JP6707410B2 - Document search device, document search method, and computer program - Google Patents

Description

The present invention relates to a data processing technique, and particularly to a document search technique.

The applicant has proposed a technique for supporting the search of patent documents (see, for example, Patent Documents 1 and 2).

Japanese Patent Laid-Open No. 2005-50175 Publication of Japanese Patent Laid-Open No. 2009-110541

In the patent search for extracting the target patent document, noise patent documents that do not correspond to the target patent document may be mixed in the extracted patent document group. In addition, a part of the intended patent document may be leaked without being extracted. Up to now, in order to grasp the existence of noise or leakage from the result of patent search, it is necessary for the user to read the extracted patent document group, and the knowledge of the user is necessary. That is, it was not easy to understand the existence of noise and leakage from the patent search result.

The present invention has been made in view of these problems, and a main object thereof is to provide a technique for assisting in efficiently grasping noises and omissions in document retrieval.

In order to solve the above-mentioned problems, a document search device according to an aspect of the present invention is a condition input by a user, and is specified from a plurality of documents that may have a citation relationship with another document. An acquisition unit that acquires a condition for extracting a document, a search unit that extracts a plurality of documents that satisfy the condition acquired by the acquisition unit from a plurality of documents stored in a predetermined storage unit, and a search unit. Based on the citation relationship between each of the extracted plurality of extracted documents and other documents, a leaked document which is not included in the plurality of extracted documents but is expected to be highly relevant to the plurality of extracted documents, A user is provided with information about a leaked document or a noise document detected by the detection unit, which is included in a plurality of extracted documents but is likely to be highly related to non-extracted documents. And an output unit that presents to.

Another aspect of the present invention is a literature search method. This method includes a step of acquiring a condition for inputting a specific document from a plurality of documents, which are conditions input by a user and may have a citation relationship with another document, and a predetermined condition. From the plurality of documents stored in the storage unit, a step of extracting a plurality of documents satisfying the condition acquired in the acquisition step, and a plurality of extracted documents extracted in the extracting step and other documents Relevance between a missing document that is not included in multiple extracted documents but is expected to be highly relevant to multiple extracted documents based on the citation relationship, and a non-extracted document that is included in multiple extracted documents The computer executes a step of detecting at least one of noise documents that are expected to be high, and a step of presenting information about a leaked document or a noise document detected in the detecting step to a user.

It should be noted that any combination of the above constituent elements, and the expression of the present invention converted between a system, a computer program, a recording medium storing the computer program, and the like are also effective as an aspect of the present invention.

According to the present invention, it is possible to assist in efficiently grasping noise and omission in document search.

It is a figure which shows the structure of the information system of an Example. It is a block diagram which shows the function structure of the literature search device of FIG. It is a figure which shows the example of the citation relation in an extraction patent. It is a figure which shows the example of the search screen which a document search device provides. It is a figure which shows the example of the search screen which a document search device provides. It is a figure which shows the example of the citation map which a document search device provides. It is a flow chart which shows operation of a document retrieval device. It is a figure which shows the example of the citation map which the literature search apparatus of a modification provides.

Before describing the embodiments of the present invention, an outline will be first described.
The document search device of the embodiment is a feature that predicts noise and omission in patent search based on the citation information and cited information of the patent group extracted by the search, and the citation with the patent group extracted by the search as the population. It has a feature to display a map.

Features 1. Noise and Leakage Prediction in Patent Search:
In various kinds of research on patents, a database storing patent documents may be searched under a search condition provided by a user to extract a desired patent document. If the search condition (search formula) given by the user is not appropriate, the patent documents that are originally intended should not be sufficiently extracted, that is, search omission may occur. In addition, the extracted patent documents may include noise patent documents that do not correspond to the target patent document.

If the accuracy of the noise is rough, a list of search results can be displayed and a certain degree of judgment can be made based on the technology classification code, the title of the invention, and the like. However, it is necessary for the user to read the extracted document in order to grasp the noise ratio with high accuracy. The leak rate can be determined to some extent based on whether or not the known patent document is extracted if the user has knowledge about the known patent document, but it is difficult to make a judgment without the knowledge. is there. As described above, in order to determine the ratio of noise and leakage in the patent search result, it is necessary for the user to read the extracted document and the knowledge of the user, and it is easy to efficiently determine the ratio of noise and leakage. Was not.

Therefore, the document search device according to the embodiment detects candidates for leaks and noises in the document search based on the citation relationship between the documents extracted by the search and other documents, and information on the detected leaks and noises (such as ratios). ) Is presented to the user. By presenting information about omissions and noises in the document search to the user, the accuracy of document extraction can be confirmed before the user actually reads the extracted document, and the validity of the search condition can be easily determined.

Features 2. Display citation map with patent search results as population:
Usually, the patent search result is often a list of a plurality of patent documents displayed in a table format. The document search device according to the embodiment presents a patent search result to a user using a citation map indicating a citation relationship between a plurality of patent documents extracted by the search and other documents. The citation map is an image in which a plurality of objects showing a plurality of patent documents extracted by the search are arranged in a manner associated with each other based on a citation relationship. This assists the user to intuitively understand the relationship between the plurality of patent documents extracted by the search.

The configuration of the document search device of the embodiment is not limited to the search of patent documents, but can be applied to the search of documents related to various industrial property rights including utility models, designs, and trademarks. Further, the configuration of the document retrieval device of the embodiment is not limited to the retrieval of documents related to industrial property rights, and documents that can form a citation relationship with other documents (for example, academic papers, degree-claimed papers, standardization techniques are established). The present invention can be widely applied to searching for contributed books and books (reference documents described in the book correspond to cited documents), various technical documents, and the like, which are technical documents proposed for doing so. This citation relationship may be designated by the applicant or author of the document, or may be designated in the process of document examination/trial (notification of reasons for refusal, etc.). In the following embodiments, a search for patent documents (publication patent publications and patent publications) will be described as an example.

Here, the “extracted patent”, “necessary patent”, “leakage patent”, and “noise patent” in the embodiment will be described. The extracted patent is a patent document extracted by a search process based on a search condition specified by a user from a plurality of patent documents stored in a database or the like. The necessary patent is a patent document that cites at least one of the plurality of extraction patents, and is a patent document that is cited by at least one of the plurality of extraction patents. The necessary patent is a patent document that is expected to meet the user's search purpose, that is, it can be said that the patent document should be originally extracted. Required patents may include both patent documents included in the extracted patent and patent documents not included in the extracted patent.

Leakage patents are required patents that are not included in the extraction patents. That is, the leaked patent is a patent document which is not included in the extracted patent but is expected to have high relevance to the extracted patent. In other words, a leaked patent is a patent document that is expected to match the user's search purpose, but is a patent document that was not extracted by the user's search conditions.

The noise patent is one of a plurality of extracted patents, but it is a patent document that cites only a non-extracted patent that is not included in the extracted patent, or a patent document that is cited only by a non-extracted patent. .. In other words, the noise patent is a patent document that has a citation relationship but is not cited by any of the plurality of extraction patents, and is not a reference of any of the plurality of extraction patents. In other words, the noise patent is a patent document expected to have a high relevance to the non-extracted patent (highly related to the non-extracted patent and low to other extracted patents). In other words, the noise patent is a patent document that is extracted under the user's search condition, but is a patent document that does not match the user's search purpose (that is, should not be originally extracted). In the above, the noise patent is a patent document that cites only non-extracting patents or a patent document that is cited only by non-extracting patents, but a noise patent is a patent document that cites only non-extracting patents. Moreover, it may be a patent document cited only by a non-extracted patent. It should be noted that patent documents of extracted patents that have no citation relationship (either extracted patents or non-extracted patents) are not usually treated as noise patents, but may be treated as noise patents. Further, a patent document of a non-extracting patent that has no citation relationship (both extracted patent and non-extracting patent) is not treated as a leaked patent, but may be treated as a leaked patent.

FIG. 1 shows the configuration of an information system 100 according to the embodiment. The information system 100 includes a document search device 10, a document DB 12, and a user terminal 16. The document search device 10 is an information processing device that provides a user with a search service for patent documents. The document search device 10 may be realized by cooperation between a web server and an application server, and may provide a web service for searching for patent documents to the user terminal 16.

The document DB 12 is a database server that stores data of a plurality of patent documents. The data of the patent document includes the main data of the application, the description, the drawing, and the abstract. Further, it includes various attribute information relating to patent documents such as filing date, examination status, and cited documents. The document search device 10 and the document DB 12 are connected via a communication network 14 such as a LAN or an intranet. Note that the document DB 12 holds information indicating a citation type regarding a cited document cited for a certain patent document. The type of citation includes the type of subject and the type of citation that set the citation relationship.For example, the citation by the applicant, the citation at the time of examination, the citation at the time of trial, the citation of notice of reasons for refusal, or the decision of refusal. Includes information indicating citations, etc.

The user terminal 16 is an information terminal operated by a user who should search for patent documents. The user terminal 16 and the document search device 10 are connected via a communication network 18 such as LAN, WAN, or the Internet. The user terminal 16 may be a PC, a smartphone, or a tablet terminal. A web browser may be installed in the user terminal 16, and the user may activate the web browser to access the website provided by the document search device 10 and use the search service for patent documents.

FIG. 2 is a block diagram showing a functional configuration of the document search device 10 of FIG. The document search device 10 includes a control unit 20, a storage unit 22, and a communication unit 24. The control unit 20 executes various data processing for realizing the search of patent documents and the display of the citation map. The storage unit 22 is a storage area that stores data that is referred to or updated by the control unit 20. The communication unit 24 communicates with an external device according to a predetermined communication protocol. The control unit 20 transmits/receives data to/from the document DB 12 and the user terminal 16 via the communication unit 24.

Each block shown in the block diagram of this specification can be realized by an element such as a CPU/memory of a computer or a mechanical device in terms of hardware, and can be realized by a computer program or the like in terms of software. Then, we are drawing the functional blocks that are realized by these collaborations. Therefore, it will be understood by those skilled in the art that these functional blocks can be realized in various ways by a combination of hardware and software.

For example, the function of each block of the control unit 20 may be implemented as a computer program, and the computer program may be installed in the storage of the document search device 10. Then, the CPU of the document retrieval apparatus 10 may read the computer program into the main memory and execute the computer program so that the function of each block of the control unit 20 is exerted. Further, the storage unit 22 may be realized by the main memory or storage (HDD, SSD, etc.) of the document search device 10.

The control unit 20 includes a request acquisition unit 26, a search unit 28, a detection unit 30, a search result generation unit 36, a search result provision unit 38, a citation map generation unit 40, and a citation map provision unit 42.

The request acquisition unit 26 acquires the search request transmitted from the user terminal 16 and including the search condition input by the user at the user terminal 16. Further, the request acquisition unit 26 acquires, from the user terminal 16, a citation map provision request that requests provision of citation maps regarding a plurality of patent documents as search results.

The search condition is a condition for extracting a specific patent document that matches a user's search purpose from a plurality of patent documents stored in the document DB 12. The search condition may be a search expression in which a plurality of search items (for example, a character string that should be included in the claims or a character string that should be included in the entire publication) are combined by an AND condition or an OR condition. Further, the search condition includes a necessary patent detection condition and a noise patent detection condition.

The necessary patent detection condition can be said to be a leaked patent detection condition, and includes the number of citations and the number of citations designated by the user. The necessary patent detection conditions range from the first patent document to the second patent document when a direct or indirect citation relationship is established between the first patent document and the second patent document. Includes the "number of generations", which is the number of stages of multi-stage quotation up to. The number of generations includes the number of cited generations indicating the number of generations in the direction in which a certain document cites another document, and the number of cited generations indicating the number of generations in the direction in which a certain document is cited by another document. ..

For example, assume a citation relationship of patent D→patent B→patent A→patent C→patent E (the side to which the tip of the arrow cites). When the number of cited generations is 1 and the number of cited generations is 1, the range up to the patent B directly cited by the patent A is extracted as the range of the cited direction, and the range up to the patent C directly cited of the patent A is extracted as the range of the cited direction. The presence or absence of a patent is determined to determine whether or not the patent A is a necessary patent. Further, when the number of cited generations is 2 and the number of cited generations is 2, the patent B directly cited by the patent A and the patent D directly cited by the patent B are included in the citation direction range, and the patent A is directly cited. The presence or absence of the extracted patent is determined with the range of the patent C, which includes the patent C, and the patent E, which directly cites the patent C, in the cited direction, and it is determined whether the patent A is a necessary patent.

The noise patent detection condition includes the number of citations and the number of citations designated by the user. Moreover, the detection conditions of the noise patent range from the first patent document to the second patent document when a direct or indirect citation relationship is established between the first patent document and the second patent document. The number of generations up to (the number of cited generations and the number of cited generations) is included. The number of citations, the number of citations, the number of citations, and the number of citations as detection conditions for required patents are different from the number of citations, the number of citations, the number of citations, and the number of citations as detection conditions for noise patents. A value may be specified.

When the request acquisition unit 26 acquires the search request, the search unit 28 extracts a plurality of patent documents satisfying the search condition included in the search request from the plurality of patent documents stored in the document DB 12. For example, the search unit 28 may transmit a query message in which keywords of one or more search items indicated by the search condition are set, to the document DB 12, and as a response to the query message, the document DB 12 satisfies the search condition. The document data may be transmitted to the document search device 10.

The search unit 28 acquires at least data indicating a citation relationship between the patent document and another document from the document DB 12 as data of the patent document satisfying the search condition. The data indicating the citation relationship includes identification information and citation types of other documents cited by the patent document, and further includes identification information and citation types of other documents in which the patent document is cited. The identification information of another document may be, for example, the application number of another patent document, a document name, or a URL. The data of the patent documents acquired by the search unit 28 in the embodiment includes data indicating the applicant (the author if not a patent document) and data on the date (for example, filing date, priority date, announcement date, publication date, etc.). Including.

The detection unit 30 determines whether or not there is a missing patent based on the citation relationship between each of the plurality of extracted patents extracted by the search unit 28 and other patent documents (that is, the citation relationship indicated by the patent document data provided from the document DB 12). Detect at least one of the noise patents. A citation relationship includes both a relationship in which a patent document cites another patent document and a relationship in which a patent document is cited by another patent document. The citation relationship may be set by the applicant of the patent document, or may be set in the process of examination or trial of the patent document. The detection unit 30 of the embodiment detects both the leak patent and the noise patent. The detection unit 30 includes a required document detection unit 31, a leak detection unit 32, and a noise detection unit 34.

The necessary document detecting unit 31 and the leak detecting unit 32 detect the presence of the leak of the patent document to be originally searched for in the search result under the search condition specified by the user. Specifically, the necessary document detection unit 31 is a patent document that cites at least one of a plurality of extracted patents, and a patent document that is cited by at least one of the plurality of extracted patents is a necessary patent. To detect. The leakage detection unit 32 detects necessary patents, which are not included in the plurality of extracted patents, among the necessary patents detected by the necessary document detection unit 31 as leakage patents.

The necessary document detection unit 31 detects a necessary patent according to the necessary patent detection condition included in the search condition. Specifically, the necessary document detection unit 31 detects a patent document satisfying both the following condition 1 and condition 2 as a necessary patent. Condition 1 is a patent that directly or indirectly cites more than the number of citations specified by the required patent detection condition among a plurality of extracted patents within the number of citation generations specified by the required patent detection condition. It corresponds to the literature. Condition 2 is directly or indirectly quoted by a number of extracted patents out of a plurality of extracted patents specified by the required patent detection condition within the range of the number of cited generations specified by the required patent detection condition. It corresponds to the patent documents described.

The noise detection unit 34 detects the presence of noise that should not be searched for in the search result under the search condition specified by the user. More specifically, the noise detection unit 34 performs the noise detection processing in such a manner that the number of non-extracted patents cited by a certain extracted patent is equal to or greater than a predetermined number and there is no (other) extracted patent cited, and When the number of non-extracted patents to which an extracted patent is cited is equal to or larger than a predetermined number and there is no (other) extracted patent to be cited, the certain extracted patent is detected as a noise patent. The noise detection unit 34 executes noise detection processing for each extracted patent.

The noise detection unit 34 detects a noise patent according to the noise patent detection condition included in the search condition. Specifically, the noise detection unit 34 detects a certain extracted patent as a noise patent when all of the following condition 1, condition 2, and condition 3 are satisfied. Condition 1 is that the number of non-extracted patents that a certain extracted patent directly or indirectly cites within the number of cited generations specified in the noise patent detection condition is equal to or greater than the number of citations specified in the noise patent detection condition. Is to be.

Condition 2 is that the number of non-extracted patents that a certain extracted patent is directly or indirectly cited within the number of cited generations specified in the noise patent detection condition is specified in the noise patent detection condition. It is more than the number of citations. In other words, the condition 2 is that the number of non-extracted patents that cite a certain extracted patent within the number of cited generations specified by the noise patent detection condition is equal to or greater than the number of cited references specified by the noise patent detection condition. It is. Condition 3 is that a certain extracted patent is not cited by any of the other extracted patents and is not cited by any of the other extracted patents.

As a modified example, the noise detection unit 34 is configured such that the number of non-extracted patents cited by a certain extracted patent is a predetermined number or more, and/or the number of non-extracted patents cited by the certain extracted patent is a predetermined number. In the above case, the certain extraction patent may be detected as a noise patent. Specifically, the noise detection unit 34 may detect a certain extracted patent as a noise patent when the certain extracted patent satisfies the above condition 1 and/or the above condition 2. Further, the noise detection unit 34 may detect a certain extraction patent as a noise patent when the extraction patent satisfies at least one of the condition 1 and the condition 2. Also in this modification, an extracted patent having a high relevance to a non-extracted patent can be detected as a noise patent.

FIG. 3 shows an example of the citation relationship in the extracted patent. This figure shows seven patent documents A, B, C, D, E, F, and Y extracted according to the search condition of the user. The quotation column in the figure shows other documents (for example, patent documents and non-patent documents) designated by the applicant of the extracted patent or cited in the process of examination/trial of the extracted patent. Further, the cited column in the figure shows other documents in which the extracted patent is cited by the applicant or in the process of examination/trial.

Here, citation number=1, citation number=1, citation generation number=1, and citation generation number=1 are designated as necessary patent detection conditions, and citation number=1, citation number=1 as noise patent detection condition. , The number of cited generations=1 and the number of cited generations=1 are designated. In the example of FIG. 3, patent documents B and X are detected as necessary patents. Among them, the patent document X is not included in the extraction patent, and thus is also detected as a leakage patent. Further, Patent Document Y cites one or more non-extracted patents but does not cite other extracted patents, and is cited by one or more non-extracted patents but not by other extracted patents. Therefore, it is detected as a noise patent.

If Patent Document C is not extracted by the search, it means that Patent Document B is not cited from the extracted patents within the first generation and is out of the necessary patents. However, when the number of cited generations of the required patent detection condition is set to 2 or more, the patent document B will be cited from the patent document C two generations ahead, and therefore the patent document C is not extracted. Detected as a necessary patent. Further, if the number of cited generations of the required patent extraction condition is set to 2 or more, since Patent Document D refers to Patent Document B two generations ahead, Patent Document D is detected as a necessary patent. To be done. In this way, the larger the set values of the number of cited generations and the number of cited generations, the wider the detection range of required patents.

Returning to FIG. 2, the search result generation unit 36 provides information on necessary patents and/or leaked patents detected by the necessary document detection unit 31 and the leakage detection unit 32, and information on noise patents detected by the noise detection unit 34. Generate search result information that includes. The search result generation unit 36 of the embodiment generates (1) information indicating a ratio relating to whether a necessary patent is included in the extracted patent, and (2) generating information indicating a ratio of noise patent in the extracted patent. To do. The search result information includes the number of patents extracted.

For example, as the above (1), the search result generating unit 36 has a ratio of required patents included in the extracted patents (=the required number of patents included in the extracted patents/the total number of required patents) (hereinafter, also referred to as a “necessary document intrinsic ratio”). May be calculated. Further, the search result generation unit 36 may generate information indicating the ratio of the omitted patent and the required patent. For example, the ratio of required patents not included in the extracted patents (=the required number of patents not included in the extracted patents/the total number of required patents) may be calculated. The required number of patents not included in the extracted patents can be said to be the number of leaked patents. Further, the ratio of required patents inside and outside (=the required number of patents included in the extracted patent/the required number of patents not included in the extracted patent) may be calculated.

Further, as the above (2), the search result generation unit 36 may calculate the ratio of noise patents to the extracted patents (=total number of noise patents/total number of extracted patents) (hereinafter also referred to as “noise literature ratio”). .. Further, the ratio of the extracted patents other than the noise patents (=(total number of extracted patents-total number of noise patents)/total number of extracted patents) may be calculated.

The search result providing unit 38 provides the search result information generated by the search result generating unit 36 to the user terminal 16 including search result information including at least one of the omitted patent (or required patent) and the noise patent. To display. The search result providing unit 38 of the embodiment transmits the search result information including the total number of extracted patents, the required document intrinsic rate, and the noise document rate to the user terminal 16. As a modification, the search result providing unit 38 may transmit the other type of ratio information to the user terminal 16 instead of or together with these pieces of information.

FIG. 4 shows an example of a search screen provided by the document search device 10. The user operates the user terminal 16 to access the document search device 10, and the document search device 10 causes the user terminal 16 to display the search screen 50. The user inputs search items/keywords/logical expressions as search conditions on the search screen 50. When the user presses the search button 52, the user terminal 16 transmits a search request including the search conditions input on the search screen 50 to the document search device 10.

FIG. 5 also shows an example of the search screen provided by the document search device 10. The search screen 50 of FIG. 5 includes search result information 64 and can be said to be a search result screen. The search result generation unit 36 sets the number of extracted patents in the first result column 54, sets the required document inherent rate in the second result column 56, and sets the noise document rate in the third result column 58. The value indicated by the search result information 64 in FIG. 5 is a value based on the extraction result in FIG. That is, the number of extracted patents is 7, the required document intrinsic rate is 50% (=1/2), and the noise document rate is 14% (=1/7).

When the list display button 60 is selected on the search screen 50 of FIG. 5, the search result providing unit 38 of the document search device 10 transmits an extracted patent list showing a list of seven extracted patents to the user terminal 16 and displays it. Let The search result providing unit 38 may further transmit a list showing a list of leaked patents and/or noise patents to the user terminal 16. Further, a patent list including the extracted patent and the leaked patent may be transmitted to the user terminal 16 and displayed, and the mark indicating the type may be attached to the leaked patent and the noise patent in the list. Furthermore, the search result providing unit 38 adds the patent patent list information (list) indicating the result of removing the noise patent from the extracted patent while adding the leaked patent to the extracted patent as the information of the patent document to be confirmed by the user. It may be provided to the terminal 16.

When the citation map display button 62 is selected on the search screen 50 of FIG. 5, the user terminal 16 transmits a citation map provision request to the document search device 10, and the document search device 10 causes the citation maps regarding the seven extracted patents. (Described later) is transmitted to the user terminal 16 and displayed. Next, the configuration of the document search device 10 for presenting the citation map to the user will be described.

Returning to FIG. 2, the citation map generation unit 40 arranges a plurality of objects indicating a plurality of extracted patents extracted by the search unit 28 when the request acquisition unit 26 acquires the citation map provision request (in other words, the citation map is displayed). ) Generate a citation map that is an image. An object may be a component of a web page, for example a box or box with a border. Note that the document search device 10 may store the search result by the search unit 28 based on the search request from a certain user terminal 16 in association with the session ID of the user terminal 16. The citation map generation unit 40 identifies the session ID of the user terminal 16 that has transmitted the citation map providing request, and uses the search result (patent document data) stored in advance in association with the identified session ID to create the citation map. May be generated.

The citation map generation unit 40 determines whether the first extraction patent and the second extraction patent among the plurality of extraction patents have a citation relationship (that is, a citation relationship or a citation relationship). A citation map in which an object indicating the first extracted patent and an object indicating the second extracted patent are arranged in association with each other is generated. When associating a plurality of objects, the citation map generation unit 40 of the embodiment connects these objects with an arrow line. The citation map providing unit 42 transmits the data of the citation map generated by the citation map generating unit 40 to the user terminal 16 to display it.

FIG. 6 shows an example of a citation map provided by the document search device 10. This figure shows a two-dimensional citation map showing the citation relationship between the documents shown in FIG. The citation map generation unit 40 arranges, on the citation map 70, seven document objects 76 (A to F, Y) corresponding to the seven extracted patents shown in FIG. Further, the citation map generation unit 40 includes, on the citation map 70, five documents corresponding to other patent documents that cite the extracted patent (patent documents not included in the extracted patent) and other patent documents cited by the extracted patent. The objects 76 (G to J, X) are arranged. The patent documents (extracted patents and other patent documents mentioned above) on which the objects are arranged on the citation map 70 are generically referred to as "target documents" hereinafter. The actual document object 76 includes identification information of patent documents such as an application number.

In the citation map 70, the citation relationship between the target documents is indicated by the arrow line 78. The starting point (original) of the arrow line 78 indicates the object document to be cited, and the ending point (orientation) of the arrow line 78 indicates the object document to be cited. That is, when the first patent document cites the second patent document, the citation map generation unit 40 arranges the object indicating the first patent document at the start point of the arrow line 78 and indicates the second patent document. The object is placed at the end of the arrow line 78.

For example, the citation map generation unit 40 arranges the object indicating the leaked patent and the object indicating the extracted patent having a citation relationship with the leaked patent in the citation map 70 by associating them with each other by connecting them with an arrow line. Further, the citation map generation unit 40 sets an object indicating a noise patent and an object indicating a target document (specifically, another patent document not included in the extracted patent) having a citation relationship with the noise patent, They are arranged on the citation map 70 in association with each other by connecting with an arrow line. As shown in FIG. 6, the citation map 70 of the present embodiment does not generate one patent document as a starting point, but shows the citation relationship of each of the plurality of patent documents extracted by the search. Objects of a plurality of patent documents are network-related.

The citation map generation unit 40 sets a plurality of objects (A to J, X, Y) indicating a plurality of target documents according to dates (application date, priority date, etc.) associated with each of the plurality of target documents. They are arranged in time series on the citation map 70 (reference field 74 in FIG. 6). The citation map generation unit 40 of the embodiment determines the position on the horizontal axis (X axis) of the citation map 70 in which the object of each target document is to be arranged according to the date associated with each target document. For example, the object of the target document having an earlier filing date is assigned a position on the left side, in other words, a smaller X coordinate value is set. Further, the object on the target document whose application date is later is assigned a position on the right side, in other words, a larger X coordinate value is set. Note that the type of date (application date, priority date, etc.) that serves as a reference for arranging objects in the quotation map 70 can be selected by the user.

Further, the citation map generation unit 40 determines the position on the vertical axis (Y axis) of the citation map 70 in which the object of each target document is to be arranged according to the applicant or the author associated with each of the plurality of target documents. To decide. A plurality of lanes corresponding to a plurality of applicants are provided on the vertical axis (applicant column 72) of the citation map 70. The citation map generation unit 40 arranges each target document in the lane of the document column 74 that matches the applicant of each target document.

It should be noted that the display order of the applicant on the Y axis is such that the object displayed by the target applicant and the displayed object by another applicant having a direct citation relationship with this object are When the lane of the applicant and the lanes of the other applicants are arranged adjacent to each other, they are arranged in the respective lanes, and the display mode is easier to see. Here, it is possible to arrange the lanes of the applicant in ascending or descending order of the applicant's name without depending on such a display mode, and further, it is possible to change the order of the lanes of the applicant by designation by the user. May be. Further, in the display of FIG. 6, one applicant is displayed in one lane, but objects of a plurality of applicants may be displayed in one lane. For example, if the number of display objects is a predetermined number or less, The persons may be displayed together in one lane, or may be displayed in the same lane when the applicants have a corporate relationship (such as a relationship between a parent company and a subsidiary). Of course, upon designation by the user, the object of the particular applicant who has been designated may be displayed in one lane.

Further, the citation map generation unit 40 arranges an object indicating a necessary patent, an object indicating a missing patent, an object indicating a noise patent, and an object indicating a target document other than the extracted patent on the citation map 70 in different modes (in other words, in other words). If it is revealed). For example, the citation map generation unit 40 arranges the object indicating the noise patent in the citation map 70 in a mode different from that of the object indicating the extracted patent that is not the noise patent.

In the citation map 70 of FIG. 6, the objects indicating the necessary patents (Document B, Document X) are indicated by thick boxes. Further, the object indicating the leakage patent (reference X) is shown by hatching. Further, the object indicating the noise patent (reference document Y) is indicated by horizontal hatching. Further, the objects indicating the target documents other than the extracted patent are indicated by the dashed boxes. It should be noted that different marks may be added for each type to the object indicating the required patent, the object indicating the omitted patent, the object indicating the noise patent, and the object indicating the target document other than the extracted patent, and different colors for each type. May be set.

In the citation map 70 of FIG. 6, the seven extracted patents are shown as an extracted document group 80, and the patents that match the original search purpose of the user are shown as a target document group 82. The target document group 82 is the one in which the leak patent is added to the extraction patent and the noise patent is excluded. The citation map generation unit 40 may display only the extracted document group 80 on the citation map 70 or only the target document group 82 on the citation map 70 according to the user's instruction. This can reduce the number of documents to be confirmed by the user.

In addition, the citation map generation unit 40 changes the mode of the citation map generation process between the first mode, the second mode, and the third mode according to the mode switching instruction of the citation map generation process transmitted from the user terminal 16. You may switch. The first mode may be a mode in which the citation map 70 in which only objects indicating a plurality of extracted patents (that is, the extracted document group 80) are arranged is generated. The second mode may be a mode in which a citation map 70 in which an object indicating a leaked patent is arranged in addition to an object indicating a plurality of extracted patents is generated.

Although the third mode is not included in a plurality of extraction patents, a patent document that cites at least one extraction patent, or an object indicating a patent document that is cited by at least one extraction patent, It may be a mode in which the citation map 70 arranged in addition to the object shown is generated. In the third mode, in addition to objects indicating a plurality of extracted patents, an object indicating a leaked patent, and an object indicating a target document which is not an extracted patent and not a leaked patent but has a citation relationship with the extracted patent (G to G in FIG. 6). This is a mode for generating the citation map 70 in which J) is arranged. That is, the citation map 70 of FIG. 6 is a citation map generated in the third mode.

When a predetermined selection operation for an object of a specific patent document on the citation map is input on the user terminal 16, the detailed information of the selected patent document is displayed on the user terminal 16. The operation of selecting an object may be a click operation on the object or a mouse over operation of placing the mouse cursor on the object. The user terminal 16 transmits the identification information (application number etc.) of the patent document selected on the citation map to the document search device 10, and the document search device 10 transmits the data of the patent document to the user terminal 16. Good. This communication between the user terminal 16 and the document search device 10 may be realized by Ajax (Asynchronous JavaScript (registered trademark) + XML) communication.

The data of patent documents provided from the document search device 10 may be the full text of the publication, or may be summary data including the title of the invention, claim 1, abstract, representative drawing, and the like. The user terminal 16 may display the patent document data provided from the document search device 10 on a new tab (a tab different from the tab in the citation map display) of the web browser. Further, the user terminal 16 may display the citation map and the patent document data (for example, the summary data) side by side on the same screen of the web browser.

The citation map providing unit 42 may also provide the user terminal 16 with the data of a plurality of patent documents arranged on the citation map when the citation map is provided. In this case, the Javascript (registered trademark) program embedded in the web page including the citation map is executed by the web browser of the user terminal 16 to display the detailed information of the patent document selected on the citation map on the web browser screen. You may let me.

The operation of the document search device 10 having the above configuration will be described.
FIG. 7 is a flowchart showing the operation of the document search device 10. A user who should search for patent documents inputs search conditions on the search screen 50 shown in FIG. 4 and inputs a search execution operation. The user terminal 16 transmits a search request including the search conditions input to the search screen 50 to the document search device 10. When the request acquisition unit 26 receives the search request transmitted from the user terminal 16 (Y of S10), the search unit 28 selects the search condition included in the search request from the plurality of patent documents stored in the document DB 12. A patent document satisfying the above conditions is searched (S12).

The data of each extracted patent searched by the search unit 28 includes data indicating a citation relationship with other documents. The necessary document detection unit 31 and the leakage detection unit 32 detect the required patent and the leakage patent according to the citation relationship of each extracted patent (S14). Further, the noise detection unit 34 detects the noise patent according to the citation relationship of each extracted patent (S16). The search result generation unit 36 calculates a value related to the number and ratio of necessary patents or omitted patents, a value related to the number and ratio of noise patents, and generates search result data including these calculated values (S18). .. The search result providing unit 38 transmits the search result data generated by the search result generating unit 36 to the user terminal 16 (S20). The search screen 50 of FIG. 5, for example, is displayed on the user terminal 16.

As described above, the document search device 10 of the embodiment presents an objective index value indicating the validity of the search condition designated by the user to the user. By checking the required document inclusion rate and the noise document rate in the search result information 64 in FIG. 5, the user can easily determine the validity of the search condition (search formula) before reading the extracted document and the like. For example, if the required document intrinsic rate is high and the noise document rate is low, it can be determined that the relevance of the search condition is high. On the other hand, if the required document intrinsic rate is low, that is, there are many leaked patents, and it can be determined that the relevance of the search condition is low. If the noise literature rate is high, it can be determined that the search item should be added. Further, in the embodiment, the number of citations, the number of citations, and the number of generations can be specified by the user as the detection conditions for the leakage patent (necessary patent) and the noise patent. As a result, it is possible to finely specify the detection conditions for the leaked patent (necessary patent) and the noise patent according to the theme of the document to be searched (for example, the type and characteristics of the invention).

When the user selects the citation map display button 62 on the search screen 50 of FIG. 5, the user terminal 16 transmits a citation map providing request to the document search device 10. At that time, the user specifies whether or not only the extracted patent is to be mapped, and the user terminal 16 transmits a citation map provision request including the designation of mapping to the document search device 10. When the request acquisition unit 26 receives the citation map provision request transmitted from the user terminal 16 (Y of S22) and only the extraction patent is the mapping target (Y of S24), the citation map generation unit 40 only extracts the extraction patent. A citation map in which is arranged is generated (S26). For example, the citation map of only the extracted document group 80 in FIG. 6 is generated.

On the other hand, if there is a mapping target other than the extracted patent (N in S24), the citation map generation unit 40 arranges the patent documents which are not included in the extracted patent but have a citation relationship with the extracted patent in addition to the extracted patent. The cited citation map is generated (S28). In S28, for example, the citation map 70 of FIG. 6 is generated. The citation map providing unit 42 transmits the data of the citation map generated by the citation map generation unit 40 to the user terminal 16 and displays it (S30). If the quotation map providing request has not been received (N of S22), the processing of S24 and thereafter is skipped. If the search request has not been received (N of S10), the processing of S12 and thereafter is skipped.

The user can easily understand the citation relationship between the extracted patents by checking the citation map. For example, extracted patents cited by many patents can be grasped as basic technology patents. Further, in the quotation map of the embodiment, the required patent, the leaked patent, and the noise patent are displayed in different modes. As a result, the user can easily identify the necessary patents, the leaked patents, and the noise patents, and intuitively understand the inherent ratio of the required patents, the ratio of the leaked patents, the ratio of the noise patents, and the like. Further, in the citation map of the embodiment, the horizontal axis is the time axis such as the filing date and the vertical axis is the applicant, so that the user can intuitively grasp additional information such as the filing date and the applicant. it can.

The present invention has been described above based on the embodiments. It should be understood by those skilled in the art that this embodiment is merely an example, and that various modifications can be made to the combination of each component and each processing process, and that such modifications are within the scope of the present invention.

As a first modification, a modification regarding a three-dimensional citation map will be described.
In the above-described embodiment, the object of the target document is arranged at a position different in the vertical axis (Y axis) for each applicant, but as a modification, the object of the target document is different position in the depth direction (Z axis) for each applicant. It may be placed at. Specifically, the citation map generation unit 40 determines the object indicating the first target document and the second target document when the applicant or the author is different between the first target document and the second target document. A three-dimensional citation map (also referred to as a three-dimensional graph) in which the objects shown are arranged at different positions in the depth direction may be generated.

FIG. 8 shows an example of a citation map provided by the modified document search device 10. The citation map of FIG. 8 corresponds to the citation map of FIG. 6 and is a three-dimensional citation map showing the citation relationship between documents shown in FIG. However, FIG. 8 shows only the applicants a, b, c, and f, and the XY plane is provided for each of those applicants. The citation map generation unit 40 assigns different positions in the depth direction (Z axis) of the citation map 70 for each applicant of the target document. For example, the citation map generation unit 40 arranges the target document of a certain applicant on the XY plane of the Z coordinate value assigned to the applicant. In FIG. 8, the XY plane of the applicant f and the target document (X) are shown by solid lines, the XY plane of the applicant a and the target documents (A, B, C) are shown by broken lines, and the XY plane and the target of applicant b are shown. Documents (C, F) are indicated by a chain line, and the applicant's c XY plane and target documents (E, Y) are shown by a chain line.

Here, in FIG. 8, target documents of each applicant are displayed on each XY plane on the Z axis for each applicant, and the display order of each applicant on the X axis is the same as that of the target applicant. The displayed object and the displayed object of another applicant having a direct citation relationship with this object are such that the XY plane of the applicant of this subject and the XY plane of this other applicant are adjacent to each other. In the case of the above configuration, it is possible to display the citation relationship between the objects arranged on the respective XY planes in a display mode that is easier to see. Here, the XY plane of the applicant on the Z-axis can be arranged in ascending or descending order of the applicant's name regardless of such a display mode, and further, the applicant's name on the Z-axis can be arranged. It is also possible to arrange the XY plane. Further, in the display of FIG. 8, one applicant is displayed on one XY plane, but objects of a plurality of applicants may be displayed on one XY plane. For example, the number of display objects is a predetermined number or less. Applicants may be displayed together on one XY plane, or may be displayed on the same XY plane when the applicants have a corporate relationship (relationship between parent company and subsidiary, etc.). .. Of course, the user's designation may be received, and the object of the particular applicant who has been designated may be displayed on one XY plane.

The lateral direction (X-axis direction) of the citation map 70 in FIG. 8 is the time direction as in FIG. 6, and the citation map generation unit 40 arranges the target documents with earlier application dates (or priority dates) closer to the origin side. .. In other words, the X coordinate value is increased as the target document has a later application date (or priority date). Further, the citation map generation unit 40 arranges a plurality of target documents having the same or similar application date (or priority date) in the vertical direction (Y-axis direction) of the citation map 70.

When the first target document of the first applicant cites the second target document of the second applicant, the citation map generation unit 40 sets the document object 76 of the second target document as a starting point. , An arrow line 78 ending at the document object 76 of the first target document. In this case, the citation map generator 40 sets an arrow line 78 from the XY plane of the second applicant to the XY plane of the first applicant.

Even in the three-dimensional citation map, the citation map generation unit 40 makes the object indicating the required patent, the object indicating the omitted patent, the object indicating the noise patent, and the object indicating the target document other than the extracted patent different from each other. Place it on the citation map with. For example, in FIG. 8, a bold box is set for the object (B, X) indicating the required patent. Further, the hatched shading is set for the object (X) indicating the leak patent, and the shading of the horizontal line is set for the object indicating the noise patent (Y).

The quotation map generation unit 40 sets the Z axis of the quotation map 70 so as to be in the depth direction viewed from the user (depth direction of the screen), and for example, makes the Z axis of the quotation map 70 parallel to the user's line-of-sight direction. It may be set to. Further, the citation map generation unit 40 may allow the users to simultaneously check the XY planes of a plurality of applicants by setting a predetermined transparency on the XY planes of the applicants. Further, the citation map generation unit 40 may set the citation map 70 so that the position on the Z axis for displaying the XY plane can be switched according to the user's operation. According to the first modification, even if a large number of target documents exist, the user can intuitively grasp the existence and number of target documents (further required patents, leaked patents, noise patents) for each applicant. Can help.

In the first modified example, different positions in the depth direction (Z axis) of the citation map 70 are assigned to each applicant of the target document, but the horizontal axis (X axis) direction of the citation map 70 is assigned to each applicant of the target document. Different positions may be assigned, or different positions in the vertical axis (Y axis) direction may be assigned. In this case, the information assigned to the X-axis and the Y-axis in FIG. 8 may be represented by the position in the direction of another axis (for example, the Z-axis).

As a second modification, another modification of the three-dimensional citation map will be described.
The citation map generation unit 40 is a divisional application in which the first target document (that is, the extracted patent, the leaked patent, or another patent having a citation relationship with the extracted patent) is a divisional application whose original application is the second target document, or , If the first target document is a priority claim application having the second target document as a basic application, the object indicating the first target document and the object indicating the second target document are defined in the depth direction ( A three-dimensional citation map (also called a three-dimensional graph) arranged at different positions (on the Z axis) may be generated.

For example, the citation map generation unit 40 has (1) an XY plane for the original application (or no divisional application) and (2) a child application (in other words, division 1) at different positions on the Z axis of the citation map 70 of FIG. The XY plane for the (generation) and (3) the XY plane for the grandchild application (in other words, the divided second generation) may be arranged. The citation map generation unit 40 arranges the object of the target document on the XY plane of (1) above when the target document is the original application, and on the XY plane of (2) above when the target document is a child application. You may arrange the object of the said target document.

Further, the citation map generation unit 40 arranges (1) the XY plane for the basic application and (2) the XY plane for the priority claim application at different positions on the Z axis of the citation map 70 of FIG. Good. When the target document is a basic application, the citation map generation unit 40 arranges the object of the target document on the XY plane of (1) above, and when the target document is a priority claim application, the XY plane of (2) above. You may arrange|position the object of the said target document on the top. Note that the XY plane in the second modification may be a citation map in the format of FIG. 6 in which the horizontal axis is the time axis and the vertical axis is the applicant axis.

In divisional applications (priority claim applications), the citation relationship with other documents is often similar to that of the original application (basic application). When the divisional application (priority claim application) and the original application (basic application) are arranged in the same position on the same plane, the number of objects on the same plane becomes large, and as a result of the complicated arrow lines, the citation map is easy to read and read. May deteriorate. According to the second modified example, by arranging an object indicating a divisional application (priority claim application) and an object indicating an original application (basic application) at different positions in the depth direction, an object and an arrow drawn on the same plane. The number of lines can be reduced and the readability and legibility of the citation map can be suppressed.

In the second modified example, different positions in the depth direction (Z axis) of the citation map 70 are assigned to different types of applications (eg, original application, divisional application, priority claiming application). Different positions in the horizontal axis (X axis) direction of 70 may be assigned, and different positions in the vertical axis (Y axis) direction may be assigned. In this case, the information assigned to the X-axis and the Y-axis in FIG. 8 may be represented by the position in the direction of another axis (for example, the Z-axis).

As a third modification, a modification regarding the display mode of the quotation map will be described.
The citation map generation unit 40 may arrange a plurality of objects in different modes for each applicant instead of arranging a plurality of objects indicating a plurality of target patents for which applicants are different in different vertical positions of the citation map. Good. For example, different colors and different marks may be added to each applicant. In addition, the citation map generation unit 40 may arrange the arrow lines indicating the citation relationship in the citation map by adding information indicating the citation type, and each arrow line may have a different mode (for example, a different color or a different color for each citation type). Different line types) may be set. The citation type includes the type of subject and the timing of citations that have established citation relationships.Specifically, the citation by the applicant, the citation at the time of examination, the citation at the time of trial, the citation of notice of reasons for refusal, and the citation of decision of refusal. Including etc. Furthermore, the citation map generation unit 40, in the vicinity of the object of each target document, information indicating the number of citations of each target document (the number of citations of other target documents) and the number of citations of each target document (other Information indicating the number of pieces cited by the target document) may be set.

A fourth modification will be described.
As mentioned above, the citation relationship between documents may be declared by the applicant or the author, and/or in place of this, is established in the process of examination or trial (notification of reasons for refusal, etc.). Sometimes. Although not mentioned in the embodiments, the necessary document detecting unit 31, the leak detecting unit 32, and the noise detecting unit 34 exclude the citation relations declared by the applicant or the author, and set the citations set in the process of examination and trial. Necessary patents, leaked patents, and noise patents may be detected based on the relationship. In other words, the citation relations declared by the applicant or the author are not reflected in the detection of necessary patents, leaked patents, and noise patents, and only the citation relations established in the process of examination and trial are necessary. It may be reflected in the detection of a patent. The citation relationship declared by the applicant or the author is relatively inaccurate, and the citation relationship established in the process of examination and trial is relatively accurate. According to the fourth modification, it is possible to improve the accuracy of detection of required patents, leaked patents, and noise patents.

When the document DB 12 further holds, as the cited document information for a certain patent document, information indicating a main cited document (so-called main reference) among a plurality of cited documents, as a different aspect of the fourth modification, The necessary document detection unit 31, the leakage detection unit 32, and the noise detection unit 34 detect necessary patents, leakage patents, and noise patents based on the citation relation with the main reference among the citation relations set in the process of examination and trial. You may. In other words, only the citation relationship with the document that is the main reference may be reflected in the detection of necessary patents, leaked patents, and noise patents. According to this aspect, it is possible to further improve the accuracy of detection of required patents, leaked patents, and noise patents.

A fifth modified example related to the fourth modified example will be described.
Although not mentioned in the above-mentioned embodiment, the citation map generation unit 40 excludes the citation relationship declared by the applicant or the author, and indicates the target document based on the citation relationship set in the process of examination and trial. A citation map may be created that associates the objects. In other words, the citation map generation unit 40 does not reflect the citation relationship declared by the applicant or the author, but sets the arrow line between the objects by reflecting only the citation relationship set in the process of examination and trial. Good. According to the fifth modified example, it is possible to increase the accuracy of the citation relationship indicated by the citation map, in other words, to generate a citation map indicating the connection of documents whose contents are actually similar.

When the document DB 12 further holds, as the cited document information for a certain patent document, information indicating a main cited document (a so-called main reference) among a plurality of cited documents, as a different aspect of the fifth modification, The citation map generation unit 40 may generate the citation map based on the citation relation with the main reference among the citation relations set in the process of examination and trial. In other words, the arrow lines between objects may be set by reflecting only the citation relationship with the document that is the main reference. According to this aspect, it is possible to show documents having a high degree of similarity in content in the citation map in association with each other, and to further enhance the usefulness of the citation map.

As a sixth modified example, a modified example in which the search condition is automatically proposed to the user will be described.
The document search device 10 may further include a first extraction unit that extracts a phrase common to a plurality of required patents. The first extraction unit may extract a common keyword (or key phrase) included in document data of required patents (or extracted patents) of a predetermined ratio (for example, 30%) or more. The search result generation unit 36 may include the word or phrase extracted by the first extraction unit in the search result data as a candidate to be added to the search condition, for example, a word or phrase to be added to the current search expression under the OR condition. The predetermined ratio may be determined based on the knowledge and experience of the literature information service company, or an appropriate value may be determined by an experiment using the information system 100. The common keyword may be a technology classification code (for example, IPC), or may be a character string described in the claims and the like.

Further, the document search device 10 further includes a second extraction unit that extracts, as characteristic words of the leaked patent, phrases that are included in the leaked patent (which is a subset of the required patent) but not included in the other required patents. You may prepare. The word/phrase extracted by the second extraction unit can be said to be a word/phrase included only in the omitted patent among the necessary patents, or a word/phrase not included in the extracted document among the necessary cases. The second extraction unit may extract a term included in the leaked patent but not included in the other necessary patents using a known technique such as a chi-square test. The search result generation unit 36 may include the word/phrase extracted by the second extraction unit in the search result data as a candidate to be added to the search condition, for example, a word/phrase to be added to the current search expression under the OR condition.

Further, the document search device 10 may further include a third extraction unit that extracts a phrase included in the noise patent but not included in the necessary patent as a characteristic word of the noise patent. The third extraction unit may extract a characteristic word of the noise patent, which is included in the noise patent but not included in the necessary patent, by using a known technique such as a chi-square test. The search result generation unit 36 may include the characteristic word of the noise patent extracted by the third extraction unit in the search result data as a keyword candidate of the negative search (excluded search, excluded search) to be added to the current search formula. .. The document search device 10 may include one or more of the first extraction unit, the second extraction unit, and the third extraction unit. According to the sixth modification, it is possible to provide the user with useful information for improving the validity of the search condition.

As a seventh modified example, a modified example will be described in which information on a leaked patent and a noise patent is presented for each search item.
As described in part with reference to FIG. 4, the search condition acquired by the request acquisition unit 26 may include a plurality of search items (and keywords). The search unit 28 may extract patent documents that satisfy each of the plurality of search items, and may extract patent documents that satisfy the combination of the plurality of search items. In addition to the configuration of the above-described embodiment, the necessary document detection unit 31 and the leakage detection unit 32 use the patent document group extracted for each search item as a population, and search based on the citation relationship of each document in the population. Necessary patents and leaked patents may be detected for each item. In addition to the configuration of the above-described embodiment, the noise detection unit 34 sets the patent document group extracted for each search item as a population, and based on the citation relation of each document in the population, the noise patent for each search item. It may be detected. In addition to the configuration of the above-described embodiment, the search result providing unit 38 may output information regarding at least one of a leakage patent (necessary patent) and a noise patent in association with each of a plurality of search items.

The search result generation unit 36 may generate information on the number or ratio of the required patents or the omitted patents for each search item (for example, the required document inclusion rate) based on the required number of patents and the number of omitted patents for each search item. .. Further, the search result generation unit 36 may generate information on the number and ratio of noise patents for each search item (for example, the noise document rate) based on the number of noise patents for each search item. The search result generation unit 36 may set the required document inclusion rate and noise document rate for each search item next to the number of hits for each search item in the search screen 50 of FIG. According to the seventh modified example, it becomes easier for the user to understand the search items (and keywords) that contribute to the appropriate narrowing down of documents and the search items (and keywords) that cause noise inclusion, and the search conditions It is possible to provide the user with useful information for improving the relevance.

As a different aspect of the seventh modified example, detection of necessary patents, leaked patents, and noise patents may be suppressed for search items in which the number of extracted patents is equal to or greater than a predetermined threshold value among the plurality of search items. This threshold value may be determined based on the knowledge and experience of the document information service company, or an appropriate value may be determined by an experiment using the information system 100. It may also be set by the user. The threshold may be 1,000 or 10,000, for example. According to this aspect, it is possible to prevent the detection time of required patents, leaked patents, and noise patents from being lengthened, and to suppress the response time until the search result data is returned in response to the search request.

Any combination of the above-described examples and modifications is also useful as an embodiment of the present invention. The new embodiment resulting from the combination has the effects of the combined examples and modifications. It is also understood by those skilled in the art that the functions to be fulfilled by the respective constituent elements described in the claims are realized by the individual constituent elements shown in the embodiments and the modified examples or by their cooperation.

10 document retrieval device, 12 document DB, 26 request acquisition unit, 28 retrieval unit, 30 detection unit, 31 required document detection unit, 32 leakage detection unit, 34 noise detection unit, 36 retrieval result generation unit, 38 retrieval result providing unit, 40 citation map generation unit, 42 citation map providing unit, 100 information system.

Claims (17)

An acquisition unit that is a condition input by a user and that acquires a condition for extracting a specific document from a plurality of documents that may have a citation relationship with another document,
A search unit that extracts a plurality of documents satisfying the conditions acquired by the acquisition unit from the plurality of documents stored in a predetermined storage unit,
Based on the citation relationship between each of the plurality of extracted documents extracted by the search unit and another document, a document that is not included in the plurality of extracted documents but is expected to have high relevance to the plurality of extracted documents And a detection unit that detects at least one of a noise document that is a document included in the plurality of extracted documents and that is expected to have high relevance to the non-extracted document,
An output unit that presents information about a leaked document or a noise document detected by the detection unit to a user,
Equipped with
The detection unit is a document that cites at least one of the plurality of extracted documents, and a necessary document detection unit that detects a document cited by at least one of the plurality of extracted documents as a necessary document. document retrieval apparatus comprising. The document according to claim 1 , further comprising a leakage detection unit that detects, as the leakage document, required documents that are not included in the plurality of extracted documents among the necessary documents detected by the necessary document detection unit. Search device. The required document detection unit is a document that cites extracted documents with a number of citations specified by a user or more, and detects a document cited by an extracted document with a number of citations specified by a user or more as the required document. The document search device according to claim 1 or 2 , wherein The acquisition unit is the number of generations from the first document to the second document when a direct or indirect citation relationship is established between the first document and the second document, Accepting the specification of the number of generations which is the detection condition of the required documents from the user,
The required document detection unit is a document that directly or indirectly cites at least one of the plurality of extracted documents within the number of generations designated by the user, and the number of generations designated by the user. The document search according to any one of claims 1 to 3 , wherein a document directly or indirectly cited by at least one of the plurality of extracted documents within the range is detected as the required document. apparatus. The output unit, document search apparatus according to claim 1, wherein the presenting information indicating a ratio as to whether required documents included in the extracted documents to the user 4. The output unit, document search apparatus according to any one of claims 1 to 4, characterized by presenting information indicating a ratio of a leak literature and required documents to the user. When the number of non-extracted documents cited by a certain extracted document is a predetermined number or more, and/or the number of non-extracted documents cited by the certain extracted document is a predetermined number or more, literature search apparatus according to any one of claims 1 to 6, characterized in that it comprises a noise detector for detecting the extracted documents the certain as the noise reference. The detection unit is configured such that the number of non-extracted documents cited by a certain extracted document is equal to or larger than a predetermined number and no extracted document is cited, and the number of non-extracted documents cited by the certain extracted document is a predetermined number. If there is no extraction cited be more than, the literature search apparatus according to any one of claims 1 to 6, characterized in that it comprises a noise detector for detecting the extracted documents the certain as the noise literature .. In the noise detection unit, the number of non-extracted documents cited by a certain extracted document is equal to or greater than the number of citations designated by the user, and/or the number of non-extracted documents cited by the certain extracted document is determined by the user. 9. The document search device according to claim 7 , wherein the certain extracted document is detected as the noise document when the number of times is equal to or more than the number of citations specified by. The acquisition unit is the number of generations from the first document to the second document when a direct or indirect citation relationship is established between the first document and the second document, The user receives the designation of the number of generations, which is the detection condition of the noise document,
In the noise detection unit, the number of non-extracted documents that an extracted document directly or indirectly cites within a range of the number of generations specified by the user is a predetermined number or more, and/or is specified by the user. Detecting the certain extracted document as the noise document when the number of non-extracted documents that are directly or indirectly cited by the certain extracted document within a range of the number of generations is equal to or more than a predetermined number. The document search device according to any one of claims 7 to 9 , which is characterized. Document retrieval apparatus according to any one of the output unit, according to claim 1 to 10, characterized in that for presenting information indicating a ratio of noise documents to the user in the extraction literature. The plurality of extracted documents are publications related to industrial property rights,
The detection unit excludes the citation relationship declared by the applicant from the citation relationship between each extracted document and other documents, and based on the citation relationship set in the examination, the leakage document and the noise document The document search device according to any one of claims 1 to 11 , wherein at least one is detected. The document search device according to claim 1 , further comprising an extraction unit that extracts a word/phrase common to a plurality of necessary documents. The document search device according to claim 2 , further comprising an extraction unit that extracts a word or phrase included in the leaked document but not included in other necessary documents as a characteristic word of the leaked document. The document search device according to claim 1 , further comprising an extraction unit that extracts a word or phrase included in the noise document but not included in the necessary document as a characteristic word of the noise document. A step of acquiring a condition for extracting a specific document from a plurality of documents which are conditions inputted by the user and may have a citation relationship with other documents,
A step of extracting a plurality of documents satisfying the conditions acquired in the acquisition step from the plurality of documents stored in a predetermined storage unit;
Based on the citation relationship between each of the plurality of extracted documents extracted in the extracting step and another document, it is not included in the plurality of extracted documents but is expected to have high relevance to the plurality of extracted documents. Leaked document which is a document, and a step of detecting at least one of a noise document that is a document that is expected to be highly related to the non-extracted document although it is included in the plurality of extracted documents,
Presenting to the user information about the leakage or noise literature detected in the step of detecting,
Was the computer is running,
The step of detecting includes the step of detecting as a required document a document that cites at least one of the plurality of extracted documents and that is cited by at least one of the plurality of extracted documents. Document retrieval method characterized by. A function that is a condition input by a user and that acquires a condition for extracting a specific document from a plurality of documents that may form a citation relationship with another document,
From a plurality of documents stored in a predetermined storage unit, a function of extracting a plurality of documents satisfying the conditions acquired by the acquisition function,
Based on the citation relationship between each of the plurality of extracted documents extracted by the extracting function and another document, it is not included in the plurality of extracted documents but is expected to have high relevance to the plurality of extracted documents. A leaked document that is a document, and a function of detecting at least one of a noise document that is a document that is included in the plurality of extracted documents but is expected to be highly related to non-extracted documents,
A function of presenting to a user information regarding a leaked document or a noise document detected by the function of detecting,
To the computer ,
The detecting function is a document that cites at least one of the plurality of extracted documents, and includes a function of detecting a document cited by at least one of the plurality of extracted documents as a required document. A computer program characterized by .

Images