JP5988235B1 - Patent Literature Search Method - Google Patents

Abstract

An inexperienced person collects patent documents in the above fields comprehensively and screens them efficiently. The patent classification of a set of five or more master patent documents collected by an arbitrary method and the concept of keywords are expanded, and the master patent documents are increased by using the patent classification and keywords that are expanded the concepts. S1 to S6, selecting the patent classification of the increased master patent document set in a standardized manner, expanding the concept of keywords, and creating a search expression that comprehensively collects the desired patent documents S7 to S11 , The patent database is searched using the search formula to collect the patent document set 3, S12, the total number of applicable patent documents included in the patent document set 3 is estimated by the sampling method, The target document number that the target document set should have is determined from the target capture rate, and the relevant document from the patent document set 3 is screened while referring to the target document number. [Selection] Figure 2

Description

The present invention relates to a search method for a large number of similar patent documents, such as those used for patent map creation.

A patent document that does not leak patent documents that conform to a specific purpose and does not conform to a search purpose included in a hit patent document collection (hereinafter referred to as a non-conforming patent document, a non-conforming patent document, a patent that conforms to a search purpose) In order to create a search expression that hits a small number of documents, the knowledge of patent classification in the field to be searched, keywords in the field to be searched, and similar words or synonyms of the keyword Although it is indispensable, it is very difficult to master the knowledge about the patent classification and keywords in all the fields encountered as search targets, and in order to reduce the leakage of relevant patent documents, If a keyword with a large concept is used, the number of nonconforming patent documents will be very large, and it will take a lot of man-hours for screening. Is information.

Based on such a situation, a method for easily searching for similar patent documents using a computer has been proposed without having deep knowledge of patent classification, keywords, synonyms and similar words of the keywords.

As an example, Japanese Patent Application Laid-Open No. 2003-141164 automatically inputs other patent documents using the F-term assigned to the patent document by simply inputting the number of the specific patent document without considering patent classification or keywords in advance. A method of searching and displaying the similarity by the number of common F-terms has been proposed, but if only similar patent documents in a specific field such as those used in a patent map are to be collected, a search using a keyword is performed. If it is not used together, it will be difficult to set the criteria for discriminating between relevant and non-conforming documents with only the number of patent classifications, and even if you actually compare the number of patent classifications in the relevant and non-conforming literature sets, you will find a clear difference. It is difficult.

JP2003-141164A

As a method for determining a patent classification used in a search formula for comprehensively collecting patent documents of a specific technology, a plurality of patent documents serving as masters to be searched are collected by an arbitrary method and assigned to the master patent documents. There is a method of using the patent classification selected from the patent classifications used in the search formula for comprehensively collecting the patent classifications. However, since a plurality of patent classifications are generally assigned to one patent document, there are more than one. Since there are many types of patent classifications in the patent classification set of the master patent documents, there is a problem that it is difficult to determine which of the patent classifications should be adopted in the search formula for the comprehensive collection There was a point.

There is a method of determining by referring to keywords used in a plurality of patent documents as search target masters as keywords of the search formula for comprehensive collection, but the number of patent documents as masters is In the case of a small number, the keyword to be referred to is sufficient in quantity, or when the number of master patent documents is large, the keyword determination method is not standardized. There was a problem that it might occur.

In addition, if a broad concept is used as a patent classification or keyword used in a search formula in order to reduce the omission of retrieval of relevant patent documents, a document set containing a large number of non-conforming patent documents will be retrieved. There was a problem that a great amount of man-hours would be spent.

Accordingly, the first object of the invention of the present application is to be unfamiliar with the field of search target by standardizing the patent classification and keyword determination method used in the search formula for comprehensively collecting patent documents in the field to be searched. It is intended to provide a method that enables a person to search a patent document set that comprehensively includes relevant documents.

The second purpose is a method of creating a conforming document set by an efficient screening method even if a lot of nonconforming documents are included in a patent document set that comprehensively includes conforming patent documents collected by an arbitrary method. Is to provide.

In order to solve the above-described problems, the present invention employs means for solving the problems having the following characteristics.

The invention described in claim 1 is a patent document search method executed on a computer in which patent document analysis software and spreadsheet software are installed, and the patent document set stored in the storage unit of the computer is stored in the patent document. Process with analysis software and spreadsheet software,
Collect five or more patent documents with at least one of the patent names belonging to the patent document set that belongs to the target patent document set and the applicant, and master patent document set. A table for organizing the keywords determined by the searcher when the keywords used in the master patent document increase search expression are the keywords used in the name of the invention of the master patent document set 1 The keyword representing the category of the invention is arranged in the first hierarchical keyword column, the keyword that can modify the keyword arranged in the first hierarchical keyword column is arranged in the second hierarchical keyword column, the third hierarchical keyword In the column, a keyword that can modify the keyword arranged in the second-level keyword column is arranged, as in the case of arrangement. If all keywords determined by the user are placed in any hierarchical keyword column, the hierarchical keyword column is expanded, and if there are keywords (compound words) belonging to multiple hierarchical keyword columns among the keywords to be placed Separating the keywords so that the keywords can be arranged in a single hierarchical keyword column, arranging the keywords in the corresponding hierarchical keyword columns, and creating a hierarchical keyword table of the names of the inventions of the master patent document set 1;
The keywords that can be integrated into the partial match keywords for each hierarchical keyword column are integrated into the partial match keywords for each hierarchical keyword column, and the duplicate keywords are not duplicated. A step of creating a partial match search keyword table by hierarchy of names of inventions in the master patent document set 1 and search items, a logical sum of all the first patent classifications in the master patent document set 1, and search For each hierarchical keyword string of the patent classification search formula 1 for the forward matching and the keyword table for the partial match search keyword table for each hierarchical name of the invention of the master patent document set 1, the search item is the name of the invention, the summary, and the claim. The range keyword and search keyword are the logical OR of the keywords in the hierarchical keyword column, and the search method is Creates an expression for partial matching, creates a logical product of the expressions in the same set for all of the sets created by sequentially removing one expression from the set of all the created expressions. A keyword search formula 1 that makes all of the obtained logical products the logical sum is created, and a master patent document increase search formula is created using the logical product of the patent classification search formula 1 and the keyword search formula 1 Searching the patent database using the step and the search formula for increasing the master patent document, screening the hit document set to create a master patent document set 2 containing only relevant documents, and the search target set as the master patent document Set 2, search keyword is cumulative patent classification (the cumulative patent classification is a document to which all patent classifications included in the master patent literature collection 2 are assigned the patent classification) When the numbers are arranged in descending order as the sort key, the logical classification of the patent classification with the first number of documents, the logical sum of the first and second patent classifications, and then the first, second and third patent classifications The logical sum of the patent classifications created by including the patent classification one rank lower in the order like logical sum), the search method is a search with forward matching, the total number of documents in the master patent document set 2 of the number of hit documents Search for patent documents to create a collection of patent documents for which the cumulative patent classification at the time when the ratio to the above (hereinafter referred to as the ratio of the number of hits by cumulative patent classification) reaches or exceeds the reference value For each of the step of deciding to make a patent classification of Formula A, the number of documents by keyword of the invention name of the master patent document set 2, the number of documents by keyword of summary, and the number of documents by keyword of claims Calculating a ratio of the master patent document set 2 to the total number of documents, determining a keyword candidate extraction range to be used in the patent document search formula A using the calculated ratio as an evaluation scale, and the keyword candidate extraction range All the keywords within the determined range are collected in one keyword table, and only the keywords determined by the searcher to be used in the patent document search formula A are extracted from the table, using the extracted keywords, A hierarchical keyword table of the master patent document set 2 is created by the same creation method as the hierarchical keyword table created in the master patent document set 1, and the master patent document set 1 is created using the created hierarchical keyword table. Master patent documents in the same way as the created keyword table for partial match search by layer Step 2 for creating a partial partial search keyword table for each hierarchy and search items, search keywords for patent classification are cumulative patent classifications determined in the patent classification of the patent document search formula A, and patent classification with the search method being forward matching For each hierarchical keyword string in the keyword table for partial matching search by hierarchy of the search formula 2 and the master patent document set 2, the search item is the keyword of the full text of the text, the search keyword is the logical sum of the keywords of the hierarchical keyword string, the search method Creates a partial match formula, and the logical product of the formulas of all the created formulas is used as the keyword search formula 2, and the logical product of the patent classification search formula 2 and the keyword search formula 2 is used to The patent document search formula A is created, and the patent document search formula A is used to search the patent database. 3 (assuming the number of documents is A) and at least M from the patent document set 3 (the M cases are sampled when the number of relevant documents included in the patent document set 3 is estimated by a sampling method) Statistically based on the tolerance of the estimated number of reference documents included in the patent document set 3 and the estimated number of relevant documents included in the patent document set 3 (hereinafter referred to as β). A patent document set S created by randomly extracting S sample samples exceeding the number of sampling documents to be determined} is a conforming document set S1 (the number of documents in the document set is hereinafter referred to as S1) and a nonconforming document set. It is classified into S2 (the number of documents in the document set is hereinafter referred to as S2), and S1 / S (the ratio is referred to as relevant document content rate and is hereinafter referred to as α) and β is used to form the patent document set 3 Included in The maximum number of estimated applicable patent documents is the sum of the product of A and α and the product of A, α, and β, and the minimum number of estimated values is the product of A, α, and the product of A, α, and β. The ratio of the number of patent documents to be included in the patent document set 3 to the number of patent documents that the target patent document set should have and the step of estimating that the difference is a difference (this ratio is referred to as a capture rate and hereinafter) (referred to as γ), and the number of patent documents that the target set of patent documents should have (the number of documents is hereinafter referred to as G) is an estimate of the number of relevant documents included in the patent document set 3 The product of the minimum number of values and γ (the number of documents is hereinafter referred to as Gmin) and the maximum number of estimated values of the relevant documents included in the patent document set 3 and γ (the number of documents is referred to as Gmax below). From the patent document set 3 and the step of setting that the number of documents is between In the first search for extracting a joint document, the search target document set is the patent document set 3, the search item is the keyword of the name of the invention, and the search keyword is in the relevant document set S1 including the keyword in the name of the invention. The logical sum of the keywords extracted by the searcher from the keywords in the extraction range determined from the ratio of the number of documents belonging to the number of documents belonging to the non-conforming document set S2 as an evaluation scale, and the keyword partially matching the keywords, the search method is complete Implemented with a formula that matches, the hit literature set name is H1 full set, the relevant literature set created by screening the H1 full set is H1 relevant set, the number of documents in the H1 relevant set is H1, and if H1 ≧ Gmin If the H1 compatible set is the target set of patent documents, and H1 ≧ Gmin does not hold, the process proceeds to the second search step.
In the second search, the search target document set is the logical difference set of the patent document set 3 and the H1 complete set (hereinafter referred to as the patent document set B), the search item is the summary keyword, and the search keyword is the search keyword A keyword extracted by a searcher from keywords in an extraction range determined by using a ratio of the number of documents belonging to the relevant document set S1 including a keyword in the summary and the number of documents belonging to the non-conforming document set S2 as an evaluation measure, and the keyword and part The logical sum of the matching keywords and the search method are implemented with the expression that makes the exact match, the hit literature set name is the H2 complete set, the relevant reference set created by screening the H2 complete set is the H2 compatible set, and the H2 compatible set If the number of documents in H2 is H2, and the sum of H1 and H2 ≧ Gmin, the logical sum set of H1 compatible set and H2 compatible set is the target , And the case in which the sum ≧ Gmin of the H1 and H2 is not satisfied, the process proceeds to third search of the step,
In the third search, the search target document set is a logical difference set of the patent document set B and the H2 complete set (hereinafter referred to as the patent document set C), the search item is the keyword of the claim, and the search keyword is , A keyword extracted by a searcher from keywords in an extraction range determined by using a ratio between the number of documents belonging to the relevant document set S1 including the keyword in the claim and the number of documents belonging to the non-conforming document set S2 as an evaluation scale; The logical OR of the keywords that partially match the keyword, and the search method are performed with an expression that perfectly matches, the hit document set name is the H3 complete set, and the relevant reference set created by screening the H3 complete set is the H3 compatible set, If the number of documents in the H3 compatible set is H3, and the sum of H1, H2 and H3 ≧ Gmin, the H1 compatible set, the H2 compatible set, and the H3 compatible set A patent document set Liwa set is the target, when the sum ≧ Gmin of H1 and H2 and and H3 not satisfied, the process proceeds to the fourth search step,
In the fourth search, the search target document set is a logical difference set of the patent document set C and the H3 full set (the set is hereinafter referred to as patent document set D), the search item is the keyword of the full text, and the search keyword is A keyword extracted by a searcher from keywords in an extraction range determined by using a ratio of the number of documents belonging to the relevant document set S1 and the number of documents belonging to the non-conforming document set S2 including the keyword in the full text as the evaluation scale, and the keyword The logical sum of keywords that partially match, and the search method is an expression that makes the search match completely, the hit document set name is the H4 all set, the matching document set created by screening the H4 all set is the H4 matching set, the H4 If the number of documents in the matching set is H4 and the sum of H1, H2, H3, and H4 ≧ Gmin, then the H1 matching set, the H2 matching set, and the H3 matching set H4 is a patent document set logical union is the target of the adaptation set of, when the sum ≧ Gmin of H1 and H2 and H3 and H4 is not satisfied, the process proceeds to fifth search step,
The fifth search is performed by creating a logical difference set (hereinafter referred to as a patent document set E) of the patent document set D and the H4 all set as a search target document set, and screening the document set. If the document set of the H5 conforming set is H5, the number of documents of the H5 conforming set is H5, and the sum of H1, H2, H3, H4, and H5 ≧ Gmin, then the H1 conforming set, the H2 conforming set, the H3 conforming set, and the H4 conforming set A search method for a patent document, characterized in that a logical sum set of H5 compatible sets is a target patent document set and the sum of H1, H2, H3, H4, and H5 ≧ Gmin is not satisfied, and is reviewed from the beginning.

The invention described in claim 2 is a patent document search method executed by a computer in which patent document analysis software and spreadsheet software are installed, and the patent document set stored in a storage unit of the computer is stored in the patent. Process with literature analysis software and spreadsheet software,
Collecting patent documents of a specific technology in an arbitrary manner to create a patent document set 3 (the number of documents is A) and at least M from the patent document set 3 {the M When the number of relevant documents included in the patent document set 3 is estimated by the sampling method, the number of documents A in the patent document set 3 that is a sampling population and the number of relevant documents included in the patent document set 3 A patent document set S created by randomly extracting S sample samples exceeding the sampling error number statistically determined from the allowable error of the estimated value (the allowable error is hereinafter referred to as β) The documents are classified into a conforming document set S1 (the number of documents in the document set is hereinafter referred to as S1) and a nonconforming document set S2 (the number of documents in the document set is hereinafter referred to as S2), and S1 / S (the ratio is referred to as a conforming document). Contains ) And β, the maximum number of estimated values of the number of relevant patent documents included in the patent document set 3 is the product of A and α and the product of A, α, and β. The step of estimating the minimum number of estimated values as the difference between the product of A and α and the product of A and α and β, and the patent document set 3 of the number of patent documents that the target patent document set should have The ratio of the total number of relevant documents included in the estimate (referred to as the capture rate, hereinafter referred to as γ) is set, and the number of patent documents that the target patent document set should have (the number of documents) (Hereinafter referred to as G) is included in the patent document set 3 and the product of the minimum number of the estimated values of the number of relevant documents included in the patent document set 3 and γ (the number of documents is hereinafter referred to as Gmin). Product of the estimated maximum number of relevant literatures and γ (the number of literatures is referred to as Gmax hereinafter) And the first search for extracting relevant documents from the patent document set 3, the search target document set is the patent document set 3, and the search item is the name of the invention. A keyword or a search keyword is a searcher from keywords in an extraction range determined by using a ratio between the number of documents belonging to the relevant document set S1 and the number of documents belonging to the non-conforming document set S2 including the keyword in the name of the invention as an evaluation measure. The logical sum of the keyword extracted by the keyword and the keyword that partially matches the keyword, and the search method is an expression that is completely matched, the hit literature set name is the H1 complete set, and the relevant reference set created by screening the full H1 set Is the H1 compatible set, and the number of documents in the H1 compatible set is H1, and if H1 ≧ Gmin, Yes, if H1 ≧ Gmin does not hold, proceed to the second search step,
In the second search, the set of documents to be searched is the logical difference set of the patent document set 3 and the H1 complete set (hereinafter referred to as the patent document set B), the search items are the keywords of the summary and claims, and the search The keyword is obtained by the searcher from the keywords in the extraction range determined by using the ratio of the number of documents belonging to the relevant document set S1 and the number of documents belonging to the non-conforming document set S2 including the keyword in the summary or claims as an evaluation measure. The logical sum of the extracted keyword and the keyword that partially matches the keyword, and the search method are executed with the exact match. The hit document set name is the H2 complete set, and the relevant reference set created by screening the complete H2 set is created. If the number of documents in the H2 matching set and the H2 matching set is H2, and the sum of H1 and H2 ≧ Gmin, the logical sum of the H1 matching set and the H2 matching set A patent document set if's target, when the sum ≧ Gmin of H1 and H2 is not satisfied, the process proceeds to the third search step,
In the third search, the search target document set is a logical difference set of the patent document set B and the H2 full set (the set is hereinafter referred to as a patent document set C), the search item is the keyword of the full text, and the search keyword is: A keyword extracted by a searcher from keywords in an extraction range determined by using a ratio of the number of documents belonging to the relevant document set S1 and the number of documents belonging to the non-conforming document set S2 including the keyword in the full text as the evaluation scale, and the keyword The keyword is logically ORed with a partial match, and the search method is an exact match. The hit document set name is the H3 complete set, the relevant reference set created by screening the H3 complete set is the H3 compatible set, the H3 If the number of documents in the matching set is H3, and the sum of H1, H2, and H3 ≧ Gmin, the logic of the H1 matching set, the H2 matching set, and the H3 matching set Set is a patent document set a target, the process proceeds when the sum ≧ Gmin of H1 and H2 and and H3 not satisfied to the fourth search step,
The fourth search is performed by creating a logical difference set of the patent document set C and the H3 all set as a search target document set, screening a relevant document set created from the document set, an H4 compatible set, and a document of the H4 compatible set If the number of cases is H4 and the sum of H1, H2, H3, and H4 ≧ Gmin, the logical sum set of the H1 matched set, H2 matched set, H3 matched set, and H4 matched set is the target patent document set, and H1 and H2 If the sum of H3 and H4 ≧ Gmin does not hold, review from the beginning.
A method for searching patent documents, comprising:

According to the analysis of the patent document, the present invention has the following effects.

In the first aspect of the invention, even those who are unfamiliar with patent classifications and keywords of a specific technology start the creation of a small number of master patent document sets, and organize the patent classifications included in the master patent document set. By standardizing the method and expanding the concept of the keyword, it is possible to comprehensively search for patent documents in a specific field, and sequentially create a low set from a set of documents with a high content of relevant patent documents. The operation is completed by the screening until the cumulative number of documents in the document set reaches the target value, and the document set with a low patent document content rate is not screened, thereby enabling efficient patent document search.

In the second aspect of the invention, an efficient screening method can be used even from a collection of patent documents comprehensively collected using high-level keywords at the expense of mixing non-conforming patent documents to prevent omission of search. Patent document search can be easily performed.

Patent document analyzer Flowchart of Patent Literature Search Method (Part 1) Flowchart of Patent Literature Search Method (Part 2) Master patent literature set 1 First patent classification table of master patent document set 1 Number of documents by keyword for the title of the invention of the master patent document set 1 ranking table Hierarchical keyword table of invention names of master patent document set 1 Keyword table for partial matching search by hierarchy of titles of the invention of master patent document set 1 List of logical ORs of keywords by level in keyword table for partial match search by level of master patent document set 1 Master patent literature increase search formula Master Patent Document Set 2 Document Number Ranking Table by All Patent Categories Document Hit Ratio Table by Cumulative Patent Classification of Master Patent Document Set 2 Master content table by keyword for the title of the invention of master patent document set 2 Master content table for each keyword in the summary of Master Patent Document Set 2 Master content table by keyword in claims of master patent document set 2 Keyword table for hierarchical analysis of title and abstract of invention of master patent document set 2 and claims Hierarchical keyword table of title, abstract and claims of invention of master patent document set 2 Keyword table for partial matching search by hierarchy of title, abstract and claims of invention of master patent document set 2 Patent Literature Search Formula A Necessary sampling number and tolerance for estimating the number of documents of a specific technology included in a specific patent document set Relevant / nonconforming document ratio table by keyword for the name of the invention of patent document set S Relevant / nonconforming document ratio table for each keyword in the summary of patent document set S Search formula for creating A * S name set Retrieval formula for creating B * S summary set

Hereinafter, embodiments of the patent document search method of the present invention will be described with reference to the drawings. However, the embodiments of the present invention show the most preferable modes of the present invention, and the present invention is not limited thereto.

As shown in FIG. 1, a computer in which patent document analysis software and spreadsheet software used for analyzing a patent document set, which is a collection of a plurality of patent documents to which patent classifications are assigned, such as a published patent gazette and a patent gazette, are installed. The structure of the patent document analyzer 1 provided with 2 is demonstrated.

This patent document analysis apparatus 1 has a computer 2 to which a control device 3, an arithmetic device 4, a storage device 5 and an I / O 6 are connected via a bus line, and on the input side of the I / O 6 of the computer 2 A keyboard 7, a mouse 8, an input device 9, a CD-ROM drive 10 including an MO drive, etc. are connected, and on the output side of the I / O 6, a display device 11 comprising a CRT (including a liquid crystal display), a printer 12 etc. are connected. Furthermore, it is possible to connect to an ASP patent database 13 such as SRPARTNER or NRI Cyber Patent Desk through the communication line 14. The patent database 13 stores bibliographic information and patent documents such as specification contents.

The main functions of the patent document analyzer 1 are as follows.
When the input device 9 is operated to open a file of a set of patent document to be analyzed using the patent document analysis software preinstalled in the storage device 5 of the computer 2, a front page is displayed. Then, select a data object such as search / list and proceed with the operation.

To create a patent map, a map is selected as a data object, a map creation condition setting screen is drawn on the display device 11, and various map creation conditions are set on the setting screen. Map (the first column is the number of documents to which each display item column such as patent classification, keyword, applicant, etc., the second column is assigned the keyword of the display item column, the first column is the second column A numerical table arranged in descending order using the number of documents as an alignment key is mapped, and switching between the numerical table and the map can be easily performed by the input device 9), or a reference number comparison map of two document sets by keyword (one column) The second column is the number of documents in the document set 1. The third column is the number of documents in the document set 2. The fourth column is the number of documents in the second column. The keyword array of the display items in the first column is a map of a numerical table that is arranged in descending order using the ratio in the fourth column as the alignment key. Can be easily made by the input device 9).

In addition, the numerical table of the various maps is copied and pasted to the sheet of the spreadsheet software displayed on a screen different from the numerical table screen, and the copied and pasted numerical table is used using the spreadsheet software. Addition or deletion of rows and columns, description of characters in each cell, deletion, arithmetic operation of numbers between cells, alignment of a specific range of data in ascending or descending order using characters in a specific row or column, etc. Processing can be done freely.

Retrieval based on various patent classifications and keywords for a patent document set stored in the storage device 5 as a search target, a logical sum set of files stored as objects of the patent document set stored in the storage device 5, and logic Difference sets and intersection sets can be created freely.

A method for creating a collection of patent documents necessary for analysis is as follows. First, the input device 9 is operated to access the patent database 13 or the published patent publication CD set in the drive 10 such as a CD-ROM, Patent literature search is performed using a search expression composed of a patent classification, a keyword, etc., the hit literature set is named and saved in the storage device 5, and the patent literature saved in the storage device 5 The set is converted into a format that can be processed by the patent document analysis software using a data format conversion program installed in the storage device 5 in advance, and the converted patent document set is screened as necessary to analyze patent documents. A set is created and stored in the storage device 5.

The patent classification in this application means any of IPC and FI sections, classes, subclasses, main groups, and subgroups, and the patent document means a document to which the patent classification is assigned. .

Next, a patent document search method according to the present invention will be described with reference to a search expression for creating a summary set of B * S in FIG.

S1, Collect five or more patent documents of at least one of the names and applicants of the invention given to one patent document by any method, and collect master patent document set 1 (see FIG. 4 as an example). create.

This is a step of creating a master patent document set for extracting patent classifications and keywords used to create a master patent document increase search expression.

The master patent document collection method in this step is an arbitrary method. For example, the patent database 13 is connected to the patent database through the communication line 14 and the patent database 13 is operated by the operation of the input device 9. A patent document search is performed, the hit patent document is stored in the storage device 5 as a file A1, the file A1 is converted into a data format that can be processed by the patent document analysis software, and a file A2 is created. There is a method of selecting five or more documents and storing the five or more patent documents in the storage device 5 as a master patent document set 1.

As a necessary condition for the master patent document set 1, it is specified that at least one of the name of the applicant or the invention of the invention is collected from the applicant and the name of the invention given to one patent document. The reason for this is that even patent documents related to the same field have different viewpoints and different patent classifications and keywords used for the applicant and invention, so there are many keywords used in the master patent document increase search formula. This is because various types of patent classifications and keywords can be collected.

Furthermore, the reason for prescribing a set of five or more patent documents as a necessary condition of the star patent document set 1 is that the number of patent document sets created for a specific purpose is in accordance with Pareto's law. Based on the idea that the leading patent classification that can hit 80% is the main patent classification of the patent document set, if there are at least five master patent documents, even if one of them is a non-conforming document, This is because 80% of the first patent classification of the master patent document set 1 is the patent classification of the relevant literature.

S2, the first patent classification table of the master patent document set 1 is created.

This step is a step of extracting a patent classification used in the search formula for increasing the master patent document. As a means for automatically performing the extraction operation of the patent classification using the patent document analyzer 1, the master patent is used. A document number ranking table (see FIG. 5 as an example) for the first patent classification of the document set 1 is created, and this table is substituted for the first patent classification table of the master patent document set 1.

The method for creating the first patent classification table of the master patent document set 1 is that the input device 9 is operated to open the file A2 from the start screen of the patent document analysis software, and the ranking patent map from the front screen of the file A2. A creation condition setting screen is displayed on the display device 11. The selection of the analysis target object on the ranking patent map creation condition setting screen selects the master patent document set 1, and the selection of the analysis display item selects the first patent classification. When the map creation is executed, a ranking ranking map of the number of documents classified by the first patent classification of the master patent document set 1 is created, and when the input device 9 is operated to switch the ranking map to a table, the top of the master patent document set 1 It is possible to create a document number ranking table by patent classification (see Fig. 5 as an example).該筆 head Patent Classification document-number ranking table is a substitute for all the largest patent classification table of the master Patent Document set 1 since that.

S3, the keyword of the title of the invention of the master patent document set 1 is extracted.

This step is a step of extracting a keyword used in the search formula for increasing the master patent document. The means for automatically extracting the keyword using the patent document analyzer 1 is the master patent document set. Document number ranking table by keyword of the name of the first invention (the first column is the keyword column of the name of the invention, the second column is the document number column to which the keyword is assigned, the first column is arranged in two columns A numerical table arranged in descending order using the number of documents in the eye as an alignment key is created as an example), and keywords of the table are used.

The method of creating the keyword-specific document number ranking table of the title of the invention of the master patent document set 1 is that the input device 9 is operated to open the file A2 from the start screen of the patent document analysis software, and the front of the file A2 The ranking patent map creation condition setting screen is displayed on the display device 11 from the screen, and the selection of the analysis target object on the ranking patent map creation condition setting screen is performed by selecting the master patent document set 1, and the selection of the analysis display item is the invention. When the name keyword is selected and the creation of the patent map is executed, the document number ranking map by keyword for the name of the invention of the master patent document set 1 is created, so the input device 9 is operated to switch the ranking map to a table. And the number of documents by keyword of the title of the invention of the master patent document set 1 It is possible to create a down King table (see FIG. 6 as an example).

S4, the keyword of the name of the invention of the master patent document set 1 is classified by hierarchy from the viewpoint of the order of modifying the keyword expressing the category of the invention, and the keyword table by level of the name of the invention of the master patent document set 1 The keywords of the hierarchical keyword table are organized from the viewpoint of partial matching and duplication, and a keyword table for partial matching search by hierarchical name of the invention of the master patent document set 1 is created.

Even if the keywords in the number-by-keyword document ranking table of the title of the invention of the master patent document set 1 are used as they are in the search formula for increasing the master patent documents, it is not possible to expect a significant increase in the number of master patent documents. From the viewpoint of the order in which the keywords representing the category of the invention are modified, the keywords in the document number ranking table according to the keywords of the invention of the document set 1 (for example, four keywords such as watch, device, elderly, life) In this case, the keyword expressing the category of the invention can be determined as a device, and the order in which the devices are modified can be determined like a device that watches the life of the elderly. Keywords that work in the same way, keywords in the layer that modifies the watch are keys that work the same as life (For example, see FIG. 7), and the keywords in the hierarchical keyword table are further divided from the viewpoint of partial matching and duplication. Organize and create a keyword table for partial matching search for each name of the invention of the master patent document set 1 (see FIG. 8 as an example), create an expression for each layer of the keyword table for partial matching search by hierarchy, The concept of the keyword of the invention name of the master patent document set 1 is expanded by the combination of the created formulas, and the master patent document is increased by using the keyword in the search formula for increasing the master patent document.

The method of creating the keyword table classified by hierarchy of the names of the inventions of the master patent document set 1 (see FIG. 7 as an example) is as follows. ) Is used to create the search formula for increasing the master patent document, the keyword extracted by the searcher is extracted, and the keyword indicating the category of the invention among the extracted keywords, that is, the invention of the product or the invention of the method is determined. Keywords are arranged in the first hierarchy keyword column, keywords other than the keyword representing the category of the invention are keywords that can modify the keyword in the first hierarchy keyword column, the keyword in the second hierarchy keyword column, and the keyword string in the second hierarchy The keyword that can modify the keyword is arranged in the third keyword column so that the searcher can Until all the keywords it is determined to use to create a search expression increased star patent documents can be placed in the appropriate hierarchical keyword string to create increased hierarchical keyword column.

When there is a compound word that spans a plurality of hierarchical keyword strings among the keywords judged by the searcher when used to create the search formula for increasing the master patent document, the compound word is decomposed into keywords belonging to a single hierarchical keyword string The decomposed keywords are allocated to the corresponding hierarchical keyword strings.

The method for creating the keyword table for partial matching search by name of the invention of the master patent document set 1 (see FIG. 8 as an example) is the keyword table by hierarchy of names of the invention of the master patent document set 1 (see FIG. 7 as an example). The keywords that can be integrated into the partial match keywords are integrated into the partial match keywords, and the duplicate keywords are arranged in order to eliminate duplication.

S5, the patent classification search formula 1 using the patent classification of the first patent classification table of the master patent document set 1 and the keywords of the partial match search keyword table by hierarchy of the names of the invention of the master patent document set 1 were used. The master patent document increase search formula is created using the logical product of the keyword search formula 1.

Patent classification search formula 1 is a search item, a search keyword for patent classification is a logical sum of all patent classifications in the first patent classification table of the master patent document set 1, and a search method is a forward match (for example, in FIG. 10) The patent classification search formula 1), the keyword search formula 1 is a search item for each hierarchical keyword column in the keyword table for partial matching search by name of the invention of the master patent document set 1 according to the invention name and summary. Create a formula (refer to K1, K2, K3, and K4 in FIG. 10 as an example) in which the keywords in the claims and the search keywords are the logical sum of the keywords in the same hierarchical keyword string and the search method is a partial match. For all of the sets obtained by removing one expression from all the expressions, the logical product of the expressions in the same set is created, and the logical sum of all the created logical products (for example, FIG. 1 It is a keyword search for a formula 1) in.

The reason for creating a set obtained by removing one expression from all the expressions created for each hierarchical keyword string in the keyword search expression 1 is that the restriction by the hierarchical keyword string corresponding to the removed expression is removed and a new set is created. This is to capture keywords, and in the description of the present application, a set that does not use one expression in sequence is created from the above expressions, K1, K2, K3, and K4, and the logical product of the expressions in the set is used. However, the number of formulas that are not used sequentially is not necessarily one, and may be an appropriate number by trial search.

The reason why the keywords used in the search formula for increasing the master patent document are limited to the keywords extracted from the keyword of the name of the invention is that, based on experience, the ratio of the number of relevant documents to the total number of hits when using the keyword Is much larger than that based on the keywords in the summary and claims, and is easy to screen, and there are a sufficient number of relevant documents necessary for collecting the keywords used in the search formula for increasing the master patent document. It is because it is obtained.

S6, searching the patent document database 13 using the search formula for increasing the master patent document, converting the hit document set into a data format that can be processed by the patent document analysis software, screening the data format converted document set Thus, a master patent document set 2 having only relevant documents is created.

This step is a step of creating a master patent document set 2 having a number exceeding the master patent document set 1, specifically, connecting to the patent database 13 through the patent analysis apparatus 1 and the communication line 14, By the operation, the patent document search of the patent database 13 is performed by the search formula for increasing the master patent document, the hit document set is stored in the storage device 5 as the file B1, and the file B1 can be processed by the patent document analysis software The file B2 is created by conversion into the file B2, the file B2 is screened to create a document set of only relevant documents and a master patent document set 2, and the document set is stored in the storage device 5.

S7, create a document number ranking table (see FIG. 11 as an example) of all patent classifications of the master patent document set 2, and based on the ranking table, a document hit ratio ratio table by cumulative patent classification (see FIG. 12 as an example) To determine the patent classification of Patent Document Retrieval Formula A that is used to search for patent documents targeting the cumulative patent classification selected as an evaluation scale. .

The method for creating a document number ranking table for all patent classifications of the master patent document set 2 is to operate the input device 9 to open the master patent document set patent document set 2 from the start screen of the patent document analysis software, The ranking patent map creation condition setting screen is displayed on the display device 11 from the front screen of FIG. 5. The selection of the analysis object on the ranking patent map creation condition setting screen is performed by selecting the master patent document set 2 and selecting the analysis display item. If all patent classifications are selected and a patent map is created, a ranking map for all patent classifications of the master patent document set 2 is created. Therefore, if the ranking map is switched to a table by operating the input device 9, the master patent document Create a ranking table for all patent categories in set 2 (see Fig. 11 as an example) It can be.

The method of creating the cumulative patent category document hit ratio table (see FIG. 12 as an example) of the master patent document set 2 is based on the total patent category document number ranking table (see FIG. 11 as an example) of the master patent document set 2. , By operating the input device 9, copying and pasting to the sheet of spreadsheet software, and pasting the document number ranking table for all patent categories, the cumulative patent category column, the document hit count column by cumulative patent category, and the cumulative patent category Increase the number of documents hit ratio (%) column. For example, F1 is the first patent category in the ranking table for the number of documents by all patent classifications, F2 is the patent classification of F2, and F3 is the third patent classification. For example, F1 is entered in the cumulative patent category column ranked first, and the logical sum of F1 and F2 is entered in the cumulative patent category column ranked second. In the classification column, the logical sum of F1, F2, and F3 is described. In the cumulative patent category-based document hit count column for each of the first, second, and third rankings, the search target document set is the master patent document set 2, search. The keyword is the patent classification of the ranking in the cumulative patent classification column, that is, F1 for the first ranking, F1 and F2 for the second ranking, and F1 and F2 for the third ranking. Enter the number of hit documents with the logical sum of F3, and the number of hit documents by cumulative patent classification column shows the total number of references in the master patent document set 2 of the number of hits by cumulative patent classification of each ranking. Fill in the percentage of

S8, master content table by keyword for the title of the master patent literature set 2 (see FIG. 13 as an example), master content table by summary keyword (see FIG. 14 as an example), master content by keyword in claims A rate table (see FIG. 15 as an example) is created, and an extraction range of keywords used in the patent document search formula A is determined using the master content rate in the table as an evaluation scale.

The keyword group used in the patent document search formula A is extracted from the keyword of the name of the invention, the abstract keyword, and the keyword of the claims used in the master patent document set 2, and the keyword Therefore, in order to efficiently extract the keywords to be used, the ratio of the number of documents by keyword of the name of the invention of the master patent document set 2 to the number of documents in the master patent document set 2 (the ratio is referred to as The master content by keyword in the title of the invention and the description, the same description is summarized and applied to the keyword in the claims), and the master content by keyword in the summary and the claims is calculated in the same manner The keyword extraction range used in the patent document search formula A is determined using the master content rate as an evaluation scale.

The reference value of the master content rate that determines the keyword extraction range of each keyword master content rate table is, based on experience, zero or more for the keyword of the name of the invention, that is, all keywords as the keyword extraction range, If the sum is about 2% or more for the summary keyword and about 5% or more for the claimed keyword, there is no particular problem in terms of omission of search and processing man-hours.

The master content table by keyword for each of the titles, abstracts, and claims of the invention of the master patent document set 2 is a keyword-specific document for each of the titles, abstracts, and claims of the master patent document set 2. A master content rate column is added to the number ranking table (see FIG. 6 as an example), and the added column is created by describing the ratio of the number of documents by keyword to the total number of documents in the master patent document set 2 (example) (See FIGS. 13, 14, and 15).

S9, the names, summaries, and claims of the master patent document set 2 as keyword candidates for use in the patent document search formula A for all keywords selected as an evaluation scale in the master content rate table for each keyword Are summarized in a keyword table for hierarchical analysis (see FIG. 16 as an example), and the keyword table of the invention of the master patent document set 2 and the summary and the keyword table by hierarchy of claims (see FIG. 17 as an example) from the keyword table. create.

In this step, all relevant keywords satisfying the standard value of the master content rate in the master content rate table for each keyword of the title, summary, and claims of the invention are included in order to exhaustively hit the relevant documents to be collected. Using the keyword extracted by the searcher determined to be used in the patent document search formula A, the master patent is created in the same manner as the method for creating the hierarchical keyword table of the titles of the master patent document set 1 in S4. Create a hierarchical keyword table (see FIG. 17 as an example) of the title, abstract and claims of the invention of the document set patent document set 2 and use it for the title, summary and claims of the master patent document set 2 Is a preparatory step to extend the concept of keywords that are

S10, the keyword table for partial matching search of the title, summary and claims of the master patent document set 2 by hierarchy from the name and summary of claims of the invention of the master patent document set 2 and the claims (as an example) (See FIG. 18).

In this step, the keywords that can be integrated into the partially matched keywords for each hierarchical keyword column are integrated into the partially matched keywords for each keyword in the hierarchical keyword table of the title and summary of the master patent document set 2 and the claims. The overlapping keyword is a step of creating a keyword table for partial matching search according to the hierarchy of the title and abstract of the invention of the master patent document set 2 and the claims without duplication.

A keyword using keywords in the keyword table for partial matching search by name of the patent classification search formula 2 using the patent classification determined in S11 and S7, and the name and abstract of the invention of the master patent document set 2 in S10 and claims. The patent document search formula A is created using the logical product of the search formula 2.

The patent classification search formula 2 is a search item, the search keyword of the patent classification is a logical sum of the patent classification determined to be the patent classification of the patent document search formula A in S7, and the search method is a forward match (for example, FIG. 19 In the patent classification search formula 2, the keyword search formula 2 is searched for each hierarchical keyword column of the keyword table for partial matching search according to the hierarchy of the title and summary of the master patent document set 2 and claims. Create an expression (see A1, A2, A3, and A4 in FIG. 19 as an example) in which the item is a full-text keyword, the search keyword is a logical sum of keywords in the same hierarchical keyword string, and the search method is a partial match. This is a logical product of all the created formulas (for example, keyword search formula 2 in FIG. 19).

S12, search the patent document database 13 using the patent document search formula A, convert the hit document set into a data format that can be processed by the patent document analysis software, and convert the patent document set 3 (the number of documents in the document set) Is A).

The patent document set 3 is created by connecting the patent database 13 to the patent database 13 via the communication line 14 with the patent analysis apparatus 1, and performing the patent document search of the patent database 13 using the patent document search formula A by operating the input device 9. The hit literature set is stored in the storage device 5 as the file C1, and the document set obtained by converting the file C1 into a data format that can be processed by the patent document analysis software is the patent document set 3, and the set is saved in the storage device 5 .

The patent document set 3 is a collection of five or more master patent documents composed of five or more master patent documents collected by a search formula created by expanding the concept of patent classification and keywords used in the master patent document set 1. Patent classification (see S7) and concept expansion based on Pareto's Law based on the patent classification and keywords of Master Patent Document Set 2 that has more than 100 (experientially more than 100 documents) Since it is a document set that has been hit by a search formula (see S11) using a keyword that has been processed, the patent classification is more than the average document set of the patent document set that has been created by the searcher's own unique technique. The number of hits of non-conforming documents by is small, and the number of hits of non-conforming documents by keywords is estimated to be large. It said to be a document set that was comprehensively collect.

S13 is a step of estimating the total number of relevant documents included in the patent document set 3 by a sampling method.

This step is a step of estimating the number of relevant documents by the sampling method as a result of searching the patent document database 13 using the patent document search formula A. M cases are included in the patent document set 3 and the patent document set 3 in the patent document set 3 which is a sampling mother set when the number of relevant documents included in the patent document set 3 is estimated by the sampling method. Patents created by random sampling of S sample documents with the number of documents exceeding the allowable error of the estimated number of relevant documents (Note 1), which is the number of sample documents statistically determined from β] The document set S is screened and the sample compatible document set S1 made up of S1 relevant documents (the set is stored in the storage device 5 as a data object of the patent document set 3). And S2 sample nonconforming document set S2 (the set is stored in the storage device 5 as a data object of patent document set 3), and the ratio of the conforming document number S1 to the sample document number S (conforms this ratio) And the tolerance β, and the maximum number of estimated values of the number of relevant patent documents included in the patent document set 3 is the product of A and α, and A and α. The minimum number of estimated values is estimated to be the difference between the product of A and α and the product of A, α, and β.

(Note 1) The allowable error is the allowable error when estimating the number of conforming documents included in Patent Document Set 3 from the analysis results of a randomly sampled sample. ± 3%, ± 5%, and ± 10% are often used (see FIG. 20 as an example).

S14, setting a ratio γ of the number of documents that should be included in the target patent document set to the number of relevant documents included in the patent document set 3 (this ratio is referred to as a capture rate and hereinafter referred to as γ), and γ is set. The estimated minimum value (denoted as Gmin) of the number of documents (denoted as G) that should be included in the set of patent documents to be defined as the minimum number of estimated values of the number of relevant documents included in the patent document set 3 and γ The product and the estimated maximum value (Gmax) are determined as the product of the maximum number of estimated values of the number of relevant documents included in the patent document set 3 and γ.

The capture rate γ is a value that can be freely set according to the purpose of use of the patent document set for patent maps. If you want to reduce search omissions, set a high value, allow some search omissions, and reduce the number of processing steps. A low value may be set.

Therefore, the pass / fail criterion for the number of documents (G) that the target patent document set should have is greater than or equal to Gmin, and less than or equal to Gmax. Since the number of documents in the set of patent documents to be approached approaches the acceptance / rejection criterion value from the smaller one, the criterion is assumed to be greater than or equal to Gmin [A * α (1-β) * γ].

A method for efficiently screening relevant documents included in Patent Document Set 3 is described below. As a specific method of the efficient screening method, Patent Documents for Screening from Patent Document Set 3 are described. Documents that are estimated to have a low content of relevant documents from the hits of a set of documents that are estimated to have a high content of relevant documents for the order of combinations of search items and search keywords in the search expression used to search the set Since the screening operation is stopped when the cumulative number of relevant documents extracted by screening is set to hit the set and clears the criterion value, it is not always necessary to screen all of the patent document set 3 And a search formula for creating a patent document set extracted from the patent document set 3 for screening. The search keyword is a keyword included in the patent document set S created by random extraction from the patent document set 3, and is a keyword that is frequently used in the conforming document set but rarely used in the non-conforming document set. Since it is used, it is possible to hit a document set having a high content of relevant documents.

S15, the first search for extracting the screening patent document set from the patent document set 3 is the search document set 3, the search item is the keyword of the title of the invention, and the search keyword is the patent document set S The keyword selected by the searcher from the keywords in the range determined as the evaluation scale of the conformity / nonconformance document ratio of the name of the invention according to the keyword conformance / nonconformity document ratio table (see FIG. 21 as an example), and the keyword partially matches Create a search expression (see FIG. 23 as an example) for which the search method is an exact match search, and perform a search using the search expression, and hit the document set (the name of the A * S name set in the set) And stored in the storage device 5 as a data object of the patent document set 3), and the extracted relevant document set (the set Literature number of grant name names adapted set is stored in the storage device 5 as data objects in Patent document set 3) in the H1.

The specific creation procedure of the name set of A * S is as follows. The patent document set 3 patent in which the patent document set 3 is stored as a data object in the display device 11 of the patent document analysis apparatus 1 by operating the input device 9. Open the information search screen and operate the input device 9 to specify the keyword of the name of the invention of the patent document set 3 as a search item. Retrieval formula that uses the logical sum of the keyword selected by the searcher and the keyword that partially matches the keyword selected from the keywords in the range determined as the evaluation scale based on the conformance / nonconformity document ratio in the document ratio table (the search method is a perfect match) As an example, see FIG. 23) and execute a search with the search formula to create a name set of A * S.

The method for creating a keyword-specific conformity / non-conformance document ratio table of the patent document set S will be described by taking the keyword of the invention as an example and operating the input device 9 to display the patent document on the display device 11 of the patent document analysis apparatus 1. A property number creation map for creating a document number comparison map for two document sets by keyword in a file in which the set S is stored as a data object is opened, and the input device 9 is operated to create the map as the patent document set S, one column. Select the keyword of the invention name for the object to be displayed in the eye, the object to be displayed in the second column is the relevant document set S1, the object to be displayed in the third column is the non-conforming document set S2, and the input device 9 is operated. Execute map drawing and switch the map to the table screen. Rate table is created, similarly to summarize, the claims, it is possible to create a keyword-specific calibration and incompatible document ratio table similar Patent document set S also Text entirety.

If S16, H1 ≧ Gmin, the name matching set is the target patent document set. If H1 ≧ Gmin is not established, the process proceeds to the next step.

S17, the second search for extracting the screening patent document set from the patent document set 3 is performed by calculating the logic of the search target document set from the patent document set 3 and the A * S name set screened in the previous step. Difference set (the name of the patent document set B is given to the set and stored in the storage device 5 as a data object of the patent document set 3), the search item is a summary keyword, and the search keyword is a summary of the patent document set S The logical sum of keywords selected by the searcher from keywords in a range selected as the evaluation scale based on the relevant / non-conforming literature ratio in the relevant / non-conforming literature ratio table by keyword (see FIG. 22 as an example) and the keyword partially matching the keyword. The search method is a search formula (see FIG. 23 as an example) for a perfect match search, and the search formula is searched and hit. A contributory set (the name of the B * S summary set is given to the set and stored in the storage device 5 as a data object of the patent document set 3) is screened, and an extracted matching document set (summary matching set is added to the set) Is stored in the storage device 5 as a data object of the patent document set 3, and the number of documents is H2.

The specific procedure for creating the B * S summary set is as follows. The patent document set 3 patent in which the patent document set B is stored as a data object on the display device 11 of the patent document analysis apparatus 1 by operating the input device 9. An information search screen is opened and the input device 9 is operated to specify a summary keyword of the patent document set B as a search item. The search keyword is a keyword-specific relevance / non-conformance document ratio table of the patent document set S. A search formula (for example, FIG. 24) in which the keyword selected by the searcher from a range of keywords selected as the evaluation scale based on the ratio of relevant / non-relevant documents is the logical sum of the keyword selected and a keyword that partially matches the keyword. B * S summary set can be created by executing a search using the search formula.

S18, if the sum of H1 and H2 ≧ Gmin, the logical sum set of the name matching set and the summary matching set is the target patent document set, but if the sum of H1 and H2 ≧ Gmin does not hold, proceed to the next step .

S19, the third search for extracting the screening patent document set from the patent document set 3 is performed by dividing the search target document set into the patent document set B and the B * S summary set screened in the previous step. Logical difference set (the name of the patent document set C is given to the set and stored in the storage device 5 as a data object of the patent document set 3), the search item is the keyword in the claims, and the search keyword is the patent document set S The keyword selected by the searcher from the keywords in the range selected as the evaluation scale based on the conformance / non-conformance document ratio of the claim-specific conformance / non-conformance document ratio table, and the keyword that partially matches the keyword, search The method is to create a search expression that is an exact match search, perform a search with the search expression, and hit a collection of documents (C * S is charged for the set) Screened and stored in the storage device 5 as a data object of the patent document set 3), and the extracted matching document set (the name of the claim matching set is given to the set and the patent document set 3 The number of documents stored in the storage device 5 as a data object is H3.

The specific creation procedure of the C * S claim set is as follows. The patent document set 3 patent in which the patent document set C is stored as a data object in the display device 11 of the patent document analyzer 1 by operating the input device 9. Open the information search screen, and operate the input device 9 to specify the claim keyword of the patent document set C as a search item. The search keyword is the suitability / nonconformity by keyword of the claim of the patent document set S The search formula is a logical OR of the keyword selected by the searcher from the keywords in the range determined as the evaluation scale based on the conformance / nonconformity document ratio in the document ratio table, and the search method is a perfect match. When a search is executed by creation and the search formula, a C * S billing set can be created.

S20, if the sum of H1, H2, and H3 ≧ Gmin, the logical sum set of the name matching set, summary matching set, and claim matching set is the target patent document set, but the sum of H1, H2, and H3 ≧ Gmin holds If not, go to the next step.

In order to reduce the number of searches for extracting a screening patent document set from the patent document set 3, the search target document set is the patent document set B, the search items are summary and claim keywords, and the search keywords are patent documents. The keyword selected by the searcher from the keywords in the range selected as the evaluation scale based on the conformity / non-conformance document ratio of the summary of the set S and the claim-specific conformity / non-conformance document ratio table, and the keyword partially matching the keyword There is also a method of creating a search expression that performs an exact match search for the logical sum and search method, and performing a search using the search expression to extract a matching document from a hit document set. In this case, the matching of the hit document set The document content is lower than the B * S summary set fit document content of S17.

S21, the fourth search for extracting the screening patent document set from the patent document set 3 is performed by selecting the search target document set as the patent document set C and the C * S claim set screened in the previous step. Logical difference set (the name of the patent document set D is given to the set and stored in the storage device 5 as a data object of the patent document set 3), the search item is the full text keyword, and the search keyword is the patent document set S The keyword selected by the searcher from the range of keywords determined as the evaluation scale based on the conformance / nonconformance document ratio in the full-text keyword conformity / nonconformance document ratio table, and the logical OR of the keywords partially matching the keyword Create a search expression that is an exact match search, perform a search with the search expression, and hit a set of documents (the full text set of D * S in the set) Named and saved in the storage device 5 as a data object of the patent document set 3), the extracted matching document set (the name of the full-text matching set is given to the set and the data object of the patent document set 3) Is stored in the storage device 5) as H4.

A specific procedure for creating the D * S full-text set is as follows. Patent Document 3 is a patent in which the patent document set D is stored as a data object in the display device 11 of the patent document analyzer 1 by operating the input device 9. Open the information search screen, and operate the input device 9 to specify the keyword of the full text of the patent document set D as a search item. Create a search expression that is the logical sum of the keywords selected by the searcher from the keywords in the range determined using the ratio of conforming / non-conforming documents in the table as an evaluation scale, and the keyword that partially matches the keyword, and that the search method is an exact match, When a search is executed using a search formula, a D * S full-text set can be created.

S22, If the sum of H1, H2, H3, and H4 ≧ Gmin, the target patent document set is the logical sum set of the name matching set, summary matching set, claim matching set, and full-text matching set, but H1, H2, and H3 If the sum of H4 and H4 ≧ Gmin does not hold, the process proceeds to the next step.

S23, the fifth search for extracting the screening patent document set from the patent document set 3 is performed by setting the search target document set as the logical difference set of the patent document set D and the screened full-text set of D * S The name of the patent document set E is given to the set and stored in the storage device 5 as a data object of the patent document set 3, and the relevant document set (screen E is added to the set is screened and extracted). The number of documents in the storage set 5 is given as the data object of the patent document set 3 given the name of the matching set (H5).
S24, if the sum of H1, H2, H3, H4, and H5 ≧ Gmin, the logical sum set of the name matching set, summary matching set, claim matching set, full-text matching set, and E matching set is the target patent document set However, if the sum of H1, H2, H3, H4, and H5 ≧ Gmin does not hold, review from the first step.

Since all the documents in the patent document set 3 are screened in S15, S17, S19, S21, and S23, the sample document set created by random extraction from the patent document set 3 is correctly analyzed. In any case, the target patent map patent document set should be obtained in any one of steps S16, S18, S20, S22, and S24.

Hereinafter, as examples, comprehensive collection of patent documents related to life monitoring such as safety confirmation of elderly people living alone, elderly people, emergency notification, bathroom monitoring, etc. of all relevant documents included in the collected document collection A patent document search method for estimating the number of cases by a sampling method with an allowable error of 10% and capturing 70% or more of the minimum number of the estimated values will be specifically described with reference to FIGS.

A. This is a step of creating a master patent document set 1 by collecting five or more patent documents that differ in at least one of the name of the invention and the applicant assigned to one patent document by an arbitrary method.

The master patent document set 1 is created by connecting the patent document analyzer 1 to the ASP patent database via the communication line 14 by operating the input device 9, confirming the safety of the elderly such as a single person or the elderly, emergency notification, Search patent documents using a small number of keywords related to life monitoring such as bathroom monitoring, save the hit patent documents in the storage device 5 as a file a1, convert the file a1 into a data format that can be processed by patent document analysis software, and save the file a2 From among the patent documents that become masters from the a2, select five patent documents that are different from at least one of the names and applicants of the invention given to one patent document, and select the five patent documents The data object of the file a2 was saved in the storage device 5 as the master patent document set 1 (FIG. 4).

B. The first patent classification table of the master patent document set 1 is created.

This step is a step of extracting a patent classification used in the search formula for increasing the master patent document. As a means for automatically performing the extraction operation of the patent classification using the patent document analyzer 1, the master patent document is extracted. A document number ranking table (FIG. 5) for the first patent classification of set 1 was created and used as a substitute.

The method for creating the first patent classification document number ranking table of the master patent document set 1 is to operate the input device 9 to open the file a2 from the start screen of the patent document analysis software and to rank from the front screen of the file a2. A patent map creation condition setting screen is displayed on the display device 11, the master patent document set 1 is selected as an analysis target object of the ranking patent map creation condition setting screen, the first patent classification is selected as an analysis display item, and an input device 9 is selected. When the creation of a patent map is executed by operation, a ranking map of the number of documents by master patent classification of the master patent document set 1 is created. Therefore, the ranking map is switched to a table by operating the input device 9, and the master patent document set Create and record the document ranking table (Fig. 5) by number of the first patent category. It was stored in the device 5.

C. The keyword of the title of the invention of the master patent document set 1 is extracted.

This step is a step of extracting a keyword used in the search formula for increasing the master patent document. The means for automatically extracting the keyword using the patent document analyzer 1 is the master patent document set. A document ranking table (FIG. 6) by keyword with the name of the invention of 1 was created, and the keywords in the table were used.

The method for creating a keyword-specific document number ranking table for the title of the invention of the master patent document set 1 is as follows. The file a2 stored in the storage device 5 from the start screen of the patent document analysis software by operating the input device 9 The ranking patent map creation condition setting screen for the file a2 is displayed on the display device 11. On the ranking patent map creation condition setting screen, the master patent document set 1 is selected as an analysis target object, and the analysis display item of the invention is selected. When a keyword for a name is selected and the input device 9 is operated to create a patent map, a keyword-specific document number ranking map for the name of the invention of the master patent document set 1 is created. Name of the invention of the master patent document set 1 by switching the ranking map to a table Create a keyword-specific document number ranking table (FIG. 6) stored in the storage device 5.

D, the keywords of the master patent document set 1 by number of keywords in the invention title ranking table, the keywords in the ranking table of the master patent document set 1 are classified by hierarchy from the viewpoint of the modification order of the keywords expressing the category of the invention. This is a step of creating a keyword table, organizing the keywords of the keyword table by hierarchy from the viewpoint of partial matching and the viewpoint of duplication, and creating a keyword table for partial matching search by hierarchy of the master patent document set 1.

The method for creating the keyword table by hierarchy of the master patent document set 1 (FIG. 7) from the keyword number ranking table by keyword of the invention of the master patent document set 1 (FIG. 6) is stored by operating the input device 9. FIG. 6 stored in the device 5 is drawn on the display device 11, and the drawn FIG. 6 is pasted on the EXCEL screen opened on another screen of the display device 11. The left side of the pasted FIG. A table having a NO column, a fourth layer keyword column, a third layer keyword column, a second layer keyword column, and a first layer keyword column is created, and the keyword having the name of the invention of FIG. 6 is described in S4. It was created by arranging according to the creation method of the hierarchical keyword table of the title of the invention of Patent Document Set 1.

Specifically, since the keyword and system of NO1 in FIG. 6 are keywords representing the category of the invention, they are arranged in the first hierarchical keyword column, and the keyword and life watch of NO2 in FIG. Since the keyword modification order is established and the system is arranged in the first hierarchy keyword string, the watch is arranged in the second hierarchy keyword string and the life is arranged in the third hierarchy keyword string.

The remote monitoring of NO3 in FIG. 6 is based on the keyword modification order of a system for remotely monitoring life, and the life is arranged in the third-level keyword column and the system is arranged in the first-level keyword column. Arranged in a two-level keyword string.

In the case of the compound word “watching alone” of NO5 in FIG. 6, since the keyword modification order of the system for watching the life of the solitary person is established, the keyword and the solitary guard are separated from the solitary person and the guard. Since the life arranged in the three-level keyword string can be modified, it is arranged in the fourth-level keyword string.

Among the keywords shown in FIG. 6, the above-described examination is performed on all the keywords to be used in the master patent document increase search formula, and the names of the inventions of the master patent document set 1 shown in FIG. Created a keyword table.

Further, duplicate keywords for each hierarchical keyword column in the hierarchical keyword table (FIG. 7) of the title of the invention of the master patent document set 1 are eliminated, and partially matched keywords are combined into one keyword. Integration, for example, the keyword in the second hierarchical column in FIG. 7, remote monitoring is integrated into monitoring, and there are five keywords and monitoring, so this duplication is eliminated and consolidated into one keyword. A keyword table for partial matching search by hierarchy of the names of the inventions of the master patent document set 1 was created (FIG. 8).

E. Using the keywords of the first patent classification table (FIG. 5) of the master patent document set 1 (FIG. 5) and the keywords of the partial match search keyword table (FIG. 8) of the names of the inventions of the master patent document set 1, the master patent documents This is a step of creating an increase search expression.

In order to create the search formula for increasing the master patent document, as the patent classification search formula 1, the search item of the patent classification is FI, the search keyword is all of the first patent classification table (FIG. 5) of the master patent document set 1 The logical sum of patent classifications, the search method is a forward matching formula (patent classification search formula 1 in FIG. 10), and the keyword search formula 2 is a partial match search by name of the invention of the master patent document set 1 For each hierarchical keyword column in the keyword table for search, the search item is the keyword of the name and summary of the invention and the claim keyword, the search keyword is the logical OR of the keywords in the hierarchical keyword column (FIG. 9), and the search method is a partial match (Formula symbols K1 to K4 in FIG. 10) are created, and the same is true for all of the sets created by sequentially removing one formula from the set of all the created formulas. Create a logical product of the expressions in the middle, create an expression 2 (Equation 2 in FIG. 10) in which all the created logical products are logical sums, the patent classification search expression 1 and the keyword search expression 2 The master patent document increase search formula (search formula of FIG. 10) was created using the logical product of.

F. Search the patent document database using the master patent document increase search formula, convert the hit document set into a data format that can be processed by the patent document analysis software, and screen the data format converted document set to apply This is a step of creating only the master patent document set 2.

The patent document analysis apparatus 1 is connected to the patent database 13 via the communication line 14, and the patent document search of the patent database 13 is performed using the search expression for increasing the master patent document by operating the input device 9. 198 cases are stored in the storage device 5 as the file b1, the file b1 is converted into a data format that can be processed by the patent document analysis software, the file b2 is created, and the file b2 is screened to extract 172 documents as masters The extracted document set was stored in the storage device 5 as the master patent document set 2.

G. A document number ranking table (FIG. 11) for all patent classifications of the master patent document set 2 is created, and a document hit ratio table (FIG. 12) for cumulative patent classifications of the master patent document set 2 is created from the ranking table. Cumulative patent classification at the time when the percentage of document hits by cumulative patent classification in the table exceeds the standard value (set to 80% in this example) confirms the safety of vulnerable individuals such as elderly people living alone, emergency notification, bathroom monitoring The patent classification of the patent document search formula A for comprehensively collecting the patent documents related to life watching such as

A method of creating a document number ranking table (FIG. 11) for all patent classifications of the master patent document set 2 is a file stored in the storage device 5 from the start screen of the patent document analysis software by operating the input device 9. The master patent document set 2 is opened, a ranking patent map creation condition setting screen is displayed on the display device 11 from the front screen of the file, and the analysis object of the setting screen selects and analyzes the master patent document set 2 When a patent map is created by selecting all patent classifications as display items, a document number ranking map for all patent classifications of the master patent document set 2 is created. Therefore, the ranking map is displayed as a table by operating the input device 9. To create a document number ranking table (FIG. 11) for all patent classifications of the master patent document set 2. File King table, and stored in the storage device 5 as data objects of said master Patent document set 2.

The method of creating the cumulative patent category document hit rate ratio table (FIG. 12) of the master patent document set 2 is based on the total patent category document number ranking table of the master patent document set 2 stored in the storage device 5 (FIG. 12). 11) is drawn on the display device 11 by operating the input device 9, and the drawn patent number ranking table of all patent classifications of the master patent document set 2 drawn is copied, and the ranking table is displayed as shown in FIG. A column adjacent to the right side of the ranking table (FIG. 11) for all patent classifications of the master patent document set 2 pasted on the EXCEL sheet drawn on a screen different from the screen on which the ranking table is drawn. After creating a table with the cumulative patent category column, the number of document hits by cumulative patent category column, and the number of literature hits by cumulative patent category ratio column The N-ranked cumulative allowed classification column describes the logical sum of the patent classifications from the 1st ranking to the N-th ranking. The document hit count column by cumulative patent classification includes the FI of the master patent document set 2 and the search keyword. Enter the number of patents in the cumulative patent category column ranked N, execute a search, enter the number of hit documents, and enter the number of hits in the cumulative patent category by the number of hits in the cumulative patent category Describes the ratio of the master patent document set 2 to the number of documents (172), continuously created until the ratio reaches the reference value (80% in this embodiment) or exceeds the ranking N, and cumulative patents of the ranking N The classification is determined to be the patent classification of the patent document search formula A. In this embodiment, N = 1 and the ratio of the number of document hits by cumulative patent classification is 97.7%, which is a reference value of 80. Since exceeded using Index # 1 of Patent Classification G08B25 / only as patent classification of the patent document retrieval formula A.

H. A master content table by keyword (FIG. 13), a master content table by keyword in summary (FIG. 14), and a master content table by keyword in claims (FIG. 15). This is a step of determining a keyword extraction range to be used in the patent document search formula A using the master content rate in the table as an evaluation scale.

Master content table by keyword (FIG. 13), master content table by keyword in summary (FIG. 14), and master content table by keyword in claim (FIG. 15) The method creates a document number ranking table by keyword for the title of the invention of the master patent document set 2 in the same procedure as step C, a summary document number ranking table by keyword for summary, and a document number ranking table by keyword for claims. The created three types of ranking tables are copied and pasted to an EXCEL sheet opened on a screen different from the screen on which the three types of ranking tables of the display device 11 are drawn, and each of the three types of pasted ranking tables is displayed. Next to the right of the number of documents in the master content rate column, In the N-th column content rate column, a table describing the ratio of the number of documents ranked N to the number of documents 172 in the master patent document set 2 was stored as objects FIG. 13, FIG. 14 and FIG. 15 of the EXCEL file A. .

The criterion for determining the extraction range of keywords used in the patent document search formula A is 0% or more in the master content column of the title of the invention, that is, the master content of all keywords and abstracts is 2% or more. The master content is set to 5% or more.

I, all keywords satisfying the master content standard value in the master content rate table for each keyword are collected in the keyword table for hierarchical analysis (FIG. 16) of the master patent document set 2, and further, the patent table The keywords used in the document search formula A are extracted, and a hierarchical keyword table (FIG. 17) of the master patent document set 2 is created.

The hierarchical analysis keyword table (FIG. 16) of the master patent document set 2 is stored in the storage device 5 as a data object of the EXCEL file A on the display device 11 by operating the input device 9. , FIG. 15 shows, the master content rate of FIG. 13 is 0% or more (all keywords), the keyword of FIG. 14 master content rate of 2% or more, and the keyword of FIG. 15 master content rate of 5% or more. Copy and paste and paste it on a single EXCEL sheet, create keywords by aligning the keywords on the vertical axis of the sheet in ascending order, and save the created EXCEL sheet in the storage device 5 as the object of the EXCEL file A, FIG. did.

Next, a method for creating a hierarchical keyword table (FIG. 17) of the master patent document set 2 is as follows. The input device 9 is operated to display the above FIG. 16 on the display device 11. A table having a NO column, a fourth layer keyword column, a third layer keyword column, a second layer keyword column, and a first layer keyword column is created, and the searcher selects the patent from the keywords shown in FIG. By using the keyword determined to be used in the document search formula A, the keyword table classified by hierarchy of the master patent document set 2 is created in the same manner as the keyword table classified by hierarchy of the invention of the master patent document set 1 of S4 is created. (FIG. 17) was created, and the created table was stored in the storage device 5 as an object of the EXCEL file A.

J, a step of creating a hierarchical partial match search keyword table (FIG. 18) of the master patent document set 2 from the hierarchical keyword table of the master patent document set 2.

The creation method of FIG. 18 operates the input device 9 to display the object of the EXCEL file A and FIG. 17 on the display device 11. In FIG. In addition, the partially matched keyword is a single keyword (for example, NO1 old man, NO2 old man, NO6 single old man, NO7 single old man, NO8 single old man in the fourth-tier keyword column are integrated into the old man. 18), the created FIG. 18 was saved in the storage device 5 as the object of the EXCEL file A, FIG.

K, the patent document search formula A (FIG. 19) is created using the patent classification determined in step G and the keywords of the hierarchical partial match search keyword table of the master patent document set 2 in step J.

In order to create the patent document search formula A, as the patent classification search formula 2, the search keyword FI is the search keyword FI by the cumulative patent category of the master patent document set 2 by the cumulative patent category by document hit ratio table (FIG. 12). Cumulative patent classification at the time when the ratio of the number of document hits exceeds a reference value (set to 80% in the present embodiment), G08B25 /, a search method creates a forward matching formula, and the master patent document set as a keyword search formula 2 For each hierarchical keyword column of the partial matching search keyword table (FIG. 18) of the name, summary, and claims of the invention of FIG. 2, the search item is the keyword of the full text, and the search keyword is the keyword of the same hierarchical keyword column. And the search method is partially matched (A1, A2, A3, A4 in FIG. 19), and the logical product of all the created expressions (in FIG. 19) Create a keyword search for formula 2), was prepared the patent document retrieval formula A using a logical product of the patent classification search formula 2 and a keyword search formula 2.

L, the patent document search formula A is used to search the patent document database 13, and the hit document set is converted into a data format that can be processed by the patent document analysis software to create the patent document set 3.

The patent database 13 is connected to the patent analysis apparatus 1 via the communication line 14, and the input apparatus 9 is operated to search the patent document 13 using the patent document search formula A. c1 is stored in the storage device 5, the file c1 is converted into a data format that can be processed by the patent document analysis software, and a patent document set 3 (3639 documents) is created and stored in the storage device 5.

M, the number of relevant documents included in the sample document set S (hereinafter referred to as S1 cases) consisting of the sample document number S randomly extracted to estimate the total number of relevant documents contained in the patent document set 3 by the sampling method And the estimated value of the total number of relevant documents included in the patent document set 3 (the ratio S1 / S = α, where α is hereinafter referred to as the relevant document content rate). Using the allowable error (hereinafter referred to as β), the number of conforming patent documents included in the patent document set 3 (hereinafter referred to as the comprehensive conforming literature number) is A * α (1 ± β). This is the step of estimating that there is.

Since the number of documents in the patent document set 3 is set to 3639 and the allowable error (β) is set to ± 10%, it is sufficient that the necessary number of sample patent documents is 100 or more from FIG. Since 121 references were randomly extracted from the reference set 3 and the relevant reference set was screened to obtain 46 relevant references, the relevant reference content rate α was 46/121. It is calculated as 0.38.

Therefore, the number of relevant documents included in the patent document set 3 was estimated to be 1,244-1521 in 3639 * 0.38 * (1 ± 0.1).

As the random extraction method of the present embodiment, the number of documents in the patent document set 3 is 3639 documents arranged in ascending order with respect to the filing date, and a method of extracting the arranged documents at intervals of 30 is adopted. .

N, the ratio of the number of patent documents that should be included in the target patent document set to the total number of relevant documents included in the patent document set 3 (this ratio is referred to as a capture rate and is hereinafter referred to as γ) γ is determined. This is a step of determining the range of the number of documents that the target patent document set should have as A * α (1-β) * γ ≦ G ≦ A * α (1 + β) * γ.

In this embodiment, since the capture rate is set to 70%, the range of the number of patent documents G that should be included in the target patent document set is 1244 * 0.7 with a minimum value of 871, and the maximum value is 1521 * 0. .7 was 1065 cases, and in the following examination, the criterion for G was 871 which is the minimum value (hereinafter referred to as Gmin).

O, the first search for extracting relevant documents from the patent document set 3 is the search target document set as the patent document set 3, the search item is the keyword of the name of the invention, and the search keyword is the invention of the patent document set S. A keyword selected by the searcher from a range of keywords determined as the evaluation scale based on the conformance / non-conformance document ratio in the name keyword-specific conformance / non-conformance document ratio table (FIG. 21); The search method is to create a search expression (FIG. 23) that is an exact match search, search is performed using the search expression, and the hit document set (the name of the A * S name set is assigned to the set and the patent document set 3 Relevant literature set extracted by screening (stored in the storage device 5 as a data object). Literature number of stores) in the storage device 5 as data object is a step of the H1.

The specific implementation procedure of this step is to operate the input device 9 to open a search screen for files stored in the storage device 5 and patent information of the patent document set 3, and operate the input device 9 to search for documents to be searched. The set is the patent document set 3, the search item is the keyword of the name of the invention, and the search keyword is the keyword whose ratio of the number of conforming document sets to the number of nonconforming document sets in the relevant / nonconforming document ratio table by keyword of the invention name of FIG. And the logical sum of the keywords extracted from the keywords that partially match the keywords (search keyword in FIG. 23), the search method is a perfect match, and when the search is executed, the number of documents in the hit document set and A * S name set 654, the number of documents in the relevant document set collected by screening 654, and the number of documents in the name match set (relevant document rate = 546/65) = 83.5%), and thus became the reviews H1 = 546.

P, H1 is compared with Gmin. If H1 ≧ Gmin, the name matching set is a set of patent documents, and if H1 ≧ Gmin is not established, the process proceeds to the next step.

Since H1 = 546 and Gmin is less than 871, proceed to the next step.

Q, the second search for extracting relevant documents from the patent document set 3 is a logical difference set between the patent document set 3 and the A * S name set (hereinafter referred to as the patent document set). 3 is stored in the storage device 5 as a patent document set B), the search item is a summary keyword, and the search keyword is a match / non-conformance document ratio table (FIG. 22) of the summary of the patent document set S. Create a search expression (FIG. 24) for a search result (FIG. 24), which is a logical OR of a keyword selected by a searcher from keywords in a range determined using the nonconforming document ratio as an evaluation scale and a keyword that partially matches the keyword. A search is performed using the search formula, and the hit document set (the name of the B * S summary set is assigned to the set as a data object of the file patent document set 3). H2 is the number of documents in the relevant document set (saved in the storage device 5) screened and extracted (the name of the summary relevant set is assigned to the set and saved in the storage device 5 as a data object of the patent document set 3). It is a step to do.

The specific implementation procedure of this step is to operate the input device 9 to open a search screen for files stored in the storage device 5 and patent information of the patent document set 3, and operate the input device 9 to search for documents to be searched. The set is a patent document set B, the search item is a summary keyword, the search keyword is a keyword whose ratio of the number of conforming document sets to the number of non-conforming document sets in the summary / matching non-conforming document ratio table by keyword in FIG. The logical sum of the keywords extracted from the partially matching keywords (search keyword in FIG. 24), and the search method is a search formula with the exact match (FIG. 24). The number of documents of 572 is screened, and 572 are screened. The number of documents in the relevant literature set and summary relevant set is 345 (relevant literature rate = 345/5). 2 = 60.3%), and thus became the H2 = 345 items.

R, the sum of the number of relevant documents H1 and H2 and Gmin are compared. If (H1 + H2) ≧ Gmin, the logical sum set of the name match set and the summary match set is the goal, but (H1 + H2) ≧ If Gmin is not established, the process proceeds to the next step.

Since H1 + H2 = 546 cases + 345 cases = 891 and the number of Gmin is 871 or more, it is a patent document set whose goal is the logical sum set of the name matching set and the summary matching set.

By comparing the method employed in Steps O and Q with the method of screening the patent document set 3 as it is, the screening efficiency evaluation scale is the ratio of the number of relevant document extractions to the number of screening documents, the former screening efficiency = (H1 + H2) ) / (Name set of A * S + summary set of B * S) = (546 + 345) / (654 + 572) = 0.73, latter screening efficiency = (546 + 345) / (number of documents in patent document set 3) = 891/3639 = 0.24, it can be seen that the former can perform much more efficient screening.

According to the patent document search method of the present invention, a person who is unfamiliar with the handling of patent classifications, keywords of search target technologies, and synonyms of the keywords can comprehensively collect patent documents in the search target field. As more people can take charge of tasks that are limited to those that have knowledge, using the screening method of the present application, the estimated number of relevant documents included in the comprehensive collection of documents is estimated. Relevant documents can be collected while observing the ratio (capture rate) of the number of collected relevant documents, so that a set of patent documents according to the purpose of use can be created.

Claims (2)

Patent document search method executed on a computer on which patent document analysis software and spreadsheet software are installed, and a set of patent documents stored in a storage unit of the computer is processed by the patent document analysis software and spreadsheet software And collect five or more patent documents that are different from at least one of the patent titles and applicants belonging to the target patent document set and belong to one patent document. A table for organizing the keywords determined by the searcher as a keyword used in the search formula for increasing the master patent document in the step of creating the document set 1 and the keyword used in the name of the invention of the master patent document set 1; The keyword representing the category of the invention is arranged in the first hierarchical keyword column of the table, and the second The keyword that can modify the keyword arranged in the first hierarchical keyword column is arranged in the layer keyword column, and the keyword that can modify the keyword arranged in the second hierarchical keyword column is arranged in the third hierarchical keyword column. As described above, until all keywords determined by the searcher are arranged in any one of the hierarchical keyword strings, the hierarchical keyword string is added, and among the keywords to be arranged, keywords belonging to a plurality of hierarchical keyword strings (compound words) ), Separate the keywords so that the keywords can be placed in a single hierarchical keyword column, and place them in the corresponding hierarchical keyword column to create a hierarchical keyword table for the names of the master patent document set 1 inventions. And steps to
The keywords that can be integrated into the partial match keywords for each hierarchical keyword column are integrated into the partial match keywords for each hierarchical keyword column, and the duplicate keywords are not duplicated. A step of creating a partial match search keyword table by hierarchy of names of inventions in the master patent document set 1 and search items, a logical sum of all the first patent classifications in the master patent document set 1, and search For each hierarchical keyword column in the keyword classification search formula 1 for the patent classification search formula 1 and the master patent document set 1 for the partial classification search keyword table according to the hierarchy, the search items are the title, summary, and claim of the invention. The range keyword and search keyword are the logical OR of the keywords in the hierarchical keyword column. Creates an expression for partial matching, creates a logical product of the expressions in the same set for all of the sets created by sequentially removing one expression from the set of all the created expressions. A keyword search formula 1 that makes all of the obtained logical products the logical sum is created, and a master patent document increase search formula is created using the logical product of the patent classification search formula 1 and the keyword search formula 1 Searching the patent database using the step and the search formula for increasing the master patent document, screening the hit document set to create a master patent document set 2 containing only relevant documents, and the search target set as the master patent document Set 2, search keyword is cumulative patent classification (the cumulative patent classification is a document to which all patent classifications included in the master patent literature collection 2 are assigned the patent classification) When the numbers are arranged in descending order as the sort key, the logical classification of the patent classification with the first number of documents, the logical sum of the first and second patent classifications, and then the first, second and third patent classifications The logical sum of the patent classifications created by including the patent classification one rank lower in the order like logical sum), the search method is a search with forward matching, the total number of documents in the master patent document set 2 of the number of hit documents Search for patent documents to create a collection of patent documents for which the cumulative patent classification at the time when the ratio to the above (hereinafter referred to as the ratio of the number of hits by cumulative patent classification) reaches or exceeds the reference value For each of the step of deciding to make a patent classification of Formula A, the number of documents by keyword of the invention name of the master patent document set 2, the number of documents by keyword of summary, and the number of documents by keyword of claims Calculating a ratio of the master patent document set 2 to the total number of documents, determining a keyword candidate extraction range to be used in the patent document search formula A using the calculated ratio as an evaluation scale, and the keyword candidate extraction range All the keywords within the determined range are collected in one keyword table, and only the keywords determined by the searcher to be used in the patent document search formula A are extracted from the table, using the extracted keywords, A hierarchical keyword table of the master patent document set 2 is created by the same creation method as the hierarchical keyword table created in the master patent document set 1, and the master patent document set 1 is created using the created hierarchical keyword table. Master patent documents in the same way as the created keyword table for partial match search by layer Step 2 for creating a partial partial search keyword table for each hierarchy and search items, search keywords for patent classification are cumulative patent classifications determined in the patent classification of the patent document search formula A, and patent classification with the search method being forward matching For each hierarchical keyword string in the keyword table for partial matching search by hierarchy of the search formula 2 and the master patent document set 2, the search item is the keyword of the full text of the text, the search keyword is the logical sum of the keywords of the hierarchical keyword string, the search method Creates a partial match formula, and the logical product of the formulas of all the created formulas is used as the keyword search formula 2, and the logical product of the patent classification search formula 2 and the keyword search formula 2 is used to The patent document search formula A is created, and the patent document search formula A is used to search the patent database. 3 (assuming the number of documents is A) and at least M from the patent document set 3 (the M cases are sampled when the number of relevant documents included in the patent document set 3 is estimated by a sampling method) Statistically based on the tolerance of the estimated number of reference documents included in the patent document set 3 and the estimated number of relevant documents included in the patent document set 3 (hereinafter referred to as β). A patent document set S created by randomly extracting S sample samples exceeding the number of sampling documents to be determined} is a conforming document set S1 (the number of documents in the document set is hereinafter referred to as S1) and a nonconforming document set. It is classified into S2 (the number of documents in the document set is hereinafter referred to as S2), and S1 / S (the ratio is referred to as relevant document content rate and is hereinafter referred to as α) and β is used to form the patent document set 3 Included in The maximum number of estimated applicable patent documents is the sum of the product of A and α and the product of A, α, and β, and the minimum number of estimated values is the product of A, α, and the product of A, α, and β. The ratio of the number of patent documents to be included in the patent document set 3 to the number of patent documents that the target patent document set should have and the step of estimating that the difference is a difference (this ratio is referred to as a capture rate and hereinafter) (referred to as γ), and the number of patent documents that the target set of patent documents should have (the number of documents is hereinafter referred to as G) is an estimate of the number of relevant documents included in the patent document set 3 The product of the minimum number of values and γ (the number of documents is hereinafter referred to as Gmin) and the maximum number of estimated values of the relevant documents included in the patent document set 3 and γ (the number of documents is referred to as Gmax below). From the patent document set 3 and the step of setting that the number of documents is between In the first search for extracting a joint document, the search target document set is the patent document set 3, the search item is the keyword of the name of the invention, and the search keyword is in the relevant document set S1 including the keyword in the name of the invention. The logical sum of the keywords extracted by the searcher from the keywords in the extraction range determined from the ratio of the number of documents belonging to the number of documents belonging to the non-conforming document set S2 as an evaluation scale, and the keyword partially matching the keywords, the search method is complete Implemented with a formula that matches, the hit literature set name is H1 full set, the relevant literature set created by screening the H1 full set is H1 relevant set, the number of documents in the H1 relevant set is H1, and if H1 ≧ Gmin If the H1 compatible set is the target set of patent documents, and H1 ≧ Gmin does not hold, the process proceeds to the second search step.
In the second search, the search target document set is the logical difference set of the patent document set 3 and the H1 complete set (hereinafter referred to as the patent document set B), the search item is the summary keyword, and the search keyword is the search keyword A keyword extracted by a searcher from keywords in an extraction range determined by using a ratio of the number of documents belonging to the relevant document set S1 including a keyword in the summary and the number of documents belonging to the non-conforming document set S2 as an evaluation measure, and the keyword and part The logical sum of the matching keywords and the search method are implemented with the expression that makes the exact match, the hit literature set name is the H2 complete set, the relevant reference set created by screening the H2 complete set is the H2 compatible set, and the H2 compatible set If the number of documents in H2 is H2, and the sum of H1 and H2 ≧ Gmin, the logical sum set of H1 compatible set and H2 compatible set is the target , And the case in which the sum ≧ Gmin of the H1 and H2 is not satisfied, the process proceeds to third search of the step,
In the third search, the search target document set is a logical difference set of the patent document set B and the H2 complete set (hereinafter referred to as the patent document set C), the search item is the keyword of the claim, and the search keyword is , A keyword extracted by a searcher from keywords in an extraction range determined by using a ratio between the number of documents belonging to the relevant document set S1 including the keyword in the claim and the number of documents belonging to the non-conforming document set S2 as an evaluation scale; The logical OR of the keywords that partially match the keyword, and the search method are performed with an expression that perfectly matches, the hit document set name is the H3 complete set, and the relevant reference set created by screening the H3 complete set is the H3 compatible set, If the number of documents in the H3 compatible set is H3, and the sum of H1, H2 and H3 ≧ Gmin, the H1 compatible set, the H2 compatible set, and the H3 compatible set A patent document set Liwa set is the target, when the sum ≧ Gmin of H1 and H2 and and H3 not satisfied, the process proceeds to the fourth search step,
In the fourth search, the search target document set is a logical difference set of the patent document set C and the H3 full set (the set is hereinafter referred to as patent document set D), the search item is the keyword of the full text, and the search keyword is A keyword extracted by a searcher from keywords in an extraction range determined by using a ratio of the number of documents belonging to the relevant document set S1 and the number of documents belonging to the non-conforming document set S2 including the keyword in the full text as the evaluation scale, and the keyword The logical sum of keywords that partially match, and the search method is an expression that makes the search match completely, the hit document set name is the H4 all set, the matching document set created by screening the H4 all set is the H4 matching set, the H4 If the number of documents in the matching set is H4 and the sum of H1, H2, H3, and H4 ≧ Gmin, then the H1 matching set, the H2 matching set, and the H3 matching set H4 is a patent document set logical union is the target of the adaptation set of, when the sum ≧ Gmin of H1 and H2 and H3 and H4 is not satisfied, the process proceeds to fifth search step,
The fifth search is performed by creating a logical difference set (hereinafter referred to as a patent document set E) of the patent document set D and the H4 all set as a search target document set, and screening the document set. If the document set is H5 compatible set, the number of documents of the H5 compatible set is H5, and the sum of H1, H2, H3, H4, and H5 ≧ Gmin, A search method for a patent document, characterized in that a logical sum set of H5 compatible sets is a target patent document set and the sum of H1, H2, H3, H4, and H5 ≧ Gmin is not satisfied, and is reviewed from the beginning. Patent document search method executed on a computer on which patent document analysis software and spreadsheet software are installed, and a set of patent documents stored in a storage unit of the computer is processed by the patent document analysis software and spreadsheet software And a method of comprehensively collecting patent documents of a specific technology by an arbitrary method to create a patent document set 3 (the number of documents is A), and at least M from the patent document set 3 {the M When the number of relevant documents included in the patent document set 3 is estimated by the sampling method, the number of documents A in the patent document set 3 that is a sampling population and the relevant documents included in the patent document set 3 Sample documents exceeding the allowable error of the estimated number of cases (the allowable error is hereinafter referred to as β), which is the number of sampling documents determined statistically} A patent document set S created by randomly extracting a case is a conforming document set S1 (the number of documents in the document set is hereinafter referred to as S1) and a non-conforming document set S2 (the number of documents in the document set is hereinafter referred to as S2. ), And using S1 / S (this ratio is referred to as “conforming document content ratio” and hereinafter referred to as “α”) and β, the maximum estimated number of conforming patent documents included in the patent document set 3 The number of cases is the sum of the product of A and α and the product of A, α, and β, and the minimum number of estimates is estimated as the difference between the product of A and α and the product of A, α, and β. The ratio of the number of patent documents that should be included in the patent document set to the estimated value of the total number of relevant documents included in the patent document set 3 (this ratio is referred to as a capture rate and will be referred to as γ hereinafter) The number of patent documents that the patent document set should have (the number of documents is described as G below) is The product of the minimum number of estimated values of the number of relevant documents included in the patent document set 3 and γ (the number of documents is hereinafter referred to as Gmin) and the estimated value of the number of relevant documents included in the patent document set 3 The step of setting that the number of documents is between the maximum number of documents and the product of γ (the number of documents is hereinafter referred to as Gmax) and the first search for extracting relevant documents from the patent document set 3 The document set is the patent document set 3, the search item is the keyword of the name of the invention, the search keyword is the number of documents belonging to the relevant document set S1 including the keyword in the name of the invention and the number of documents belonging to the non-conforming document set S2. Expression that the keyword is extracted by the searcher from the keywords in the extraction range determined by using the ratio as an evaluation measure, the logical sum of the keyword partially matching the keyword, and the search method is a complete match In this case, the hit literature set name is H1 complete set, the relevant literature set created by screening the H1 full set is H1 relevant set, the number of documents in the H1 relevant set is H1, and if H1 ≧ Gmin, H1 relevant set Is the target patent document set, and if H1 ≧ Gmin does not hold, the process proceeds to the second search step,
In the second search, the set of documents to be searched is the logical difference set of the patent document set 3 and the H1 complete set (hereinafter referred to as the patent document set B), the search items are the keywords of the summary and claims, and the search The keyword is obtained by the searcher from the keywords in the extraction range determined by using the ratio of the number of documents belonging to the relevant document set S1 and the number of documents belonging to the non-conforming document set S2 including the keyword in the summary or claims as an evaluation measure. The logical sum of the extracted keyword and the keyword that partially matches the keyword, and the search method are executed with the exact match. The hit document set name is the H2 complete set, and the relevant reference set created by screening the complete H2 set is created. If the number of documents in the H2 matching set and the H2 matching set is H2, and the sum of H1 and H2 ≧ Gmin, the logical sum of the H1 matching set and the H2 matching set A patent document set if's target, when the sum ≧ Gmin of H1 and H2 is not satisfied, the process proceeds to the third search step,
In the third search, the search target document set is a logical difference set of the patent document set B and the H2 full set (the set is hereinafter referred to as a patent document set C), the search item is the keyword of the full text, and the search keyword is: A keyword extracted by a searcher from keywords in an extraction range determined by using a ratio of the number of documents belonging to the relevant document set S1 and the number of documents belonging to the non-conforming document set S2 including the keyword in the full text as the evaluation scale, and the keyword The keyword is logically ORed with a partial match, and the search method is an exact match. The hit document set name is the H3 complete set, the relevant reference set created by screening the H3 complete set is the H3 compatible set, the H3 If the number of documents in the matching set is H3, and the sum of H1, H2, and H3 ≧ Gmin, the logic of the H1 matching set, the H2 matching set, and the H3 matching set Set is a patent document set a target, the process proceeds when the sum ≧ Gmin of H1 and H2 and and H3 not satisfied to the fourth search step,
The fourth search is performed by creating a logical difference set of the patent document set C and the H3 all set as a search target document set, screening a relevant document set created from the document set, an H4 compatible set, and a document of the H4 compatible set If the number of cases is H4 and the sum of H1, H2, H3, and H4 ≧ Gmin, the logical sum set of the H1 matched set, H2 matched set, H3 matched set, and H4 matched set is the target patent document set, and H1 and H2 If the sum of H3 and H4 ≧ Gmin does not hold, review from the beginning.
A method for searching patent documents, comprising:

Images