JP3929418B2 - Information search program and medium on which information search program is recorded - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a system for investigating a target material, etc., and in particular, an investigator actually reads the contents of a document, such as an investigation of an objection material related to industrial property rights, etc., a prior application search, etc. The present invention relates to a system that assists investigations that need to be conducted and facilitates investigation work.
[0002]
[Prior art]
Conventionally, a patent information database search system that can search patent information and the like by machine search has been developed and used. Taking a patent information database as an example, it is constructed by storing patent information generated for each patent application as a database in the storage means of the host computer.
[0003]
In addition to general bibliographic items, such patent information databases store various keywords (for example, free keywords, fixed keywords, F-terms, etc.) for each file. The patent information to be searched can be searched. In addition, in addition to the bibliographic items described above, CD-ROM format patent gazettes issued in recent years, CD-ROMs related to patent information edited by manufacturers (hereinafter referred to as electronic patent information), etc. The specification, abstract and drawings are also stored as electronic information, and the corresponding patent information can be searched based on various search items.
[0004]
Here, when trying to investigate invalid materials or opposition materials using the above-mentioned patent information database or electronic patent information, the researcher creates a search formula and sees the abstract of the contents that hit the search formula. If necessary, the full gazette is read to determine whether it can be used as invalid or opposition material. As a result of this search, when the number of hits is small, the publication is directly read. If the corresponding publication is not found, it is repeated to recreate the search formula and read the hit publication. The prior application search for the invention to be applied will be performed in the same procedure.
[0005]
Here, the patent search process will be described in some detail. First, the investigator understands the contents and features of the present application, and combines the classification symbols and keywords stored in the above-mentioned patent database and electronic patent information to obtain, for example, several tens to hundreds of target technologies. Create a search expression so that it is included in the answer set. Of course, the number of hits as a result of the search is small, and it is only necessary that the desired technology is covered in this hit data set. Often, similar technologies in the field are not extracted. For this reason, the researcher usually creates a search formula so that the target technology is included in the hit data set, with tens or hundreds at most as a guide. When a set exceeding several hundreds is created, it takes a lot of time for subsequent screening, and it is difficult to perform an efficient search. However, if it is certain that the target similar technology has been extracted, the abstract or the gazette will be read for all the hit applications, and the similar technology will be searched. This is done manually. If a similar technique cannot be extracted after reading a large number of publications, the search formula is recreated and the same operation is repeated. The investigation work continues until it can be confirmed that no similar technique can be extracted no matter how much searching. Especially in the case of publicly known cases, objections, and invalid material investigation, in addition to the time required to create the search formula, it will add work time for people to read the machine search results carefully, examine the contents further, and discard them. Become. Therefore, it takes a lot of time to complete the survey, and the cost of the survey becomes great. Even for survey experts who are proficient in classification systems and survey methods, the fact of survey work is that it takes a lot of time to examine and process the results of machine search.
[0006]
By the way, there is a pattern matching as a general search processing method widely used conventionally. This includes a keyword used in a search, a search expression, etc. (hereinafter abbreviated as “search keyword”) and a keyword registered for each search unit such as a document / document (hereinafter abbreviated as “registered keyword”). And whether or not the search unit is to be extracted as a search result based on whether or not they match (completely or partially). Therefore, if the search keyword and the registered keyword are inconsistent (completely or partially) but have a synonym or the like (including synonyms and the like), there is a problem that the search cannot be performed. That is, since there are so many terms in the expression of the technical idea, it is very difficult to extract a target similar technique in the machine search process by pattern matching.
[0007]
As an example of this pattern matching, when the character string “primary key” is designated as the search keyword, the registered keywords such as “primary key”, “key”, “key item” are the characters “primary key”. Since it includes at least a part of the column, it is determined as a similar keyword. However, registered keywords that are technical synonyms such as “main item” and “search item” cannot be extracted by search when the character string “main key” is designated as a search keyword.
[0008]
Further, since the above-described registered keyword is used for a search act, it needs to be registered in advance for each search unit such as a document and a document, and it takes time for registration. (Hereafter, this effort is referred to as “keyword registration effort”.)
[0009]
In addition, as a method for determining the registered keyword, a part of a keyword having a characteristic for each search unit such as a document or a document is extracted, or an administrator sets some keywords for each search unit. ing. For this reason, in general, a registered keyword often represents only a part or one aspect of a search unit such as a document / document. (Hereafter, this point is referred to as “partial expression of registered keywords”.) Therefore, when a keyword or the like constituting a search unit other than the registered keyword is designated as a search keyword, one of the registered keywords described above is used. Due to the partial expression, the search unit is not extracted as a result of the search.
[0010]
Furthermore, as in the example of the “primary key”, it is necessary to use the registered keywords that are technical synonyms and the like in a unified manner. (Hereinafter, this point is referred to as “Necessity of Unification of Registered Keywords”.) As an example of the necessity of unification of registered keywords, if the term “primary key” is used as a registered keyword, For example, the registered keywords “item” and “search item” cannot be used.
[0011]
However, if the unification necessity of this registered keyword is not observed, it is not preferable for the investigator who conducts the search, especially when judging novelty or inventive step of industrial property rights etc. The point will become more apparent.
[0012]
Here, novelty refers to whether a particular invention is new depending on whether a particular invention and one or more cited inventions are recognized and a difference occurs when they are compared. It is a patent requirement to judge whether or not. In the determination of novelty, even if the inventions are conventionally recognized as being the same invention by visually confirming the contents by an investigator (determined as having no novelty), they are cited as specific inventions. When a technical term or the like having the same meaning is used as a registered keyword, a highly accurate search may not be performed. In the domestic patent database that is prevailing at the time of filing of the present application, F terms, international patent classifications, FI symbols, and other technical contents are given as thesaurus symbols. However, since a new invention cannot always be classified into an old technical system, it is difficult to hit a patent or the like with a complete focus even if keywords and classification symbols are combined. Furthermore, when a paper or general technical document with a small number of classification symbols is to be investigated, it is difficult to hit an appropriate document by machine search from a data set narrowed down to a small number. Therefore, it is necessary to create a data set of a certain number of cases, and the investigator should examine whether or not there are relevant documents in the data set.
[0013]
Also, the inventive step is to identify a specific invention and one or more cited inventions, and to compare the two to clarify the points of agreement and differences between the items for specifying each invention. The patent requirement is determined by whether or not the difference is easy for those skilled in the art. In this novelty judgment, an invention that is cited as a specific invention, even among inventions that have been identified as having little difference by a visual inspection by an investigator (determined that there is no inventive step) However, when other technical terms having the same meaning are used as registered keywords, appropriate search and extraction may not be performed.
[0014]
In general, there are a lot of creators of documents such as specifications of industrial property rights, etc. As in the example of the “primary key” above, technical terms that have the same and similar meanings have a wide variety of terms. It is expressed by. As described above, since the necessity of unifying registered keywords is not taken into account, there is an inconvenience that it is very difficult to search for the same invention or a similar invention by the conventional search method.
[0015]
Further, when the registered keyword for each search unit is a part of the search unit, etc., there is a disadvantage that it is difficult to conduct a high-accuracy survey due to the partial expression of the registered keyword.
[0016]
For these reasons, there is a demand for the emergence of a system that can quickly and accurately extract survey target materials such as reference materials. In addition, there is a demand for an apparatus that can automatically or simplify novelty and inventive step analysis based on specific patent information.
[0017]
In this regard, in [Patent Document 1], a plurality of pieces of information to be searched are stored in advance in association with some keyword groups included in the plurality of pieces of information, and keywords of the search target information are stored. There is disclosed a method of performing a search from a plurality of pieces of information to be searched based on a keyword that is aggregated and displayed and selected by an investigator.
[0018]
However, this method requires time and effort for registering the association of keyword groups for each of a plurality of pieces of information to be searched in advance, and the time and effort for keyword registration cannot be solved. Furthermore, since a specific keyword group included in a plurality of pieces of information is used for the search, it is not possible to solve the partial expressibility and necessity of unification of registered keywords.
[0019]
In [Patent Document 2], for each document unit to be searched, a pattern expressing the presence / absence of association with a predetermined keyword group is created, and the instructor instructs the pattern. A method of searching for a document unit having a pattern similar to the pattern is disclosed.
[0020]
However, there is an inconvenience that it is difficult to reuse a past pattern when a predetermined keyword group must be determined in advance or when the predetermined keyword group is changed in the middle. Further, like [Patent Document 1], it takes time and effort to register pattern correspondence for each of a plurality of pieces of information to be searched in advance, and the time and effort for keyword registration cannot be solved. Furthermore, since a pattern expressing the presence / absence of association with a predetermined keyword group is used for the search, it is not possible to solve the partial expression and necessity of unification of registered keywords.
[0021]
[Patent Document 1]
JP-A-9-73453
[0022]
[Patent Document 2]
JP-A-61-182131
[0023]
[Problems to be solved by the invention]
As described above, in the conventional search method as described above, by simply specifying a specific keyword or search expression, patent information including a keyword or the like corresponding to this search expression is extracted. For this reason, judgment of novelty, inventive step, etc. is generally carried out by the investigator actually confirming the contents of the extracted patent information visually, etc., and the extracted patent information (patent publication etc.) ) Has a lot of trouble in understanding and grasping it, and it is difficult to judge novelty and inventive step in a short time and appropriately.
[0024]
In addition, the act of determining an appropriate keyword and search formula according to the data set to be investigated must be performed by the investigator, and there is a disadvantage that the burden on the investigator is heavy.
[0025]
Furthermore, the following inconveniences exist as problems that are not solved in [Patent Document 1] and [Patent Document 2].
[0026]
First, since the trouble of keyword registration cannot be solved, there is a disadvantage that the keyword registration must be performed every time data to be investigated is generated.
[0027]
Furthermore, since the partial expressivity of the registered keywords cannot be solved, there is a problem in that a survey is likely to be omitted and a highly accurate survey is difficult to perform.
[0028]
In addition to this, if the necessity of unifying registered keywords is not considered in the data set to be surveyed, it is difficult to conduct highly accurate surveys because multiple technical terms such as synonyms are used. There was an inconvenience.
[0029]
OBJECT OF THE INVENTION
The present invention improves the inconveniences of the prior art and [Patent Document 1] and [Patent Document 2], particularly solves the trouble of keyword registration, and does not require the trouble of registering registered keywords in advance. An object is to provide a suitable means capable of performing a search from a data set to be investigated.
[0030]
In addition, in order to solve the partial expression of registered keywords and conduct high-accuracy surveys, the purpose is to provide a suitable means by which the investigator can select an appropriate search keyword according to the data set to be surveyed. And
[0031]
Furthermore, even in the case where the necessity of unification of registered keywords is not considered in the data set to be investigated, it is preferable that the researcher can appropriately select technical terms having synonyms and the like as search keywords. It aims to provide a simple means.
[0032]
[Means for Solving the Problems]
In order to achieve the above object, a system according to the present invention includes storage means for storing reference data including an information block composed of one or more keywords and search target data including a plurality of information blocks composed of one or more keywords. ing. The system further includes an input unit for inputting an association between a keyword constituting the reference data and a keyword constituting the search target data. Further, the system includes a computer that executes processing for searching for similar information blocks from the search target data based on the information blocks of the reference data. In addition, the system includes display means for displaying the search results. In the system as described above, the information search program according to the present invention executes a first extraction step of extracting a keyword constituting an information block of the reference data from the reference data stored in the storage means in the computer. Let Further, the program causes the computer to execute a step of extracting keywords constituting a plurality of information blocks of the search target data from the search target data stored in the storage unit. Further, the program displays one or more keywords extracted from the reference data and one or more keywords extracted from the search target data on the display unit, and displays each keyword of the displayed reference data. And the search target data Extracted from keyword Association with To perform the step of letting the investigator select. In addition, the program is associated with the computer. , Search target data Extracted from Based on keywords, The keyword was extracted A search step for searching for the similar information block from search target data is executed.
[0033]
According to the present invention, since the similar information block is searched from the search target data based on the keyword of the search target data associated with each keyword of the reference data, unlike the conventional case, the investigator is more A highly accurate search can be performed. Also, unlike the prior art, the investigator can perform a search without registering a keyword for each information block of the search target data. Further, unlike the conventional case, an investigator can select an appropriate search keyword corresponding to the search target data. In addition to this, unlike the conventional case, the searcher can appropriately select a keyword of the search target data having a relationship such as a synonym for each keyword of the reference data as a search keyword.
[0034]
Further, the information search program according to another invention may be configured such that the search step is based on a keyword extracted from the reference data and a keyword extracted from search target data associated with each keyword. The similar information block is searched from data.
[0035]
According to the present invention, in order to search the similar information block from the search target data based on the keyword of the reference data and the keyword of the search target data associated with each keyword, unlike the conventional, The investigator can perform a search with higher accuracy.
[0036]
Furthermore, an information search program according to another invention is characterized in that the first extraction step includes a step of accepting from the input means that the investigator edits a keyword extracted from the reference data.
[0037]
According to the present invention, an investigator edits the keyword of the reference data,
Unlike the conventional case, the investigator can perform a search with higher accuracy.
[0038]
An information search program according to another invention is characterized in that the first extraction step includes a step of accepting from the input means that an investigator discards a keyword extracted from the reference data.
[0039]
According to the present invention, by investigating the keyword of the reference data,
Unlike the conventional case, the investigator can perform a search with higher accuracy.
[0040]
Furthermore, an information search program according to another invention includes a step of causing the computer to accept an association with a weight representing an importance level in search processing from an investigator for each keyword extracted from the reference data. And the said search step searches the said similar information block from the said search object data based on the keyword extracted from the said search object data linked | related for every keyword of the said reference data, and the said weight. It is characterized by performing.
[0041]
According to the present invention, since the similar information block is searched from the search target data based on the keyword of the search target data and the weighting, which are associated with each keyword of the reference data, unlike the conventional case. , The investigator can perform a more accurate search. Also, unlike the conventional case, the search result is corrected by weighting, so that a more detailed search can be performed.
[0042]
An information search program according to another invention includes a step in which the search step gives information indicating a similarity for each information block when a similar information block is searched from the search target data. Then, the program causes the computer to execute a step of displaying the search results in the order of similarity.
[0043]
According to the present invention, since the search results are displayed in the order of similarity, the investigator can easily understand the search results. Also, unlike the conventional case, the researcher can easily access the individual breakdown of the search results.
[0044]
Furthermore, in an information search program according to another invention, the information block of the reference data is specific patent application information, and the plurality of information blocks of the search target data are a plurality of patent application information or scientific and technical literature information. It is characterized by being.
[0045]
According to the present invention, for a specific patent application information that is an information block of the reference data, similar patent application information from a plurality of patent application information or science and technology documents that are a plurality of information blocks of the search target data Or it becomes possible to search scientific and technical literature. As a result, unlike the conventional case, it is possible to reduce the investigation time spent by the investigator and reduce the burden.
[0046]
An information search program according to another invention is characterized in that the information block of the reference data is one or more claims of a specific patent application.
[0047]
According to the present invention, for one or more claims of a specific patent application that is an information block of the reference data, from a plurality of patent application information or scientific and technical documents that are a plurality of information blocks of the search target data, It is possible to search for similar patent application information or scientific and technical literature. This makes it possible to search for similar patent application information or scientific and technical literature for each claim, unlike the prior art.
[0048]
Furthermore, in the information search program according to another invention, when the information block of the reference data, which is one or more claims of a specific patent application, is a dependent claim, the first extraction step is referred to by the dependent claim. Adding an independent term to the information block of the reference data.
[0049]
According to the present invention, when an information block of the reference data which is one or more claims of a specific patent application is a dependent claim, an independent claim cited by the dependent claim is added to the information block of the reference data It becomes possible. Thus, unlike the conventional case, even if the information block of the reference data is a dependent claim, it is possible to search for similar patent application information or scientific and technical literature for each claim.
[0050]
As a result, the above-described object is achieved.
[0051]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to FIGS.
[0052]
FIG. 1 is a configuration diagram of an information search system according to an embodiment of the present invention. As shown in FIG. 1, the information search system 1 comprises a storage means 2, an input means 3, a display means 4, and a computer 5 for controlling them. These means and the computer 5 are connected to each other, and an information search function 15 is realized by cooperating while exchanging data. The storage unit 2 stores reference data 21 that is a reference for search processing and search target data 22 that is a search target of reference data.
[0053]
Next, the storage means 2 in FIG. 1 and the reference data 21 and search target data 22 stored therein are shown in detail in the detailed view of the storage means 2 in FIG. Here, the reference data 21 includes an information block 21a composed of a plurality of keywords. In addition, the search target data 22 includes a plurality of information blocks 22a including a plurality of keywords. The search target data 22 is search data narrowed down in advance by a researcher by using keywords, classification symbols, etc. in the search formula. To do.
[0054]
Here, the keywords included in the reference data 21 and the keywords included in the search target data 22 are composed of one or more words, symbols, numerical values, or combinations thereof. Examples of this include “key item”, “system”, “search result”, and the like. In FIG. 2, the keyword “a” included in the reference data 21 and the keyword “a ′” included in the search target data 22 are in a relationship of synonyms and the like (including synonyms and the like). Similarly, the keyword b included in the reference data 21 and the keyword b ′ included in the search target data 22 have a relationship such as a synonym. The relationship of synonyms and the like here means that the keywords are in a relationship of synonyms and the like, such as the relationship between the “primary key” and “main item” and “search item” described above.
[0055]
Further, the reference data information block 21a represents an information unit that is a reference for information search. In addition, a plurality of information blocks 22 a included in the search target data 22 represent information units to be searched for the information block 21 a of the reference data. Here, the information search function 15 described above is based on the reference data information block 21a and the computer 5 based on the keyword of the reference data 21 and / or the keyword of the search target data 22 having a relationship such as a synonym thereof. This is a function for determining the information block 22a of similar search target data. The information block 21a for the reference data and the information block 22a for the search target data are preferably divided for each document or document. In this case, the information search function 15 determines the information block 22a of the search target data that is a similar document of the specific document based on the information block 21a of the reference data that is the specific document.
[0056]
FIG. 3 is a block diagram when the information search system of FIG. 1 is applied to a personal computer. 1, the storage means 2 corresponds to the memory 6 and the hard disk 10, the input means 3 corresponds to the keyboard 9 and the like, the display means 4 corresponds to the display 7, and the computer 5 corresponds to the CPU 8. And each component is connected by the bus | bath 13, and it is comprised so that it can communicate mutually. Further, the reference data 21 and the search target data 22 stored in the storage unit 2 in FIG. 1 are stored in the hard disk 10 as the reference data 11 and the search target data 12 in a file or database format. Similarly, the information search function 15 is stored in the hard disk 10 as the information search program 14. In the system in FIG. 3, the CPU 8 interprets the information search program 14 to perform calculation processing and control each component. Further, when the CPU 8 interprets and executes the information search program 14, the information search program 14 and data are read from the hard disk 10 into the memory 6.
[0057]
Here, the keyboard 9 includes a mouse or the like generally used in a personal computer. The display 7 is a CRT display, a liquid crystal display, or the like. Further, the illustrated reference data keyword 11 a conceptually indicates that the keyword included in the reference data 11 is extracted by the CPU 8 and read into the memory 6. The same applies to the keyword 12a of the search target data.
[0058]
Next, the operation of each component in this embodiment will be described with reference to the flowchart of FIG.
[0059]
When a person who intends to conduct a survey using the information search system of FIG. 3 (hereinafter abbreviated as “surveyor”) uses a keyboard 9 as an input means and inputs a command to start a survey, the CPU 8 The command is interpreted and the information search program 14 is activated.
[0060]
Here, the CPU 8 interprets and executes the activated information search program 14, whereby the four processes shown in the flowchart of FIG. 4 are performed. The four processes are a first extraction step S1 for extracting a keyword from reference data, a step S2 for extracting a keyword from search target data, and an association between the keyword of the reference data and the keyword of the search target data. Step S3 and step S4 for searching the information block of the search target data based on the associated keyword of the search target data. The information search function 15 is realized by the above four processes. Further, the first extraction step S1 for extracting a keyword from the reference data and the step S2 for extracting the keyword from the search target data are processed in different steps in advance, and the reference is immediately performed by the execution of the information search program 14. It is also possible to proceed from step S3 in which the association between the data keyword and the search target data keyword is selected.
[0061]
Below, the outline of each step of S1 to S4 will be described first.
[0062]
First, the first extraction step S1 for extracting a keyword from reference data is a step of extracting a plurality of keywords constituting an information block included in the reference data 11 stored in the hard disk 10. The keywords extracted by this step are used in S3 and S4. Hereinafter, the first extraction step S1 for extracting a keyword from reference data is abbreviated as “first extraction step from reference data”.
[0063]
Next, the step S2 for extracting keywords from the search target data is a step of extracting a plurality of keywords constituting a plurality of information blocks included in the search target data 12 stored in the hard disk 10. The keywords extracted by this step are used in S3 and S4. Hereinafter, step S2 for extracting a keyword from search target data is abbreviated as “extraction step from search target data”.
[0064]
Subsequently, in step S3 for selecting the association between the keyword of the reference data and the keyword of the search target data, the association between the keyword extracted from the reference data 11 in S1 and the keyword extracted from the search target data 12 in S2. Is a step for allowing the investigator to select. The association selected by this step is used in S4. Hereinafter, step S3 for selecting the association between the keyword of the reference data and the keyword of the search target data is abbreviated as “association selection step”.
[0065]
Subsequently, step S4 of searching for the information block of the search target data based on the keyword of the associated search target data is the same as the information block of the reference data 11 based on the keyword of the search target data 12 associated in S3. This is a step of searching for similar information blocks of the search target data 12. Through this search step, it is determined whether each information block of the search target data 12 is similar to the information block of the reference data 11. Hereinafter, step S4 for searching for the information block of the search target data based on the associated keyword of the search target data is abbreviated as “search step”.
[0066]
The above is the outline of each step from S1 to S4, and each step will be described in detail below.
[0067]
First, in the first extraction step S1 from the reference data, the CPU 8 described above interprets and executes the information search program 14, thereby performing the processes of S11 to S15 shown in the flowchart of FIG. Hereinafter, based on the flowchart of FIG. 5, each process of S11 to S15 is demonstrated.
[0068]
First, in S11, the CPU 8 displays the reference data input screen 30 of FIG. 7 on the display 7 and waits for a reference data input command from the investigator. Here, the investigator uses the keyboard 9 to instruct to read the reference data through the reference data read button 34. Furthermore, the investigator also specifies information for specifying reference data such as a path and a file name. Then, the CPU 8 receives the investigator's instruction to read the reference data, and displays the designated reference data on the reference data display unit 33. Further, the CPU 8 analyzes the reference data, divides the reference data for each claim based on the characters “claim” and symbols in black brackets, and displays a list of claims on the claim list display unit 31. . An example of the reference data input by the investigator is a specification such as a patent. As an example of such reference data reception processing, when the researcher designates the data shown in FIG. 8 as reference data, the reference data input screen 30 is in a state as shown in FIG.
[0069]
Next, when the CPU 8 displays the input screen of FIG. 9, the investigator's claim selection command is an instruction to select which of the claims of the reference data displayed on the claim list display unit 31 is to be used. Wait for. Here, the investigator commands selection of one or more claims by using a keyboard 9 or the like and double-clicking each claim in the claim list display section 31. Then, the CPU 8 receives an instruction from the investigator to select a claim, extracts the specified claim data from the reference data display unit 33, and displays it on the selected claim display unit 32.
[0070]
At the time of this extraction / display process, the CPU 8 performs branch determination S12 as to whether or not a dependent claim is included in one or more claims selected by a previous investigator's claim selection instruction. Is called. Here, whether or not a dependent claim is included is determined by checking whether the claim data extracted from the reference data display unit 33 is another claim such as the characters “claim” or “claims 1 to 3”. Is determined by whether or not If the CPU 8 determines in S12 that the dependent claim is included, the claim data cited by the dependent claim is also extracted from the reference data display unit 33 and displayed on the selected claim display unit 32 together. (S13). In S13, when a claim cited by a dependent claim further cites another claim (when it is a dependent item of another claim), data of the cited claim is also obtained. In addition, it is displayed on the selected claim display section 32. Such processing is repeated until the claim being cited becomes an independent claim. As an example of the dependent term determination S12 as described above, when the investigator designates “Claim 2” from the claim list display unit 31 on the reference data input screen 30 in the state of FIG. As shown in FIG. 10, the data of claims 1 and 2 are reflected. In this way, the examiner's claim selection instruction is reflected on the selected claim display section 32. The claim data displayed on the selected claim display unit 32 represents an information block of reference data, which is an information unit serving as a reference for information search in the information search function described above.
[0071]
Then, on the reference data input screen 30 in FIG. 10, the investigator can edit the claim data displayed on the selected claim display section 32 by using the keyboard 9 or the like. In other words, characters, symbols, etc. can be added to or deleted from the claim data displayed on the selected claim display section 32.
[0072]
Subsequently, on the reference data input screen 30 in FIG. 10, the researcher uses the keyboard 9 or the like and clicks the keyword extraction button 35 to instruct the keyword extraction from the information block of the reference data. Then, the CPU 8 receives the investigator's keyword extraction instruction, and divides the claim data displayed on the selected claim display unit 32 into keywords (S14). As a method of this division, there is a method in which the CPU 8 divides into a keyword for each particle or blank character included in the claim data. As an example of such keyword division, the claim data displayed on the selected claim display unit 32 in the input screen of FIG. 10 is divided into keywords as shown in the divided keyword list 37 shown in FIG.
[0073]
Then, when the claim data is divided into keywords, the CPU 8 causes the display 7 to display the search condition determination screen 40 of FIG. 12, the claim data is displayed in the claim data display section 42, and the divided keyword list 37 is displayed in the keyword list display section 43. (S15). In the search condition determination screen 40 of FIG. 12, the investigator can edit the claim data displayed on the claim data display unit 42 using the keyboard 9 or the like. In other words, characters, symbols, etc. can be added to or deleted from the claim data displayed on the claim data display unit 42. Here, when the claim data displayed on the claim data display unit 42 is edited, the investigator uses the keyboard 9 or the like and clicks the keyword re-extraction button 41 to remove the claim data from the claim data display unit 42. Keyword re-extraction, and the extracted keyword list can be re-displayed on the keyword list display unit 43.
[0074]
In this way, when the CPU 8 displays a list of keywords on the keyword list display unit 43, the CPU 8 waits for an investigator's keyword selection command, which is an instruction for the keyword to be used among the keywords. Here, the investigator commands the selection of one or more keywords by, for example, double-clicking each keyword in the keyword list display unit 43 using the keyboard 9 or the like. Then, the CPU 8 receives the investigator's keyword selection instruction, extracts the specified keyword from the keyword list display unit 43, and displays it in the component column of the search condition display unit 48. As described above, when the researcher sequentially clicks, for example, each keyword of “key item”, “order”, and “search result” from the keyword list display unit 43, the configuration of the search condition display unit 48 as shown in FIG. The selected keyword is displayed in the element column. In this case, it is appropriate that the keyword selected by the investigator is selected not to display all the keywords displayed in the keyword list display unit 43 but to represent the features of the invention. In this way, keyword selection and editing are exclusively determined by the investigator, and are not automatically determined by the information search program 14.
[0075]
The above is the description of each process in the first extraction step S1 from the reference data. Here, each keyword displayed in the component column of the search condition display unit 48 in FIG. 13 is the keyword extracted in the first extraction step S1 from the reference data. Hereinafter, the extraction step S2 from the search target data will be described.
[0076]
Also in the extraction step S2 from the search target data, the CPU 8 described above interprets and executes the information search program 14 to perform the following processing. First, the CPU 8 receives input of search target data to be searched for the reference data received in S1 from the investigator through the keyboard 9 or the like. Examples of the search target data include a plurality of patent gazettes. In this case, the patent gazette for each application is an individual information block in the search target data. Further, as described above, each information block in the search target data is an information unit whose similarity with the information block of the reference data is determined.
[0077]
And CPU8 divides | segments the search object data input by the investigator into a keyword similarly to S14 in 1st extraction step S1 from the above-mentioned reference | standard data. As a method of this division, there is a method in which the CPU 8 divides into keywords for each particle or blank character included in the search target data. FIG. 14 shows the keyword list 38 of the search target data divided in the extraction step S2 from the search target data as described above. Here, it is assumed that the keyword displayed in FIG. 14 is the keyword extracted in the extraction step S2 from the search target data.
[0078]
The above is description of each process in extraction step S2 from search object data. In the extraction step S2 from the search target data, the investigator only specifies the search target data, and other processing is performed based on the information search program 14. Therefore, there is no inconvenience even if the information search program 14 performs the first extraction step S1 from the reference data after the extraction step S2 from the search target data. That is, the first extraction step S1 from the reference data and the extraction step S2 from the search target data are in no particular order. Hereinafter, the association selection step S3 will be described.
[0079]
Also in this association selection step S3, the CPU 8 described above interprets and executes the information search program 14 to perform the following processing. Here, when S1 and S2 are finished, the search condition determination screen 40 of FIG. 13 is displayed on the display 7, and the CPU 8 waits for an association selection command from the investigator. Here, the investigator uses the keyboard 9 or the like to select an arbitrary keyword (hereinafter abbreviated as “optional keyword”) from among the keywords displayed in the component row of the search condition display unit 48. The association selection screen display is instructed by clicking the synonym button 44 or the like. When detecting this, the CPU 8 extracts the arbitrary keyword data from the search condition display unit 48 and displays the association selection screen 50 held in advance by the information search program 14 on the display 7. (Described later)
[0080]
The association selection screen 50 includes a target keyword display unit 51, a hit keyword list display unit 52, a non-hit keyword list display unit 53, and an association execution button 54. Here, the target keyword display unit 51 is a part that displays an arbitrary keyword previously selected from the component row of the search condition display unit 48 by the investigator. In the association selection screen 50 of FIG. 15, it can be seen that “key item” is selected as an arbitrary keyword. The hit keyword list display unit 52 is a part that displays a list of keywords that have been hit as a result of searching the keywords extracted in S2 using an arbitrary keyword as a search key. Further, the non-hit keyword list display unit 53 is a part that displays a list of keywords that have not been hit as a result of searching the keywords extracted in S2 using an arbitrary keyword as a search key. The general pattern matching described above is used as a search method for the keyword extracted in S2. As a result, the keywords extracted in S2 are sorted and displayed on either the hit keyword list display unit 52 or the non-hit keyword list display unit 53, respectively. This distribution process is determined based on the similarity criterion selected by the investigator. As shown in FIG. 15, the association selection screen 50 includes a plurality of radio buttons so that the investigator can select the similarity criterion. The association selection screen 50 in FIG. 15 has three radio buttons with a similarity of (weak), (medium), and (strong), and the similarity criterion can be selected depending on which radio button the investigator selects. . As an example of keyword assignment based on the similarity criterion, when the similarity criterion is set to (strong) for an arbitrary key called “primary key”, the character “primary key” in the keywords extracted in S2 is completely Only the keywords included in are displayed on the hit keyword list display section 52. Similarly, when the similarity criterion is set to (weak) for an arbitrary key “primary key”, a part of the character “primary key” (for example, “primary” or “primary key”) is extracted from the keywords extracted in S2. The keyword including the key “)” is displayed on the hit keyword list display unit 52.
[0081]
When the previous association selection screen 50 is displayed on the display 7, the CPU 8 displays the previously extracted arbitrary keyword data on the target keyword display unit 51. Further, the CPU 8 reads the similarity criterion initially set by the information search program 14 and the keyword extracted in S2, and based on the similarity criterion, the keyword extracted in S2 is obtained by the previous pattern matching. The hit keyword list display unit 52 and the non-hit keyword list display unit 53 are sorted and displayed. The reason for setting such a similarity criterion is that when the search target data includes many keywords similar to the keyword selected as the constituent element, only the keywords with high similarity are displayed on the hit keyword list display unit 52. This is because the searcher can select a keyword for search that is necessary immediately. Therefore, when there is almost no keyword similar to the keyword selected as the component in the search target data, the keyword including the low similarity is displayed on the hit keyword list display unit 52 so that the investigator can select it. . Furthermore, one of the radio buttons can be selected by the information search program 14 so that the number of keywords displayed in the hit keyword list display section 52 falls within a certain range.
[0082]
Further, when the investigator selects the similarity criterion selection radio button on the association selection screen 50 using the keyboard 9 or the like and changes the similarity criterion, the CPU 8 again performs the first step based on the similarity criterion. The keywords extracted in S2 are read and sorted and displayed on the hit keyword list display unit 52 and the non-hit keyword list display unit 53 by the previous pattern matching. In this way, the investigator can change the similarity criterion initially set by the information search program 14.
[0083]
Subsequently, the investigator uses the keyboard 9 or the like, and among the keyword lists extracted in S <b> 2 displayed on the hit keyword list display unit 52 and the non-hit keyword list display unit 53, the target keyword display unit 51. A keyword that is considered to have a relationship such as a synonym with the arbitrary keyword displayed in (1) is selected. As this selection method, there is a method of selecting by checking a check box provided for each keyword displayed on the hit keyword list display unit 52 and the non-hit keyword list display unit 53. Usually, the keyword selected by the investigator as a constituent element is not displayed in the hit keyword list display section 52. The keyword selected as the constituent element by the researcher does not need to be displayed, but is processed as being selected by the researcher. If the keyword selected as the constituent element by the investigator is also displayed, the investigator can confirm the presence or absence of the selected constituent element in the search target data. When this check is detected, the CPU 8 stores the association between the arbitrary keyword and the checked keyword. When the check is released, the CPU 8 deletes the association between the arbitrary keyword and the keyword for which the check is released. Taking the association selection screen 50 in FIG. 15 as an example, “search key”, “search item”, and “feature part” are selected as keywords for “key item” which is an arbitrary keyword, and the association is stored. It is in a state.
[0084]
In this way, by using the keyboard 9 or the like, the researcher selects a keyword that is considered to be in a relationship such as a synonym with an arbitrary keyword, and clicks the association execution button 54 on the association selection screen 50, etc. The other keyword is stored in association with the arbitrary keyword. In response to this, the CPU 8 saves the association, and the other keyword associated with the arbitrary keyword is the row of the arbitrary keyword in the search condition display unit 48 on the search condition determination screen 40, and the synonym item Each is displayed in an area that intersects the column. As an example of this, in the search condition display section 48 on the search condition determination screen 40 in FIG. 16, three keywords “search key”, “search item”, and “feature part” are assigned to an arbitrary keyword “key item”. However, it is shown that the two keywords “procedure” and “order” are associated with the arbitrary keyword “order”. At the same time, the search condition display section 48 on the search condition determination screen 40 in FIG. 16 also shows that no keyword is associated with the arbitrary keyword “search result”. As described above, when there is no association keyword for an arbitrary keyword, the search database is searched using only the word / phrase selected as the constituent element, that is, the keyword “search result” in this case. (Described later)
[0085]
The above is the description of each process in the association selection step S3. Here, for each arbitrary keyword displayed in the component column of the search condition display unit 48 in FIG. 16, the keyword displayed in the synonym item in the row for each arbitrary keyword is the arbitrary keyword in S3. This is a keyword of search target data associated with each. Hereinafter, the search step S4 will be described.
[0086]
Also in this search step S4, the CPU 8 described above interprets and executes the information search program 14, whereby the following processing is performed. Here, the search step S4 comprises the processes of S41 to S43 shown in the flowchart of FIG. Hereinafter, based on the flowchart of FIG. 6, each process of S41 to S43 is demonstrated.
[0087]
First, in the search condition determination screen 40 of FIG. 16, the CPU 8 issues an investigator's weight assignment command, which is an instruction to assign a weight to each arbitrary keyword displayed in the component row of the search condition display section 48. wait. Here, the investigator uses the keyboard 9 or the like to select the keyword by clicking each keyword displayed in the component row of the search condition display unit 48, and then operates the weighting meter 45. By giving a numerical value from the keyboard 9 or the like, a weighting instruction is given. Here, as shown in FIG. 16, the weighting meter 45 is configured to select a value indicated by the weighting indicator bar from numerical values in a specific range by operating the central weighting indicator bar left and right. . When receiving the weighting instruction from the investigator, the CPU 8 displays the value shown in the weighting meter 45 in the weight column in the corresponding keyword row of the search condition display unit 48 (S41). As an example of this, in the search condition determination screen 40 of FIG. 16, “8” is set for the optional keyword “key item”, “6” is set for the optional keyword “order”, and optional keyword “search result”. In contrast, when “3” is weighted by the investigator, the search condition determination screen 40 is in a state as shown in FIG. Further, for the keywords displayed in the component column of the search condition display unit 48 that are not particularly specified by the investigator, if the initial setting values are automatically set by the information search program 14, it is convenient for the researcher. It is desirable.
[0088]
Here, the weighting step S41 by the investigator is included in S4 for convenience of explanation, but may be present at the end of S1 or other steps of S2 and S3. In addition, the investigator can save various data displayed on the search condition determination screen 40 or read the previously stored various data when appropriate when displayed on the search condition determination screen 40. For example, when the investigator clicks the setting save button 46 on the search condition determination screen 40 using the keyboard 9 or the like, the CPU 8 transfers the various data displayed on the search condition determination screen 40 to the hard disk 10. Save to. When the investigator clicks the setting reading button 47 on the search condition determination screen 40 using the keyboard 9 or the like and specifies the setting file, the CPU 8 transfers the setting file from the hard disk 10 to the memory 6. Read and display the contents of the setting file on the search condition determination screen 40.
[0089]
Next, the CPU 8 waits for the searcher's search start command, which is a search start instruction for performing similarity determination for each information block of the search target data based on the conditions of the search condition display unit 48. Here, the investigator uses the keyboard 9 or the like to instruct the start of search by clicking the search start button 49 on the search condition determination screen 40 or the like. Then, the CPU 8 receives a search start instruction from the investigator, and extracts the search condition data displayed on the search condition display unit 48. As an example of this, when a search start instruction is given from the investigator in the state of the search condition determination screen 40 in FIG. 17, search condition data 60 as shown in FIG. 18 is extracted. Here, a new management number column exists in the search condition data 60, which is given a management number to make each record of the search condition data 60 unique.
[0090]
Subsequently, the CPU 8 performs similarity determination for each information block of the search target data based on the search condition data 60 (S42). As a similarity determination method, the number of synonyms for a specific component in the search condition data 60 is N, and a specific information block (hereinafter referred to as a “specific information block”) having search target data is used. As a result of a search using N keywords, the score of the specific information block for that component is recorded as “(M / N) * weight” for each specific information block. . Similarly, the same search and recording are performed for the “order” and “search result” which are other components. Finally, the score for each component is totaled for each specific information block, and the similarity score of the specific information block is obtained.
[0091]
For example, in the search condition data 60 shown in FIG. 18, the number of synonyms is three (“search key”, “search item”, “feature part”) in the row where the component is “key item”. The number of N is N = 4 by adding the “key item” selected as the component. On the other hand, in a specific information block having search target data, when the information block is searched using “search key”, “search item”, “feature part”, and “key item” as a search keyword, “search key” ”,“ Search item ”,“ characteristic part ”, and“ key item ”are all present, it can be seen that M = 4. From the search condition data 60 shown in FIG. It can be seen that the weight of the “key item” is 8. Therefore, the score of the row in which the constituent element in this specific information block is “key item” is “(4/4) * 8” when substituted in the formula of “(M / N) * weight”, and the solution of the formula is It turns out that it is "8". Similarly, the score is calculated in the specific information block in the row of the “order” component and the “search result” row. And the total score (similarity score) of the specific information block is calculated by summing up the scores calculated for each row of the above three components. Here, the scores calculated for each row of the three components are simply summed up, but the product of the scores calculated for each row of the three components may be obtained to correct the total score. it can. However, since the degree of similarity is calculated for texts such as patent specifications and technical documents that can express the same technology in a wide range of synonyms, there is almost no meaning to strictly determine the score calculation method, so various calculation methods can be adopted .
[0092]
In the previous S42, the number (N−1) in which the specific component is not added to N, which is the number of synonyms for the specific component in the search condition data 60, may be N. For example, in the search condition data 60 shown in FIG. 18, the number of synonyms is three (“search key”, “search item”, “feature part”) in the row where the component is “key item”. N = 3, and using the three keywords selected by the investigator as search keywords, the previous S42 is performed. Thus, by not using the component itself as a keyword, it is possible to perform similarity determination using only the keyword in the search reference data.
[0093]
In this way, when the CPU 8 calculates the similarity score for each information block of the search target data based on the search condition data 60, the CPU 8 sorts the information blocks of the search target data in descending order of the similarity score, and displays 7 is displayed as a list (S43). As this example, there is a display method as shown in FIG. In the information block similarity order list 70 of the search information data in FIG. 19, the value of the item “document number” means the initial display order of the information block, and the value of the item “scoring rank” is the information. It means the total score order of blocks. For example, from the row of the information block whose “document number” is “373”, it is the information block initially displayed in the 373rd row, and it is understood that the total score is the highest compared to the other information blocks. . Similarly, from the row of the information block whose “document number” is “373”, the “public number” is “H03-1234” and the “publication date” is “H03 / 10/10”. Further, it can be read that the “title of the invention” is “A system”. Furthermore, the item “link” holds identification information of individual information block data such as a path and a file name. Thus, the investigator uses the keyboard 9 or the like and selects individual information blocks on the information block similarity order list 70, so that the CPU 8 can immediately display the information blocks on the display 7.
[0094]
Through S43 as described above, the researcher can recognize the information blocks of the search target data having a high similarity score with respect to the information blocks of the reference data in the order of the similarity score.
[0095]
As described above, according to the present embodiment, for each keyword extracted from the reference data, a plurality of keywords extracted from the search reference data can be related, and based on the keywords from the related search reference data. Thus, since the information block of the search reference data can be searched, there is an effect that the target information block can be searched and extracted without fail. Therefore, the investigator can view the contents included in the search reference data in the order that requires attention for each information block unit, and can greatly reduce the time required for the reading investigation.
[0096]
【The invention's effect】
Since the present invention is configured and functions as described above, according to this, the keyword extracted from the search target data can be associated with each keyword extracted from the reference data. Based on the keyword, the similarity can be determined for each information block of the search target data.
[0097]
Further, it is not necessary to register a search keyword in advance for each information block of search target data, and all search target data can be set as a keyword target.
[0098]
Furthermore, even when the necessity of unifying registered keywords is not considered in the search target data, the researcher can appropriately select technical terms having synonyms and the like as search keywords.
[0099]
In addition, for each keyword extracted from the reference data, similarity determination can be adjusted by assigning a weight to the keyword extracted from the associated search target data.
[0100]
Due to the above points, it is possible to conduct a more detailed investigation than before, and it has the effect of being able to search and extract the target information block from the search target data without fail, which is superior to the conventional information search A system can be provided.
[Brief description of the drawings]
FIG. 1 is a configuration diagram of an information search system according to an embodiment of the present invention.
FIG. 2 is a detailed view of storage means 2 in FIG.
FIG. 3 is a block diagram when the information search system of FIG. 1 is applied to a personal computer.
FIG. 4 is a flowchart illustrating processing of the information search system.
FIG. 5 is a flowchart showing details of S1 in FIG. 4;
FIG. 6 is a flowchart showing details of S4 in FIG.
7 is a screen view of a reference data input screen 30. FIG.
FIG. 8 is a data example of reference data.
9 is a screen view of a reference data input screen 30. FIG.
10 is a screen diagram of a reference data input screen 30. FIG.
FIG. 11 is a keyword list obtained by dividing reference data into keywords.
12 is a screen diagram of a search condition determination screen 40. FIG.
13 is a screen view of a search condition determination screen 40. FIG.
FIG. 14 is a keyword list obtained by dividing search target data into keywords.
15 is a screen view of an association selection screen 50. FIG.
16 is a screen diagram of a search condition determination screen 40. FIG.
17 is a screen diagram of a search condition determination screen 40. FIG.
18 is a data example of search condition data 60. FIG.
FIG. 19 is a display example of an information block similarity order list 70 of search information data.
[Explanation of symbols]
1 Information search system
2 storage means
3 Input means
4 display means
5 Computer
6 memory
7 Display
8 CPU
9 Keyboard etc.
10 Hard disk
11 Standard data
11a Keywords for reference data
12 Search target data
12a Keywords for search target data
13 Bus
14 Information search program
15 Information search function
21 Standard data
21a Reference data information block
22 Search target data
22a Information block of search target data
30 Standard data input screen
31 Claim List Display Unit
32 Selection claim display section
33 Reference data display
34 Standard data reading button
35 Keyword extraction button
36 Standard data
37 Split keyword list
38 Keyword list of search target data
40 Search condition decision screen
41 Keyword re-extraction button
42 Claim data display section
43 Keyword list display
44 Synonym button
45 Weighting meter
46 Save settings button
47 Setting read button
48 Search condition display area
49 Search start button
50 Association selection screen
51 Target keyword list display area
52 Hit Keyword List Display
53 Non-hit keyword list display
54 Association button
60 Search condition data
70 List of information block similarity by search information data

Claims (10)

Storage means for storing reference data including an information block including one or more keywords and search target data including a plurality of information blocks including one or more keywords;
An input means for inputting an association between a keyword constituting the reference data and a keyword constituting the search target data;
A computer that executes a process of searching for similar information blocks from the search target data based on the information blocks of the reference data;
In a system comprising display means for displaying the search results,
In the computer,
A first extraction step of extracting a keyword constituting an information block of the reference data from the reference data stored in the storage means;
Extracting keywords constituting a plurality of information blocks of the search target data from the search target data stored in the storage means;
One or more keywords extracted from the reference data and one or more keywords extracted from the search target data are displayed on the display means, and the search target data is displayed for each keyword of the displayed reference data. Allowing an investigator to select an association with a keyword extracted from ,
An information search program for executing a search step for searching for the similar information block from the search target data from which the keyword is extracted based on the associated keyword extracted from the search target data . The information search program according to claim 1,
The search step includes
An information search, wherein the similar information block is searched from the search target data based on a keyword extracted from the reference data and a keyword extracted from the search target data associated with each keyword. program. In the information search program according to claim 2,
The first extraction step includes
The information search program characterized by including the step which receives from the said input means that the investigator edits the keyword extracted from the said reference | standard data. In the information search program according to claim 2,
The first extraction step includes
The information search program characterized by including the step which receives from the said input means that an investigator discards the keyword extracted from the said reference | standard data. The information search program according to claim 1,
The computer is
Receiving an association with a weight representing the importance in the search process from the investigator for each keyword extracted from the reference data,
The search step executes a step of searching for the similar information block from the search target data based on the keyword extracted from the search target data associated with each keyword of the reference data and the weighting. Information search program characterized by letting In the information search program according to claims 1 to 5,
The search step includes
Providing information indicating similarity for each information block when searching for similar information blocks from the search target data,
In the computer,
An information search program for executing a step of displaying the search results in the order of similarity. In the information search program according to claims 1 to 6,
The information block of the reference data is specific patent application information,
An information search program characterized in that a plurality of information blocks of the search object data are a plurality of patent application information or scientific and technical literature information. The information search program according to claim 7,
An information search program characterized in that the information block of the reference data is one or more claims of a specific patent application. The information search program according to claim 8, wherein
If the information block of the reference data that is one or more claims of a particular patent application is a dependent claim,
The first extraction step includes
An information search program comprising a step of adding an independent term cited by the dependent term to the information block of the reference data. A recording medium recording the information search program according to claim 1.

Images