JP2007087252A - Information analysis method, information analysis program, recording medium with the program recorded thereon, and information analysis apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information analysis method capable of accurately and quickly extracting correlation of information units included in a patent database, in acquiring latest literature and patent information. <P>SOLUTION: This information analysis method defines one "representative inventor" for one patent, analyzes correlation between the inventors and organizes the patent for each the "representative inventor", to presume a main patent application strategy of an applicant, as for patent information under a specific retrieval subject, based on the premise that a person who most frequently appears is a key person on the research and development in a group of inventors having mutually joint inventor relationship. A similar method is applicable to a general technical literature information, and useful for quick and efficient information research activities. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an information analysis method and an information analysis program for quickly and easily grasping the whole image of a large amount of document information database such as technical information and patent information searched under a specific search problem. , The recording medium, and the information analysis apparatus.

  As the 21st century is called “the era of knowledge,” a competitive business development cannot be considered without a thorough intellectual property strategy. Especially in the patent field, with the globalization of various industries, a large number of strategic patent applications continue to be filed mainly by large corporations, but depending on the field and target, the number is no longer a steady “handling” survey. May exceed the limit that can be accommodated. Particularly for local SMEs, the burden may be too great, and it is left behind by a fierce intellectual property strategy.

  Although it is a large number of patents of many companies, it is the field research and development "inventor" who is usually directly involved in each research that creates the invention specification and is involved in the application. This information analysis method focuses on the “information unit” in each patent information of the “inventor” and analyzes the correlation of the joint invention etc., thereby quickly and efficiently summarizing the patent application trends. I will try to find it. The same method can be applied to “information units” such as “applicant” and “author” and “affiliation institution” in general technical literature information, and is useful as a method for grasping technical trends from a large amount of literature information. It is.

  If you can get an overview of the technical literature publication trend and patent application trend, you can grasp the research and development trend of the field or company you are paying attention to, and the trend of the central person (= keyman) in that field.・ It will also be easier to find a “gap” in the application strategy. In addition, it becomes easy to grasp the existence of “idle” technical information that is out of the main path of the announcement / application strategy, which may lead to activation of reuse of existing intellectual property. This information analysis method is centered on a system that performs such a procedure quickly and efficiently by computer program processing. Hereinafter, a patent information survey will be described as an example.

  When conducting a patent information search / retrieval retrospectively, the application / publication status of a desired field is often searched using a specific applicant, a specific classification code, a specific keyword, and the like. However, in those cases, unless there is a great deal of attention in the setting of search conditions or a search specialist who has a high level of technology, it is common to include “search leakage” and “noise”.

  This method is different from the above, and pays attention to a researcher (inventor) involved in research and development. A researcher (inventor) is usually an expert on a specific research and development theme, is involved in a certain research and development theme over the medium to long term, and creates and applies for a patent as a result. Next, when creating and applying for patents, there are usually many cases related to related research and development, and it is not common to create and apply for patents in completely different fields from a certain time. Also, research and development is usually conducted in groups rather than alone, and the group also usually has a leader, and is often a key person in a series of research and development and patent applications. If it pays attention to, it will be an index to grasp the outline of strategic application.

  As a method for visually grasping an outline from many literature / patent information and taking out a specific information system in an easy-to-understand manner, for example, a method (see Patent Literature 1) is known. Further, as a method capable of examining the document search result in a short time regardless of the field, a method (see Patent Document 2) is known.

Japanese translation of PCT publication No. 2002-532788 JP 2005-215765 A

  In order to secure the results of R & D through investing as an in-house right, and to avoid infringing on other companies' patents and causing unnecessary disputes, it is impossible to always keep track of the latest published patent information. However, in recent years, the number of patents published in Japan has risen to more than 400,000 annually, and the number of applications is too large to be dealt with only by hand search. How to make research work more efficient and faster is a challenge for those involved in the research department.

  It is said that the usage aspect of patent information mainly consists of (1) rights information, (2) technical information, and (3) management information. These usage aspects are all important for business development, but if you want to analyze the technology trends of a specific industry or other companies, you should use keyword search or technology classification code (IPC, F-term) search. Is common. However, keyword search is easy to search with intuitive words, and even an amateur can do it, but if the words expressing the exact concept are not simple, there will be many search omissions or search results (noise) other than the intended purpose. It is easy to increase. In addition, in searches using technical classification codes such as IPC and FI, searches with less leakage and noise are generally possible according to the defined classification codes. There are other caveats, such as variations in how to apply.

  And no matter how cleverly controlled keyword terms and technical classifications are used, it is often impossible to avoid omissions because they cannot express all of the contents of patents. This information analysis method is one of the methods to make research work more efficient and quick, and to compensate for the weaknesses of these general search methods and to grasp the technological trends of other companies as a comprehensive information system relatively easily. Developed as

  In addition, in the case of an actual patent database, particularly in the case of a patent database in Japan, in the individual information of the inventor, applicant, etc., the name and company name inconsistently written in Kanji (Kanji old font New typeface), change of applicant name due to merger / division of companies, typographical errors and omissions when going through agent (patent attorney), inventor when foreign company applies for Japanese language conversion, applicant's katakana notation There are many technical problems that impede database processing as character string information, such as inconsistencies in the abbreviations and abbreviations of the abbreviations and transposition of notations. In the information analysis method of the present invention, it is possible to perform information analysis extremely efficiently by combining a “correction procedure including a semi-automatic reasoning / correction algorithm” that can cope with these problems. The work to extract the correlation of the information units contained in was realized accurately and quickly.

The information analysis method of the present invention includes:
A. One or more information units X in a data record in the database are interfered with all databases together with subordinate information consisting of zero or more information units (hereinafter referred to as coexistence information units) other than X included in the same data record. Tally processing means for tallying,
B. When the aggregated information units are arranged as a list in descending order of the number of appearances in all databases, one information unit with a certain rank N is one of the coexistence information units subordinate to the information unit with a higher rank. If not, the dependency definition processing means that handles the information unit of the rank N as one of the representative information units,
C. Among one or more information units in one data record, B. If there is one of the representative information units obtained in step (1), processing means for defining it as one piece of information characterizing one data record, or D. Among one or more information units in one data record, B. If one of the representative information units obtained in step 1 does not exist, the information unit having the highest number of appearances in all the databases among one or more information units in the one data record is included in the one data record. And a processing means for defining the data as one of information that characterizes this one data record,
E. [C. Or D. The relation of one representative information unit corresponding to one data record obtained by the processing means in all the databases for each representative information unit and the classification, size, etc. of other fields included in the data record. A processing means to arrange and manage,
F. B. And [C. Or D. ] For each representative information unit in a tree (tree) so that the correlation between one representative information unit obtained by the processing means and the coexistence information unit subordinate to it can be visually understood. Processing means for display output;
It is equipped with.

  In the information analysis method of the present invention, one of the processes for displaying and outputting the analysis result of the information analysis method of the present invention in a tree shape (tree shape) is described in a markup language such as HTML or XML, and is displayed. The number of occurrences for each information unit is displayed in a graph together with a numerical value display, and it is possible to visually identify whether each information unit is at the head position in the corresponding information unit group in the literature or patent. The display is made.

  The information analysis method of the present invention is essential in the information unit in the data record to be analyzed by reversing the name components such as new and old fonts, typographical characters, omissions, first and last names, and inconsistent Romanization notation in Japanese. When the same information unit is described as another similar information unit, a correction procedure is performed through an algorithm that infers and corrects it semi-automatically.

  In the information analysis program of the present invention, the procedure of the information analysis method of the present invention is realized by a computer program.

  The recording medium of the present invention is one in which the information analysis program of the present invention is electronically recorded and can be read by a general-purpose computer.

  The information analysis apparatus of the present invention is such that the information analysis program of the present invention is operated by a computer and the information analysis method of the present invention can be used.

  The network-compatible information analysis apparatus of the present invention is operated by a server connected to the computer network, and the information analysis program of the present invention and its attached program are transmitted from the computer on the user side connected via the computer network. The analysis method and the analysis result can be used.

  The basic principle of the information analysis method of the present invention is a simple method, and contents that can be grasped manually by a database of about several tens of cases. However, as the number of records in the database increases, the work becomes dramatically complicated, and if it is hundreds or thousands, it becomes virtually impossible to analyze. In the information analysis method, most of the above operations are performed by a program process in a very simple and short time. The original database can be easily obtained from various literature / patent databases on the Internet, CD-ROM media, etc., and the database format may be prepared as a general-purpose “comma separated (CSV) format” file. Therefore, the load of preparation work is not large.

  When a person involved in R & D or patent management watches about the publication of a particular field or the status of patent publication, look at the literature, patent abstract information, etc. that are regularly distributed, such as the SDI (Selective Information Distribution) service. In many cases, we will collect literature and patent specifications for those that are of interest and conduct a thorough reading survey. Such a method is necessary to obtain the latest individual information, but it is often easy to fall into the situation of “looking at the tree and not looking at the forest”. This information analysis method focuses on the authors belonging to each organization and the inventor belonging to each applicant, and analyzes the database together with the correlation between the author and the inventor, so that "view the forest" can be easily done. It is possible to provide an easy-to-understand and easy-to-understand information system that could not be easily obtained by conventional literature / patent database search methods. Hereinafter, an example in which the information unit of the present invention is the inventor of a patent will be mainly described.

  This information analysis method requires at least 1 for patent information searched under a specific search problem, such as patents of applicant A for the past 5 years or patents for the technical field related to product B for the past 3 years. ) Publication number, 2) Title of invention, 3) All inventor, 4) Applicant's database is created, and all patents are analyzed for joint invention relationships between the inventors (= information units) (= Aggregating process on coexistence information unit) and trying to determine the R & D organization chart on the patent application strategy (= Dependency definition process). Of the inventor groups (hereinafter referred to as R & D groups) that have a joint invention relationship with each other, the person who has the highest number of appearances as an inventor is the central person in the R & D strategy. Define one “representative inventor” for each patent (= representative information unit), classify patents for each “representative inventor”, and propose an overview of the applicant's main patent application strategy. We are going to measure.

  Specifically, patent information analysis is performed as follows. It is the front-line researchers who are engaged in development research of the applicant (= company) and are creating patents (especially their research and development groups) that control and reflect the application trends of each applicant who applies for a patent. The central figure, key man, technical expert = “representative inventor”).

  For example, suppose there is a person named M who has been involved in research and development of polyvinyl alcohol (Poval), one of K's main business fields, for many years and has written many patents. The fact that Mr. M suddenly became involved in a central patent related to artificial leather, a completely different product field from a certain year, and in a few years, he was involved in another invention related to drug development. In general, hard to think. Therefore, it is expected that following the patent application track of Mr. M and his main joint inventor, it is possible to quickly and accurately grasp the main technical trends related to the Poval of Company K.

The flow of data processing is shown based on the following example.
[1] First, a database as shown in the following table is prepared as a “comma separated value (CSV) format” file.

Publication number Invention name Inventor Applicant 07-000100 Production method of polyvinyl alcohol A, B, C, DK Company 07-000200 Composition of polyvinyl alcohol F, G K Company 07-000300 Water-soluble resin paint D, C, E Company K 07-000500 Polyvinyl alcohol film F, H Company K 07-000600 Water-soluble resin Company B, A, D Company K

[2] There are a total of 14 inventors related to these five patents, and a net of 8 people. Each of them is listed in the order of the number of appearances together with the persons who are related to the joint invention, as shown in the following table.

Inventor Number of appearances Co-inventor
D 3 A, B, C, E
A 2 B, C, D, E
B 2 A, C, D, E
C 2 A, B, D, E
F 2 G, H
E 1 A, B, C, D
G 1 F, H
H 1 F, G

[3] Based on the above definition, the “representative inventor” is defined, and the organization chart estimated from the joint invention relationship and the subordinate relationship is drawn as shown in the following tree diagram (the number of appearances and the number of appearances are the same in parentheses) In case of character code order). The black circle (●) notation indicates what is described as “first inventor” in the group of inventors of one patent, and the white circle (○) indicates not, and this graph notation (the total number of circles is It is possible to estimate the position of each inventor in terms of technical content by the number of appearances). That is, if the “representative inventor” is the central figure in a series of related patents, the “first inventor” is the actual central figure in that single patent. According to this tree diagram, Mr. D is the central figure of the project, but is in a managerial position away from the R & D site, while Mr. F is the central figure of the project and the research and development leader who is directly in charge. However, it can be estimated.

Representative Inventor Joint Inventor D (3) ┬ A (2)
● ○○ Ｂ B (2) ● ○
├ C (2) ○○
└ E (1) ○
F (2) Ｇ G (1) ○
●● └ H (1) ○

[4] Finally, the patent is classified for each “representative inventor” to characterize the patent (the following table). By reviewing the abstracts and specifications of each patent, the trend of application for patents related to Poval of K Company is grasped. Furthermore, Mr. D is the central person in the project related to the manufacturing process, and Mr. F is the central person in the project related to the application development. And the like are obtained.

Representative Inventor No. Publication number Name of invention Applicant D 1 07-000100 Production method of polyvinyl alcohol Company K
2 07-000300 Water-soluble resin paint Company K
3 07-000600 Water-soluble resin Company K F 1 07-000200 Composition of polyvinyl alcohol Company K
2 07-000500 Polyvinyl alcohol film Company K

  If there are a large number of “representative inventors”, a summary of patent application trends of the company can be grasped more clearly by displaying a pie chart based on numerical data. Furthermore, if the same analysis is performed at regular intervals and a comparative study is performed in time series, the transition of research and development themes can be estimated.

  The above is the basic principle of this information analysis method. This is a very simple method, and it can be grasped manually if it is a database of several tens of cases, but in the actual patent database processing, it was expected due to some uncertainties in character processing Often results are not obtained. For example, there are at least three old characters “Saito, Saito, Saito” in the sex of “Saito”. In the database obtained as a result of the search, “Taro Saito” and “Taro Saito” If there are two people, even if the information in the other fields in the database, that is, the applicant, the joint inventor, and the contents of the application, are found to be the same person, they will be different in terms of computer processing. In the information analysis process, it exists as another inventor, and becomes a factor that gives an analysis result far from reality. Similarly, sexual changes due to marriage, etc., and typographical errors and omissions due to mistakes made by simple intellectual property managers and agents are the same factors.

  In addition, when a foreigner is an inventor, it is necessary to use katakana notation when filing an application in Japan, but in the five examples shown in the following list, all are the same person, Due to differences in handling by application agents such as differences in separators (blanks, commas, dots), inconsistencies in katakana conversion standards, inconsistencies in abbreviations, transposition of first and last names, etc., processing as a separate person in terms of computer processing Will be.

David Sea Mage Yakak David, Sea, Mage Yakak David Sea Mageyakak David Sea Megiyakak Mageyakak David Sea

  The same problem exists in the case of information analysis for applicants, and is a technical problem that needs to be taken into account when the corporate name is changed due to merger, division, etc. In the information analysis method of the present invention, prior to the aggregation of information units, through the “correction procedure including a semi-automatic inference / correction algorithm”, the essentially same information units as described above are found and corrected at an early stage. • A process for control is incorporated. This “semi-automatic reasoning / correction algorithm” stores all information units included in all data records in the target database in a control learning table, compares them with each other in character units, and has a similarity of more than a certain percentage. If approved, it includes a mechanism to prompt the correction by notifying the operator.

  In addition, after completion of data collection, data analysis is performed and verification processing is performed as appropriate. If there is an information unit that can be regarded as the same information unit from surrounding information, additional correction and control processing can be performed. is there. Such a control process can be applied to the subsequent process with a learning function once, and the task of extracting the correlation of information units included in the patent database is performed accurately and quickly. Has contributed. FIG. 1 shows these operations as a flowchart.

  This information analysis method is used not only in stand-alone processing, but also in the environment connected to the network as shown in FIG. 2, in response to a search request from an unspecified number of users, It is also possible to provide a relationship. Hereinafter, the flowchart (FIG. 1) will be described in more detail.

  FIG. 1 shows the flow of the information analysis method in the present invention. As the data of the analysis source, patent information obtained from the Internet or commercially available information services is aggregated under a specific search problem, and at a minimum, 1) the publication number, 2) the name of the invention, 3) all inventors, 4 ) Save on the recording medium as a “Comma Separated Value (CSV) Format” file consisting of the applicant's four items (Function 1).

  Next, this “comma separated value (CSV) format” file is converted as table information in the database, and the above four items are recorded as independent fields, together with work fields having attribute information necessary for data processing. Store (functions 2 and 3).

  A field including an information unit is divided into individual information units and recorded together with attribute information of coexisting information units. At this time, the name control algorithm (function 5) for each individual information unit is used for semi-automatic search and inference, and correction candidates are automatically presented to the operator. With the control learning table (function 6) for inference, this work need only be performed once for one type of information unit, and is automatically processed from the next time onward. Information obtained by the aggregation process (function 4) is stored in the aggregation result storage database (function 7).

  The information units that have been aggregated are evaluated for joint invention relationships and subordinate relationships by analysis result evaluation processing (function 8), and supplementary information unit name control (function 5) is repeated as necessary. Re-execute the aggregation process (function 4) to obtain a more accurate analysis result.

  When the aggregation is finally completed, this result is output as a list for each representative information unit, or output in an HTML file format as a diagram showing a tree-like correlation between the representative information unit and the coexistence information unit (function 9).

  The analysis database (function 3), the aggregated result storage database (function 7), and the analysis result storage database (function 10) can be reproduced and displayed by the analysis result browsing process (function 11), as shown in FIG. By constructing it as a service function via the network as shown, it can be provided as an information service to users at remote locations.

  As described above, the processing procedure when the information analysis method of the present invention is applied to patent information has been described. However, this method is not limited to patent information, and can be easily applied to general technical literature information. is there. The target language is not limited to Japanese, but the same procedure can be applied to foreign languages such as English.

  As an example, among the published patents, published patents, republished patents, and registered utility models issued between January 6 and August 4, 2005, the search subject “Summary or Claim,” “Photocatalyst The result of applying this information analysis method to 617 documents extracted under “including the keyword“ ”is shown. As a result, the following number of cases was obtained, and a list (individual names are concealed) of the inventors (including the inventor, the same shall apply hereinafter) ranked higher in the number of appearances is shown in FIG.

Number of references: 617 Total inventors: 1,626 Net inventors: 1,118

  Based on the results of this aggregation, the inventors analyzed the joint invention relationship and subordinate relationship, and defined one “representative inventor” for each patent (including utility model, the same applies hereinafter). FIG. 4 shows a rearranged list of patents. Among joint inventors, those with an asterisk (*) at the end are subordinate to another joint invention tree structure with the joint inventor as the leader (= “representative inventor”). (It is omitted in this specification for the sake of space).

  Finally, the relationship between the “representative inventor” and the co-inventor is displayed as a tree diagram, and information on whether each inventor appears as the “first inventor” in the patent described as the inventor is displayed. FIG. 5 is a graph showing the number of appearances. By displaying in this way, what kind of inventor group (= R & D group) belongs to what applicant in the searched subject, under what application trend (= application strategy) In addition, it is possible to grasp an outline of what patents are being applied for. And who is the “representative inventor” (= manager in charge of the R & D group) and who is the true central figure from the situation of the “first inventor” (= person in charge of the R & D site) It is also possible. These pieces of information themselves are valuable information that is difficult to obtain by the conventional method, but based on the information, if a search is performed again with improved search conditions, a more accurate patent information search can be performed.

It is a flowchart of the information investigation method of this invention, the program which implements this, and an apparatus. It is explanatory drawing at the time of utilizing the information investigation method of this invention, the program which implements this, and an apparatus via a network. It is a list (individual name is concealed) of the inventor ranked in the top appearance frequency obtained as a result of Example 1 of the present invention. It is a part of the patent list (individual name, patent number, applicant name is concealed) for each representative inventor obtained as a result of Example 1 of the present invention. FIG. 5 is a part of a tree diagram (individual names are concealed) showing the correlation between the representative inventor and the joint inventor and the number of applications obtained as a result of Example 1 of the present invention.

Claims (9)

A. One or more information units X in a data record in the database are interfered with all databases together with subordinate information consisting of zero or more information units (hereinafter referred to as coexistence information units) other than X included in the same data record. Tally processing means for tallying,
B. When the aggregated information units are arranged as a list in descending order of the number of appearances in all databases, one information unit with a certain rank N is one of the coexistence information units subordinate to the information unit with a higher rank. If not, the dependency definition processing means that handles the information unit of the rank N as one of the representative information units,
C. Among one or more information units in one data record, B. If there is one of the representative information units obtained in step (1), processing means for defining it as one piece of information characterizing one data record, or D. Among one or more information units in one data record, B. If one of the representative information units obtained in step 1 does not exist, the information unit having the highest number of appearances in all the databases among one or more information units in the one data record is included in the one data record. And a processing means for defining the data as one of information that characterizes this one data record,
E. [C. Or D. The relation of one representative information unit corresponding to one data record obtained by the processing means in all the databases for each representative information unit and the classification, size, etc. of other fields included in the data record. A processing means to arrange and manage,
F. B. And [C. Or D. ] For each representative information unit in a tree (tree) so that the correlation between one representative information unit obtained by the processing means and the coexistence information unit subordinate to it can be visually understood. Processing means for display output;
An information analysis method characterized by comprising: 2. The information analysis method according to claim 1, wherein all of the information units according to claim 1 are in a specific field in a data record in the database. The information analysis method according to claim 1 or 2, wherein the information unit according to claim 1 or 2 is one of the following.
A. Author B. in literature information Affiliated organization in literature information Inventor D. in patent information Applicant in patent information One of the processes for displaying and outputting the analysis result by the information analysis method according to claim 1 in a tree form (tree form) is described in a markup language such as HTML or XML, and is displayed information unit The number of appearances for each information is displayed in a graph together with a numerical display, and each information unit is displayed in a group of information units in the corresponding document or patent so that it can be visually identified whether it is at the head position. The information analysis method according to claim 1, wherein the information analysis method is provided. 5. Information units according to claims 1 to 4, wherein information units that are essentially the same are similar due to new and old typefaces, typographical errors, omissions, inversion of name components such as first and last names, and inconsistency in Romanized notation at Japanese localization. 5. The information analysis method according to claim 1, further comprising a step of performing correction through an algorithm for inferring and correcting the information unit when the information unit is described as another information unit. 6. An information analysis program, wherein the procedure of the information analysis method according to claim 1 is realized by a computer program. 7. A recording medium in which the information analysis program according to claim 6 is recorded electronically and is readable by a general-purpose computer. An information analysis apparatus according to claim 6, wherein the information analysis program according to claim 6 is operated by a computer, and the information analysis method according to claims 1 to 5 can be used. The information analysis method according to any one of claims 1 to 5, wherein the information analysis program according to claim 6 and the program attached thereto are operated by a server connected to a computer network, and from a user computer connected via the computer network, A network-compatible information analysis apparatus characterized in that the analysis result can be used.

Images