the first keyword data may be extracted after the extraction of the second keyword data, or the third keyword data may not be extracted from the beginning, and then the first keyword data may be extracted. Alternatively, the second keyword data may be extracted.

As described above, according to the above-described embodiment, the keyword data in the title, abstract and claims of the invention can be extracted as it is, and no artificial element is involved in the extraction. For example, it is possible to perform a keyword search based on the scope of claims, which is the scope of rights, and it is possible to perform a machine search that accurately grasps the gist of the invention. As a result, complicated and subdivided patent information can be searched with high efficiency and high accuracy, and keyword search efficiency can be improved, and for example, patent maps relating to keywords and patent maps such as claim maps can be created. It can be done easily.

Further, by setting three kinds of keyword data variously, the keyword data can be extracted according to the technical level of the invention, the retrieval accuracy, etc., and the extraction device 1 and the retrieval device 10 having high versatility can be obtained. You can Further, the extraction device 1 can be incorporated in the personal computer 11, the search device 10 itself can be constructed at low cost, and the search cost can be reduced.

In the above embodiments, the case where the title of the invention, the abstract, and the scope of claims are used as the file data read from the text file has been described, but the present invention is not limited to this and, for example, claims The present invention may be applied only to the above range, or may be applied to the keywords specifically used for the description of the embodiments (drawings) in the specification. When applied to the description of this embodiment, by accumulating the extracted keywords, it becomes possible to search and extract patent information related to improvement of a specific structure (for example, in the field of mechatronics).

In the above embodiment, three types of keyword data are set and compared, but, for example, four or more types of keyword data may be set and compared with file data. Furthermore, it goes without saying that the hardware configuration and the like in the above embodiments are also examples, and various modifications can be made without departing from the spirit of the present invention.

[0029]

As described above in detail, according to the keyword data extracting method and the keyword data extracting device of the patent database of the present invention, the keyword data can be automatically extracted from the claims and the like as they are. Therefore, the gist of the invention can be accurately grasped, and the keyword search efficiency can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram of a keyword data extraction device for a patent database according to the present invention.

FIG. 2 is a block diagram of a patent database search device using the extraction device.

FIG. 3 is a flowchart for explaining the operation of the extraction device.

FIG. 4 is a conceptual diagram of table A showing the second keyword data.

FIG. 5 is a conceptual diagram of table B showing the third keyword data.

FIG. 6 is a conceptual diagram showing a part of the file data before the extraction.

FIG. 7 is a conceptual diagram showing a state in which the first keyword data is extracted.

FIG. 8 is a conceptual diagram showing a state in which the second keyword data is extracted.

FIG. 9 is a conceptual diagram showing a state in which the third keyword data is deleted.

FIG. 10 is a conceptual diagram showing a state where the same keyword data is deleted.

[Explanation of symbols]

1 --- Extractor 2 --- Keyword data storage means 3--External storage means 4--Control Means 5 ・ ・ ・ ・ ・ ・ Storage means 6 ・ ・ ・ ・ ・ ・ ・ ・ Input means 7 ・ ・ ・ ・ ・ ・ ・ Output means 10 ・ ・ ・ ・ ・ ・ Search device 11 ・··· PC 12 ··· Keyboard 13 ··· Display (CRT) 14 ··· Printer 15 ··· CD-ROM 16 ... Central processing unit (CPU) 17 ... Random access memory (RA
M) 18 ... Read-only memory (ROM)

Claims (2)

[Claims] 1. A keyword data extraction method for extracting keyword data from document information of a patent application stored in an external storage means, wherein first keyword data consisting of kanji, katakana, code, romaji, etc. is extracted from the document information. And a step of extracting second keyword data consisting of a hiragana character string from the document information, and a step of leaving the third keyword data consisting of a specific kanji in a non-extracted state. Patent database keyword data extraction method. 2. A keyword data extraction device for extracting keyword data from document information of a patent application stored in an external storage means, wherein the keyword data storage means sets and stores a plurality of types of keyword data, and the keyword data storage means Based on the keyword data, first keyword data consisting of kanji, katakana, code, romaji, etc. and second keyword data consisting of a hiragana character string are extracted from the document information of the external storage means, and at the same time from a particular kanji And a control unit capable of keeping the third keyword data in a non-extracted state.