JP6053131B2 - Information processing apparatus, information processing method, and program - Google Patents

Description

  The present invention relates to an information processing apparatus for selecting a patent document using a screening result of a patent search.

  Conventionally, devices for managing the results of patent searches have been developed (see, for example, Patent Document 1). Furthermore, an apparatus for creating a search expression for collecting patent documents similar to a specific patent document has been developed (see, for example, Patent Document 2).

JP 2007-224004 A JP 2011-59843 A

  However, in the conventional patent search, related patents and unrelated patents are classified manually, which is a very time-consuming operation. Furthermore, even if there is an error in a patent document that has been classified as an unrelated patent, it is rarely reconsidered, leaving the wrong classification result.

The information processing apparatus according to the first aspect of the present invention is a related patent identification in which one or more related patent identification information, which is information specifying a related patent document that is determined to be related as a result of a screening operation in patent search, is stored. One or more stored in the related patent identification information storage unit using elements included in the patent document identified by the information storage unit and one or more related patent identification information stored in the related patent identification information storage unit A search expression generation unit that generates a search expression that can acquire a patent document identified by related patent identification information, and a search patent document that is a patent document acquired by using the search expression generated by the search expression generation unit 1 is a feature vector of a related patent document identified by a search patent identification information acquisition unit that acquires search patent identification information that is information to be stored and each related patent identification information stored in the related patent identification information storage unit A feature vector that acquires the above related patent feature vector and that acquires one or more search patent feature vectors that are feature vectors of the search patent document identified by each search patent identification information acquired by the search patent identification information acquisition unit Of the one or more search patent feature vectors acquired by the acquisition unit and the feature vector acquisition unit, the search patent corresponding to the search patent feature vector similar to the feature of the one or more related patent feature vectors acquired by the feature vector acquisition unit A selection unit that selects at least search patent identification information that is identification information and does not match the related patent identification information stored in the related patent identification information storage unit, and an output that outputs information related to a result selected by the selection unit An information processing apparatus.
With this configuration, the information processing apparatus can search for a patent document similar to the related patent document selected as the related patent in the patent search and select whether the similar patent document is a related patent document. Thereby, for example, the information processing apparatus can efficiently acquire a patent document similar to the related patent document among the patent documents that have not been searched in the screening work in the patent search, and can select the related patent from the acquired patent documents. .

In addition, the information processing apparatus of the second aspect of the present invention is related to the first aspect of the invention, the related patent identification information stored in the related patent identification information storage unit among the search patent identification information selected by the selection unit. As the number of search patent identification information items that do not match increases, an evaluation unit that further evaluates lower is further provided, and the output unit is an information processing device that outputs a result of evaluation performed by the evaluation unit.
With this configuration, the information processing apparatus can perform a lower evaluation as the number of patent documents similar to related patent documents that have not been investigated in the sorting operation in the patent search increases. Thereby, the information processing apparatus can perform a lower evaluation as the number of related patents that are not included in the search target increases, for example, in the patent search.

Further, in the information processing apparatus according to the third aspect of the present invention, in the first or second aspect of the invention, the selection unit includes a feature vector acquisition unit out of one or more search patent feature vectors acquired by the feature vector acquisition unit. It is the search patent identification information corresponding to the search patent feature vector similar to the feature of the acquired one or more related patent feature vectors, and is identified by the related patent identification information stored in the related patent identification information storage unit The information processing apparatus also selects search patent identification information corresponding to a patent document.
With this configuration, the information processing apparatus can also select related patent documents selected as related patents in patent search. Thereby, for example, the information processing apparatus can not select a patent document that has been erroneously selected as a related patent in a sorting operation in a patent search.

Further, in the information processing apparatus according to the fourth aspect of the present invention, in contrast to the second aspect of the invention, the selection unit acquires 1 of the one or more search patent feature vectors acquired by the feature vector acquisition unit. Patent documents identified by the related patent identification information stored in the related patent identification information storage unit, which is the search patent identification information corresponding to the search patent feature vector similar to the characteristics of the related patent feature vector. Corresponding search patent identification information is also selected, the evaluation unit is the search patent identification information acquired by the search patent identification information acquisition unit, and does not match the search patent identification information selected by the selection unit, the related patent identification information This is an information processing apparatus that performs a lower evaluation as the related patent identification information stored in the storage unit increases.
With this configuration, the information processing apparatus can also evaluate related patent documents selected as related patents in patent searches. Thereby, for example, the information processing apparatus can perform a lower evaluation as the number of patent documents erroneously selected as related patent documents in a patent search increases.

The information processing apparatus according to the fifth aspect of the invention is an unrelated patent document that is a patent document that is determined to be irrelevant as a result of the sorting work in the patent search for any one of the first to fourth inventions. And further includes an unrelated patent identification information storage unit that stores one or more unrelated patent identification information that is information for identifying the search expression generation unit. A search expression that does not acquire at least a part of one or more unrelated patent documents stored in the unrelated patent identification information storage unit by further using an element included in the patent document identified by the unrelated patent identification information. An information processing apparatus to be generated.
With this configuration, the information processing apparatus can generate a search expression that takes into account unrelated patents. Thereby, for example, the information processing apparatus can omit unnecessary patent documents from the search result.

Further, in the information processing apparatus of the sixth invention, in contrast to the fifth invention, the feature vector acquisition unit is identified by each non-related patent identification information stored in the non-related patent identification information storage unit. One or more unrelated patent feature vectors, which are feature vectors of related patent documents, are further acquired, and the selection unit acquires 1 of the one or more search patent feature vectors acquired by the feature vector acquisition unit. This is an information processing apparatus that selects search patent identification information corresponding to a search patent feature vector that is not similar to the features of the above unrelated patent feature vectors.
With this configuration, the information processing apparatus can select whether the search patent document is a related patent in consideration of an unrelated patent. Thereby, for example, the information processing apparatus can perform more accurate selection by using an unrelated patent feature vector.

In addition, the information processing apparatus according to the seventh aspect of the invention relates to any one of the first to fourth aspects of the invention. And further includes an unrelated patent identification information storage unit that stores one or more unrelated patent identification information that is information for identifying the non-related patent identification information storage unit. One or more unrelated patent feature vectors that are feature vectors of unrelated patent documents identified by the patent identification information are further acquired, and the selection unit includes, among the one or more search patent feature vectors acquired by the feature vector acquisition unit, The information processing apparatus selects search patent identification information corresponding to a search patent feature vector that is not similar to a feature of one or more unrelated patent feature vectors acquired by a feature vector acquisition unit.
With this configuration, the information processing apparatus can select whether the search patent document is a related patent in consideration of an unrelated patent. Thereby, for example, the information processing apparatus can perform more accurate selection by using an unrelated patent feature vector.

In the information processing apparatus according to the eighth aspect of the invention, in the sixth or seventh aspect of the invention, the selection unit is a feature vector acquisition unit among one or more unrelated patent feature vectors acquired by the feature vector acquisition unit. The unrelated patent identification corresponding to the unrelated patent feature vector, which is similar to the feature of one or more related patent feature vectors acquired by, and not similar to the feature of one or more unrelated patent feature vectors acquired by the feature vector acquisition unit An information processing apparatus that also selects information.
With this configuration, the information processing apparatus can also select patent documents selected as unrelated patents in the patent search. As a result, for example, the information processing apparatus can select a patent document that has been selected as an unrelated patent document in the screening work in the patent search but may not be an unrelated patent.

In addition, the information processing apparatus of the ninth aspect of the invention is an information for specifying an unrelated patent document that is a patent document that is determined to be irrelevant as a result of the sorting work in the patent search for the second or third invention. And further includes an unrelated patent identification information storage unit storing one or more unrelated patent identification information, and the feature vector acquisition unit includes each unrelated patent identification information stored in the unrelated patent identification information storage unit. One or more unrelated patent feature vectors that are feature vectors of the unrelated patent document to be identified are further acquired, and the selection unit acquires a feature vector from among the one or more unrelated patent feature vectors acquired by the feature vector acquisition unit. The related patent feature vector is similar to the feature of one or more related patent feature vectors acquired by the department, and is not similar to the feature of one or more unrelated patent feature vectors acquired by the feature vector acquisition unit. Also selected unrelated patent identification information corresponding to the vector, the evaluation unit is a low evaluation The more unrelated patent identification information selected by the selecting unit, an information processing apparatus.
With this configuration, the information processing apparatus can also evaluate patent documents selected as unrelated patents in patent searches. Thereby, for example, the information processing apparatus is sorted into unrelated patent documents by sorting work in patent search, but the lower the number of patent documents that may not be unrelated patents, the lower the evaluation can be made.

Further, in the information processing apparatus according to the tenth aspect of the present invention, the feature vector acquisition unit acquires each patent document identified by information specifying each patent document, relative to any one of the first to ninth inventions The information processing apparatus acquires one or more elements from each patent document and acquires each feature vector using the one or more elements.
With this configuration, the information processing apparatus can create a feature vector based on the elements described in the patent document. Thereby, for example, the information processing apparatus can acquire a feature vector without using an external server or the like.

In addition, in the information processing apparatus according to the eleventh aspect of the present invention, in any one of the first to tenth aspects, the feature vector acquisition unit converts similar elements acquired from the respective patent documents into the same vector element. This is an information processing apparatus that acquires each feature vector as a corresponding element.
With this configuration, the information processing apparatus can acquire a feature vector having another expression having the same meaning as the same vector element among elements described in the patent document. Thereby, for example, similar feature vectors can be acquired from similar documents created using different expressions.

  The information processing apparatus according to the present invention can be selected when the patent document is a related patent.

Conceptual diagram of a system including an information processing apparatus according to Embodiment 1 Block diagram of the information processing apparatus in the embodiment The flowchart which shows operation | movement of the information processing apparatus in the embodiment The flowchart which shows the operation | movement regarding the process which produces | generates the search formula of the information processing apparatus in the embodiment The flowchart which shows the operation | movement regarding the process which acquires the feature vector of the information processing apparatus in the embodiment The flowchart which shows the operation | movement regarding the process which selects the related patent document of the information processing apparatus in the embodiment The figure which shows an example of the patent document stored in the patent document storage part in the embodiment The figure which shows an example of the similar element stored in the similar element storage part in the embodiment The figure which shows an example of the information which identifies the patent document stored in each patent identification information storage part in the embodiment The figure which shows an example of the feature vector of each patent document which the feature vector acquisition part in the embodiment acquired The figure which shows an example of the display of the information processing apparatus in the embodiment The figure for demonstrating the relationship of each patent document in the embodiment The figure which shows an example of the external appearance of the computer system in the said embodiment The figure which shows an example of a structure of the computer system in the said embodiment.

  Hereinafter, embodiments of an information processing apparatus and the like will be described with reference to the drawings. In addition, since the component which attached | subjected the same code | symbol in embodiment performs the same operation | movement, description may be abbreviate | omitted again.

(Embodiment 1)
In the present embodiment, an information processing apparatus 1 that selects a related patent using a result of sorting work in a patent search will be described.

  FIG. 1 is a conceptual diagram of a system including an information processing apparatus 1 according to the present embodiment. In FIG. 1, an information processing apparatus 1 and one or more user terminals 2 are connected via a network 100. The network 100 is a wired or wireless communication line, such as the Internet, an intranet, a LAN (Local Area Network), a public telephone line, or the like. The user terminal 2 may be any terminal that can be connected to the network 100. For example, the user terminal 2 may be a desktop personal computer, a notebook personal computer, a smartphone, or a PDA.

  FIG. 2 is a block diagram of the information processing apparatus 1 in the present embodiment. The information processing apparatus 1 includes a reception unit 101, a related patent identification information storage unit 102, an unrelated patent identification information storage unit 103, a search formula generation unit 104, a patent document storage unit 105, a search patent identification information acquisition unit 106, and a similar element storage. Unit 107, feature vector acquisition unit 108, selection unit 109, evaluation unit 110, and output unit 111.

  The receiving unit 101 receives related patent identification information that is information for specifying one or more related patent documents. The receiving unit 101 may receive unrelated patent identification information that is information for specifying one or more unrelated patent documents. In addition, the reception unit 101 may receive a search expression used to acquire related patent documents and non-related patent documents. The accepting unit 101 may accept the type of patent search. The type of patent search may be information for identifying the type of patent search. Related patent documents are patent documents related to the technology or invention to be searched. Unrelated patent documents are patent documents that are not related to the technology or invention to be searched. The related patent identification information and the unrelated patent identification information received by the receiving unit 101 are information for identifying patent documents that are determined to be related or not related as a result of a screening operation in a manual patent search. However, it may be a result of being automatically selected by the information processing device 1 or other devices. A patent document is information relating to an application document such as a patent filed with the JPO. The types of patent documents include published patent gazette, patent gazette, utility model published gazette, utility model registered gazette, published patent gazette, published utility model gazette, republished patent gazette, republished utility model gazette, and the like. The country of patent documents issuance is not limited to Japan, the United States, China, Europe, South Korea, etc. The language and data format of patent documents are not limited. The information for specifying the patent document may be anything as long as the information can specify one patent document. The information specifying the patent document may be, for example, the patent document itself or a patent document ID that specifies the patent document. The patent document ID may be an application number, a public number, a registration number, or the like, which is an ID managed by a storage device or the like in which the patent document is stored. Also good. A patent search is a search for documents related to a certain technology or invention. For example, there are several types of patent searches depending on the purpose. The types of patent searches are, for example, prior art searches, infringement prevention searches, invalidation material searches, and the like. Prior art search refers to investigating whether prior art exists before filing a patent application. Infringement prevention investigation refers to investigating whether or not to infringe other patents before putting goods or services on the market. The invalidation document search is a search for invalidating the registered patent of another person. In addition, the patent search may be an expression indicating only the operation of selecting related patent documents and non-related patent documents, and is an expression indicating that a search expression is constructed and collecting patent documents to be selected. It may also be an expression that further includes analyzing the survey object.

  The search formula is information used to specify patent documents stored in a database or the like. The search expression includes an element for narrowing down a search target. The search expression may further include a logical operator that relates two or more elements or negates the element, and one or more logical operators and one or two or more elements are combined into one block. It may include a block delimiter element that is information for handling as. Note that the data structure, data format, definition method, and the like of the search expression are not limited.

  The element may be a term, a patent classification, information for narrowing down bibliographic information, or a block element described later. The term is preferably a keyword that specifies a survey target, but may be any term. The term may be associated with a search field. The search field is a target for searching for terms in the search formula. For example, “Summary”, “Claims”, “Summary + Claims”, or “Full text”. The patent classification may be anything that classifies patent documents and can be used when searching. For example, the patent classification may be any of codes such as IPC, FI, F-term, US class, ECLA, and PCP, or may be a part of the code. The part of the code may be, for example, a section, a class, a subclass, or a main group in the case of IPC. Further, a part of the code may be, for example, a theme code or a theme code and a viewpoint as long as it is F-term. Bibliographic information is information necessary for specifying a document. For example, the bibliographic information may be an application number, a public number, a registration number, information on an agent such as an agent name, and an applicant such as an applicant name. May be information on the inventor, such as the inventor name, the filing date, the publication date, or other information. The information for narrowing down the bibliographic information may be the bibliographic information value itself or a value having a range. A value having a range is a value that designates a patent document filed in 2010, such as the filing date “2011.10.1 to 2010.10.1”.

  The logical operator may be, for example, an AND operator, an OR operator, or a NOT operator. The AND operator and the OR operator are logical operators that relate two or more elements. For example, the AND operator is an operator for obtaining a product set of two elements. The OR operator is an operator for obtaining a union of two elements. For example, the NOT operator is an operator that negates an element. For example, the NOT operator is an operator for obtaining a set that does not include a set of specific elements. The block delimiter element may be, for example, parentheses or a text box on a graphical user interface. An area delimited by block delimiter elements is defined as a block element. For example, “(term A OR term B) AND (term B) AND (term C)” in parentheses is a block element, and “(” and “)” are block delimiters. is there.

  In addition, the receiving unit 101 accumulates related patent identification information in the related patent identification information storage unit 102. When receiving unrelated patent identification information, the receiving unit 101 stores the unrelated patent identification information in the unrelated patent identification information storage unit 103.

  The accepting unit 101 usually accepts information transmitted from the user terminal 2 via the network 100, but accepts information input from an input device such as a keyboard, mouse, touch panel, etc., via a wired or wireless communication line. It may be reception of transmitted information or reception of information read from a recording medium such as an optical disk, a magnetic disk, or a semiconductor memory.

  The related patent identification information storage unit 102 stores one or more related patent identification information. The related patent identification information is information for identifying a related patent document that is a patent document that is determined to be related as a result of the sorting operation in the patent search. The related patent identification information storage unit 102 is preferably a non-volatile recording medium, but can also be realized by a volatile recording medium.

  The unrelated patent identification information storage unit 103 stores one or more unrelated patent identification information. The unrelated patent identification information is information for identifying an unrelated patent document that is a patent document that has been determined to be unrelated as a result of the sorting operation in the patent search. The non-related patent identification information storage unit 103 is preferably a nonvolatile recording medium, but can also be realized by a volatile recording medium.

  The search expression generation unit 104 is stored in the related patent identification information storage unit 102 using elements included in the patent document identified by one or more related patent identification information stored in the related patent identification information storage unit 102. A search expression capable of acquiring a patent document identified by one or more related patent identification information is generated. The search expression generation unit 104 acquires all patent documents identified by the related patent identification information stored in the related patent identification information storage unit 102 (hereinafter also referred to as stored related patent documents). However, it is also possible to generate a search expression for obtaining a part of stored related patent documents. For example, the search expression generation unit 104 may acquire elements from the stored related patent documents and connect the acquired elements with an OR operator to generate a search expression.

  Further, the search expression generation unit 104 further uses the elements included in the patent document identified by the one or more unrelated patent identification information stored in the unrelated patent identification information storage unit 103, A search expression that does not acquire at least a part of one or more unrelated patent documents stored in the storage unit 103 may be generated. The search expression generation unit 104 acquires all patent documents (hereinafter, also referred to as stored unrelated patent documents) identified by unrelated patent identification information stored in the unrelated patent identification information storage unit 103. Although it is preferable to generate a search expression that does not, a search expression that does not acquire a part of the stored unrelated patent documents may be generated. For example, the search expression generation unit 104 acquires an element from a stored unrelated patent document, makes an element obtained by connecting each acquired element with an OR operator a block element as a block delimiter element, and applies a NOT operator to the block element The search expression may be generated by connecting the search expression generated using the stored related patent document to a block element with a block delimiter element using an AND operator. When generating a search expression, elements included in stored related patent documents and stored non-related patent documents may not be included in the generated search expression, and block elements to which the NOT operator is applied. It is not necessary to include only. In this way, the search expression generation unit 104 generates a search expression that includes all of the related patent documents and does not include unrelated patent documents as much as possible.

  Note that the search expression generation unit 104 may select only the important terms and include the elements to be included in the search expression in the generated search expression. The important element may be, for example, an element described in a patent document having a ratio equal to or greater than a threshold value, or an element described in the number of times equal to or greater than the threshold value in one patent document. May be an element that is equal to or less than a threshold value, may be an element that has a TF / IDF value equal to or greater than a threshold value, and may be an element that satisfies a combination of two or more conditions among the above conditions. Each threshold value may be a value that is arbitrarily set by the user or may be a value that is empirically set by the developer. The threshold value may be set for each element type. The document for which the IDF value is calculated may be a previous patent document stored in the patent document storage unit 105, or a stored related patent document and a stored unrelated patent document. . In addition, the elements described in the number of patent documents exceeding the threshold may be elements described in stored related patent documents exceeding the threshold, and may be described in stored unrelated patent documents exceeding the threshold. The element may be an element described in both patent documents exceeding the threshold value. The elements described in the stored related patent documents and the stored non-related patent documents are preferably not included in the block elements to which the NOT operator is applied, but are not included in the block elements to which the NOT operator is applied. May be included.

  The search expression generation unit 104 may acquire and include an element similar to the element acquired from the patent document in the element of the search expression. For example, the search expression generation unit 104 may acquire an element similar to the acquired element from the similar element storage unit 107, or may acquire it from the external device that provides the similar element via the network 100. good. Similar elements acquired by the search expression generation unit 104 are the same as similar elements in the similar element storage unit 107 described later. In addition, the search formula generation unit 104 may select so that the number of elements included in the search formula is minimized. Minimizing the number of elements to include in a search formula is preferred when the element is a patent classification, but it may be when the element is a term or even when the element is bibliographic information. It may be a case where the above elements are combined.

  When the accepting unit 101 accepts a search expression and the search field is associated with the accepted search expression term, the search expression generating unit 104 may associate the search field with the generated search expression term. The search field to be generated may be associated with the search field expanded or may not be associated. Expanding the search field means increasing the number of patent documents to be searched. Specifically, expanding the search field may mean that the search field is “full text” when the search field associated with the accepted search term is “summary + claim”. good. Further, when the receiving unit 101 receives a type of patent search, the search expression generation unit 104 may associate a search field that matches the type of patent search with a term, and search for bibliographic information that matches the type of patent search. It may be included in the search expression. For example, when the type of patent search is invalidation material search, the search formula generation unit 104 searches for patent documents whose search field “full text” is associated with the term and whose application date is from the current date to 20 years ago. A search expression for searching may be generated. When the accepting unit 101 accepts a search formula and has bibliographic information, the search formula generating unit 104 may include the bibliographic information in the search formula for generating the bibliographic information. Does not have to be included. Note that the search expression generation unit 104 may generate a search expression by associating a search field set in advance with a search expression term to be generated. Further, the search formula generation unit 104 may generate a search formula including bibliographic information set in advance. Setting in advance may be arbitrarily set by the user or may be set by the developer based on experience. For a method for generating a search expression such as a method for minimizing the number of elements included in the search expression, see Japanese Patent Application Laid-Open No. 2011-59843 for a method that is not described in detail in this description. The search expression generation unit 104 can be usually realized by an MPU, a memory, or the like. The processing procedure of the search expression generation unit 104 is usually realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The patent document storage unit 105 stores one or more patent documents. The patent documents stored in the patent document storage unit 105 may be information including at least a part of the patent documents, or may be all. The patent document storage unit 105 stores the patent document so that the patent document can be specified by the information specifying the patent document. The patent document may be stored in the patent document storage unit 105 so as to be able to be specified from the information specifying the patent document by the information included in the patent document. The patent document is associated with the information specifying the patent document, It may be stored so that it can be identified from information identifying the patent document. The patent document storage unit 105 is preferably a non-volatile recording medium, but can also be realized by a volatile recording medium. The process of storing patent documents in the patent document storage unit 105 does not matter. For example, the patent document storage unit 105 may store a patent document via a recording medium, or may store a patent document transmitted via a communication line or the like, or The patent document input via the input device may be stored.

  The search patent identification information acquisition unit 106 acquires search patent identification information that is information for specifying a search patent document that is a patent document acquired using the search formula generated by the search formula generation unit 104. That is, the search patent identification information acquisition unit 106 searches the search formula generated by the search formula generation unit 104, specifies a hit patent document, and acquires search patent identification information that specifies the patent document. The search patent identification information acquisition unit 106 may acquire the search patent identification information from the patent document storage unit 105, or may acquire the search patent identification information from an external device that provides information specifying the patent document. good. As a result, the searched patent identification information overlaps with related patent identification information stored in the related patent identification information storage unit 102 or unrelated patent identification information stored in the unrelated patent identification information storage unit 103. It does not have to be duplicated. That is, the search formula generated by the search formula generation unit 104 may be a search formula for searching stored unrelated patent documents. The search patent identification information acquisition unit 106 can usually be realized by an MPU, a memory, or the like. The processing procedure of the search patent identification information acquisition unit 106 is usually realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The similar element storage unit 107 stores one or more similar elements that are similar elements. A similar element may be a synonym, a word having a relationship between a broader term and a narrower term, a patent category indicating the same technical field, or a transition or integration in the transition of a patent category. It may be a patent classification or the like. The similar element storage unit 107 preferably stores a heading element and an element similar to the heading element in association with each other, but one or more similar elements have a one-to-one correspondence. It may be attached. The similar element storage unit 107 is preferably a non-volatile recording medium, but can also be realized by a volatile recording medium. The process of storing similar elements in the similar element storage unit 107 does not matter. For example, the similar element storage unit 107 may store similar elements via a recording medium, may store similar elements transmitted via a communication line, or the like. Similar elements input via the input device may be stored.

  The feature vector acquisition unit 108 acquires and searches one or more related patent feature vectors that are feature vectors of related patent documents identified by each related patent identification information stored in the related patent identification information storage unit 102. One or more search patent feature vectors, which are feature vectors of a search patent document identified by each search patent identification information acquired by the patent identification information acquisition unit 106, are acquired. The feature vector acquisition unit 108 also obtains one or more unrelated patent feature vectors that are feature vectors of unrelated patent documents identified by each unrelated patent identification information stored in the unrelated patent identification information storage unit 103. Furthermore, you may acquire. Note that the feature vector acquisition unit 108 may acquire a feature vector using information specifying a patent document from a feature vector storage unit (not shown), or use information specifying a patent document from an external device (not shown). The feature vector may be acquired, or the feature vector may be generated from a patent document. When acquiring a feature vector from an external device (not shown), the external device may generate a feature vector in response to a request from the feature vector acquisition unit 108, and stores a previously generated feature vector. May be. When creating and obtaining feature vectors from patent documents, the feature vector obtaining unit 108 obtains each patent document identified by information for identifying each patent document, and obtains one or more elements from each patent document. Each feature vector is acquired using the one or more elements. The element may be a term, a patent classification, or bibliographic information.

  When the element is a term, the term acquired from each patent document by the feature vector acquisition unit 108 may be, for example, a predetermined part of speech (for example, a noun or the like), and may be a series of predetermined parts of speech. There may be technical terms or other terms. If the term is a technical term, for example, the technical term is stored in a recording medium (not shown), and the term can be acquired by determining whether the technical term is included in the related patent document. The term may be acquired by using an algorithm for extracting technical terms from a document. For the latter method, see, for example, the following document. Hirokazu Ohata, Hiroshi Nakagawa, “Extraction of technical terms by different number of words”, Information Processing Society of Japan Research Report, 2000-NL-136, p. 119-126. Nakagawa Hiroshi, Mori Yasunori, Yumoto Yasuaki, “Terminology Extraction Based on Appearance Frequency and Connection Frequency”, Natural Language Processing, Vol. 10 No. 1, p. 27-45, January 2003. When acquiring the term, the feature vector acquisition unit 108 may acquire the term by using the TF value or the TF / IDF value as the importance and excluding the term having a low importance. The document for which the IDF value is calculated may be all the patent documents stored in the patent document storage unit 105. The stored related patent documents, the stored unrelated patent documents, and the search patent identification information may be used. It may be a patent document to be identified (hereinafter also referred to as a search patent document). In addition, when an unrelated patent feature vector is not acquired, the stored unrelated patent document may not be a target for calculating the IDF value. The term of low importance may be a term whose importance value is a threshold value or less, or may be a predetermined number of terms arranged in order of importance. The threshold may be a predetermined value, for example, or may be a value obtained by multiplying the maximum value of importance by a number smaller than 1 (for example, 0.9 or 0.8). The predetermined number may be a predetermined number, for example, and a value smaller than 1 (for example, 0.01 or 0.001) is added to the total number of terms acquired by the feature vector acquisition unit 108. It may be a multiplied number. The feature vector acquisition unit 108 acquires terms from a character string including at least a part of a patent document. The character string including at least a part of a patent document may be, for example, the full text described in the patent document, a character string identified by a heading, or a combination of a plurality of headings. It may be a character string. The heading may be anything as long as it identifies an area in a patent document such as, for example, the name of an invention, a claim, or a summary.

  The feature vector acquisition unit 108 may acquire each feature vector obtained from each patent document with similar elements as the same vector element. When similar elements are set to the same vector element, the feature vector acquisition unit 108 may use the information stored in the similar element storage unit 107 to change the similar elements into the same vector element, and is not illustrated. A similar element may be acquired via a network from an external device that provides the element, and the similar element may be made the same vector element using the information. Specifically, the feature vector acquisition unit 108, when “database, database, DB, repository, dictionary” is stored in the similar element storage unit 107, “database”, “database”, and “DB” ”,“ Repository ”, and“ dictionary ”are all processed as the same term.

  Each feature vector acquired by the feature vector acquisition unit 108 may be a vector having dimensions for all the elements acquired by the feature vector acquisition unit 108, and the number of dimensions may be different for each patent document. . The element of the feature vector may be, for example, the number of occurrences of the term, the TF value of the term, or the TF / IDF value of the term, and whether the term exists in the patent document. It may be a number indicating whether or not, specifically “1 (present)”, “−1 (not present)”, or the like.

  The feature vector acquisition unit 108 may acquire a feature vector from the entire patent document (full text), or may acquire a feature vector from a part of the patent document. The part of the patent document may be, for example, an international patent classification, abstract, claims, or a combination thereof.

  Note that the feature vector acquisition unit 108 may perform a process of acquiring the same feature vector more than once when calculating the feature vector of the same patent document, and accumulates the calculated feature vector in a storage unit (not shown). By doing so, it may be obtained without performing the same processing. Specifically, the feature vector acquisition unit 108 may perform a process of acquiring a feature vector only once when a stored related patent document and a stored unrelated patent document and a search patent document overlap. You may perform the process which acquires a degree feature vector. The feature vector acquisition unit 108 can be usually realized by an MPU, a memory, or the like. The processing procedure of the feature vector acquisition unit 108 is usually realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The selection unit 109 is a search patent similar to the feature of one or more related patent feature vectors acquired by the feature vector acquisition unit 108 among one or more search patent feature vectors acquired by the feature vector acquisition unit 108. At least search patent identification information that corresponds to the feature vector and does not match the related patent identification information stored in the related patent identification information storage unit 102 is selected. In other words, the selection unit 109 selects information that identifies a patent document that has a possibility of a related patent from the search patent identification information that has not been selected as a related patent document in the screening work in the patent search. The selection unit 109 is similar to the feature of one or more related patent feature vectors acquired by the feature vector acquisition unit 108 among one or more search patent feature vectors acquired by the feature vector acquisition unit 108. The search patent identification information corresponding to the patent document identified by the related patent identification information stored in the related patent identification information storage unit 102 and the search patent identification information corresponding to the search patent feature vector is also selected. good. That is, the selection unit 109 may select information for identifying a patent document correctly classified as a related patent document from the stored related patent documents. The selection unit 109 is similar to the feature of one or more unrelated patent feature vectors acquired by the feature vector acquisition unit 108 among one or more search patent feature vectors acquired by the feature vector acquisition unit 108. Search patent identification information corresponding to the search patent feature vector not to be selected may be selected. That is, the selection unit 109 does not have to select information for specifying a search patent document having characteristics of an unrelated patent document. The selection unit 109 is similar to the feature of one or more related patent feature vectors acquired by the feature vector acquisition unit 108 among one or more unrelated patent feature vectors acquired by the feature vector acquisition unit 108. Then, the unrelated patent identification information corresponding to the unrelated patent feature vector that is not similar to the feature of the one or more unrelated patent feature vectors acquired by the feature vector acquisition unit 108 is also selected. That is, the selection unit 109 may select a patent document having a high possibility of a related patent document among the patent documents that are determined to be unrelated patent documents in the sorting operation in the patent search. It may be considered that one or two or more feature vectors are included in a class that is similar to a feature of a feature vector. Hereinafter, explanation will be made using classes. The selection unit 109 selects information specifying each patent document, but may select each patent document.

  A class is a set of one or more feature vectors calculated. Classes do not contain different types of feature vectors. The phrase “not including different types of feature vectors” means, for example, that a class of related patent feature vectors does not include a feature vector that is not a related patent feature vector. If there is no unrelated patent identification information storage unit 103, the selection unit 109 may determine that the complement of the class of related patent feature vectors is a set of unrelated patent feature vectors. Further, a complementary set of the union of the related patent feature vector class and the unrelated patent feature vector class may be determined as a patent document that is neither a related patent document nor an unrelated patent document. The class may be a set that is a result of classification by a learning device using machine learning described later.

  There is no limitation on the method in which the selection unit 109 selects a patent document using the feature vector. For example, the selection unit 109 may select using vector similarity or may use machine learning. Hereinafter, the selection unit 109 will be described in two parts: (A) a method of selecting using vector similarity, and (B) a method of selecting using machine learning. Note that the selection unit 109 may classify each patent feature vector into two or more classes.

(A) Selection using feature vector similarity The selection using feature vector similarity means that the similarity between a patent feature vector to be judged and a representative vector of a class of related patent feature vectors is equal to or less than a threshold value. In some cases, it may be determined that the patent feature vector to be judged is a related patent, and the similarity between the patent feature vector to be judged and the representative vector of the class of the related patent feature vector is If the degree of similarity of the representative vector of the class of the unrelated patent feature vector is smaller, the patent feature vector to be judged may be determined to be a related patent and may be selected. The similarity between the representative vector of the patent feature vector class and the representative vector is less than or equal to the threshold, and is smaller than the similarity of the representative vector of the class of the unrelated patent feature vector Sometimes, the patent feature vector to be determined may be selected based on the related patent. The similarity calculation method may use a COS scale, may use a Pearson correlation coefficient, or may use another calculation method for calculating a similarity between vectors such as a deviation pattern similarity. good. Since the calculation method for calculating each similarity is a known technique, a description thereof will be omitted. The class of the related patent feature vector may be 1 or 2 or more. When creating two or more classes, for example, feature vectors may be clustered using division optimization clustering (such as k-means method), or feature vectors may be clustered using hierarchical clustering (such as shortest distance method). Alternatively, the feature vectors may be clustered using other known clustering methods. Note that details of each clustering method are well-known techniques, and thus description thereof is omitted. The same applies to the class of unrelated patent feature vectors. The representative vector may be a feature vector that appears most frequently in the class, or may be an average vector of the class. The average vector may be a normal vector average, or a unit vector average calculated by treating all as unit vectors. Further, when determining the representative vector, the selection unit 109 calculates the representative vector once, selects a representative vector that is calculated again by excluding a feature vector whose similarity with the representative vector is higher than a threshold value from the class. May be used for The threshold used at this time is one or more values (for example, 1.5 or 2.0) in the average value of the similarity between the representative vector created first and the feature vector used to create the representative vector. Or a value arbitrarily set by the user or the developer. Note that the threshold for determining whether or not they are similar may be a predetermined value or a value calculated by the selection unit 109. When the threshold value is a predetermined value, the threshold value may be a value arbitrarily set by a user or a developer. When the threshold value is a value calculated by the selection unit 109, it may be the similarity between the representative vector of each class and the least similar feature vector belonging to the class, and the representative vector of each class and the class It may be the average value of the similarity with the least similar feature vector belonging to it, or it may be the minimum value of the similarity between the representative vector of each class and the least similar feature vector belonging to that class. May be the maximum value of the similarity between the representative vector and the least similar feature vector belonging to the class.

(B) Selection using machine learning Selection using machine learning means learning using a learning device such as a neural network, SVM (Support Vector Machine), or SVR (Support Vector Regression), and learning is completed. On the other hand, the patent feature vector to be judged is classified to determine whether it belongs to the class of the related patent feature vector and to select it. A neural network is a learning model that aims to express some characteristics of brain function by computer simulation. There are various types of models and methods in the neural network, any of which may be adopted. For example, as the type of neural network, a perceptron may be employed, back-property may be employed, or a Boltzmann machine may be employed. SVM is a learning model that learns a classification pattern using teacher data, sets boundary lines for classification, and performs classification. SVR is a learning model that learns classification patterns using teacher data and classifies them into three or more classes. Since each machine learning algorithm is a known technique, description thereof is omitted. When the selection unit 109 selects using machine learning, the feature vector of the patent document stored in the related patent identification information storage unit 102 and the unrelated patent identification information storage unit 103 is learned as teacher data. The feature used by the selection unit 109 for learning is a value corresponding to each element acquired by the feature vector acquisition unit 108. After the learning is completed, when the patent feature vector to be determined is determined to be the related patent feature vector, the selection unit 109 is similar to the feature included in the related patent feature vector. And / or judging that the feature is not similar to the feature of the unrelated patent feature vector, and selecting the patent feature vector to be judged.

  Note that when the number of dimensions of each feature vector is different, the selection unit 109 unifies the number of dimensions. The selection unit 109 can be usually realized by an MPU, a memory, or the like. The processing procedure of the selection unit 109 is usually realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The evaluation unit 110 performs a lower evaluation as more search patent identification information that does not match the related patent identification information stored in the related patent identification information storage unit 102 among the search patent identification information selected by the selection unit 109 increases. The evaluation may be to acquire an evaluation value. That is, the evaluation unit 110 may acquire a lower evaluation value as there are more search patent documents that have not been selected in the selection work in the patent search. In addition, when the selection unit 109 selects the related patent document that has already been stored, the evaluation unit 110 is the search patent identification information acquired by the search patent identification information acquisition unit 106 and is selected by the selection unit 109. As the related patent identification information stored in the related patent identification information storage unit 102 does not match the retrieved patent identification information, the lower the evaluation, the lower the evaluation may be. That is, the evaluation unit 110 may acquire a lower evaluation value as there are more unrelated patent documents that are erroneously determined to be related patent documents in the screening work in the patent search. When the selection unit 109 also selects stored non-related patent documents, the evaluation unit 110 may perform a lower evaluation as the number of non-related patent identification information selected by the selection unit 109 increases. That is, the evaluation unit 110 may acquire a lower evaluation value as there are more related patent documents that are erroneously determined to be unrelated patent documents in the selection work in the patent search. Note that the evaluation value may be acquired by calculating an evaluation value calculated using the result selected by the selection unit 109, and corresponds to the result selected by the selection unit 109 using an evaluation table. You may acquire the evaluated value. The evaluation table may be a table in which the selection result of the selection unit 109 is associated with the value to be deducted, or may be a table in which the selection result of the selection unit 109 is associated with the value to be added. good. The selection result of the selection unit 109 may be the number of search patent identification information selected by the selection unit 109, or the number of information specifying the stored related patent documents that the selection unit 109 did not select. It may be the number of information specifying the stored non-related patent documents selected by the selection unit 109.

  The evaluation unit 110 may perform evaluation using the precision, the recall, or the F value. The relevance rate may be, for example, “(related patent document selected in sorting operation in patent search + related patent document selected in selection unit 109) / search patent document” or “(selected in sorting operation in patent search). Related patent document + related patent document selected only by the selection unit 109) / (patent document to be selected in the search operation in the patent search not including the search patent document + search patent document) ”. In addition, the recall rate is “related patent document selected in the screening work in the patent search / (related patent document selected in the screening work in the patent search + related patent document selected only in the selection unit 109)”. Also good. The F value may be (2 × matching rate × recall rate) / (matching rate + recall rate). Note that the evaluation unit 110 may output one value as an evaluation value, or may output a plurality of values as evaluation values.

  The evaluation unit 110 can usually be realized by an MPU, a memory, or the like. The processing procedure of the evaluation unit 110 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The output unit 111 outputs information regarding the result selected by the selection unit 109. The information related to the result selected by the selection unit 109 may be information obtained by the evaluation unit 110 evaluating the result selected by the selection unit 109. Output refers to display on a display, projection using a projector, printing on a printer, transmission to an external device, storage in a recording medium, delivery of processing results to other processing devices or other programs, etc. It is a concept that includes. The output unit 111 may be considered as including or not including an output device such as a display or a speaker. The output unit 111 can be realized by output device driver software, or output device driver software and an output device.

  FIG. 2 is a flowchart showing an example of the operation of the information processing apparatus 1 in the present embodiment. Hereinafter, the operation will be described with reference to FIG.

  (Step S201) The receiving unit 101 determines whether or not related patent identification information and unrelated patent identification information, which are patent search results, have been received. If a patent search result is accepted, the process proceeds to step S202. If not received, step S201 is repeated.

  (Step S202) The receiving unit 101 acquires related patent identification information from the received patent search result, and accumulates the acquired related patent identification information in the related patent identification information storage unit 102.

  (Step S <b> 203) The receiving unit 101 acquires unrelated patent identification information from the received patent search result, and accumulates the acquired unrelated patent identification information in the unrelated patent identification information storage unit 103.

  (Step S204) The search expression generation unit 104 generates a search expression using related patent documents and unrelated patent documents. Details of the method for generating the search expression will be described later with reference to the flowchart of FIG.

  (Step S205) The search patent identification information acquisition unit 106 acquires search patent identification information using the search formula created in step S204.

  (Step S206) The feature vector acquisition unit 108 acquires a related patent feature vector using the related patent identification information stored in the related patent identification information storage unit 102. Details of the method for acquiring the related patent feature vector will be described later with reference to the flowchart of FIG.

  (Step S207) The feature vector acquisition unit 108 acquires an unrelated patent feature vector using the unrelated patent identification information stored in the unrelated patent identification information storage unit 103. Details of the method for acquiring the unrelated patent feature vector will be described later with reference to the flowchart of FIG.

  (Step S208) The feature vector acquisition unit 108 acquires a search patent feature vector using the search patent identification information acquired in step S205. Details of the method for acquiring the search patent feature vector will be described later with reference to the flowchart of FIG.

  (Step S209) The selection unit 109 unifies the dimensions of the feature vectors.

  (Step S210) The selection unit 109 acquires an average vector of related patent feature vectors.

  (Step S211) The selection unit 109 acquires an average vector of unrelated patent feature vectors.

  (Step S212) The selection unit 109 selects an original related patent from the search patent documents acquired in Step S205. Details of the method of selecting the original related patent will be described later using the flowchart of FIG.

  (Step S213) The selection unit 109 selects the original related patent from the related patent identification information stored in the related patent identification information storage unit 102 in advance. Details of the method of selecting the original related patent will be described later using the flowchart of FIG.

  (Step S214) The selection unit 109 selects the original related patent from the unrelated patent identification information stored in the unrelated patent identification information storage unit 103 in advance. Details of the method of selecting the original related patent will be described later using the flowchart of FIG.

  (Step S215) The evaluation unit 110 evaluates whether the selection of the patent search result received in step S201 is appropriate using the selection result of step S212 to step S214.

  (Step S216) The output unit 111 outputs the result evaluated in step S215.

  FIG. 4 is a flowchart showing an example of the operation of generating the search expression (step S204) in FIG. Hereinafter, a process for generating a search expression will be described with reference to FIG.

  (Step S <b> 301) The search expression generation unit 104 acquires related patent documents from the patent document storage unit 105 using the related patent identification information stored in the related patent identification information storage unit 102.

  (Step S <b> 302) The search expression generation unit 104 acquires an unrelated patent document from the patent document storage unit 105 using the unrelated patent identification information stored in the unrelated patent identification information storage unit 103.

  (Step S303) The search expression generation unit 104 assigns 1 to the counter m.

  (Step S304) The search expression generation unit 104 determines whether the related patent document acquired in Step S301 includes the mth related patent document. If there is an m-th related patent document, the process proceeds to step S305, and if not, the process proceeds to step S307.

  (Step S305) The search expression generation unit 104 acquires an element from the m-th related patent document.

  (Step S306) The search expression generation unit 104 increments the counter m by 1. Then, the process returns to step S304.

  (Step S307) The search expression generation unit 104 substitutes 1 for a counter n.

  (Step S308) The search expression generation unit 104 determines whether or not the unrelated patent document acquired in step S302 includes the nth unrelated patent document. If there is an nth unrelated patent document, the process proceeds to step S309, and if not, the process proceeds to step S311.

  (Step S309) The search expression generation unit 104 acquires an element from the nth unrelated patent document.

  (Step S310) The search expression generation unit 104 increments the counter n by 1. Then, the process returns to step S308.

  (Step S311) The search expression generation unit 104 acquires a patent document including the related patent document element acquired in step S305, and acquires a search expression that does not acquire the patent document including the unrelated patent document element acquired in step S309. Generate. Then, the process returns to the upper process.

  FIG. 5 is a flowchart showing an example of the operation of acquiring the feature vector (step S206, step S207, step S208) of FIG. Hereinafter, a process for generating a search expression will be described with reference to FIG. Note that “related patent” is assigned to X in FIG. 5 when called from step S206, and “unrelated patent” is assigned when called from step S207, and called from step S208. In this case, “search patent” is substituted. The following description is based on the assumption that it is called from step S206 in which “related patent” is assigned to X. Step S207 and step S208 are to be read respectively.

  (Step S <b> 401) The feature vector acquisition unit 108 acquires related patent documents from the patent document storage unit 105 using the related patent identification information stored in the related patent identification information storage unit 102.

  (Step S402) The feature vector acquisition unit 108 substitutes 1 for the counter p.

  (Step S403) The feature vector acquisition unit 108 determines whether the related patent document acquired in step S401 includes the p-th related patent document. If there is a p-th related patent document, the process proceeds to step S404. If not, the process returns to the upper process.

  (Step S404) The feature vector acquisition unit 108 acquires elements from the p-th related patent document.

  (Step S405) The feature vector acquisition unit 108 converts similar elements into unified elements using the similar elements stored in the similar element storage unit 107.

  (Step S406) The feature vector acquisition unit 108 acquires the TF / IDF value of the element acquired in step S405.

  (Step S407) The feature vector acquisition unit 108 acquires the p-th related patent feature vector having the number of dimensions acquired in step S405 and the value of each dimension being the TF / IDF value acquired in step S406.

  (Step S408) The feature vector acquisition unit 108 increments the counter m by 1. Then, the process returns to step S403.

  FIG. 6 is a flowchart showing an example of the operation of selecting the related patent in FIG. 3 (Step S212, Step S213, Step S214). Hereinafter, the process of selecting a related patent will be described with reference to FIG. 5 is substituted with “search patent” when called from step S212, and is called with “patents selected in advance as related patents” when called from step S213. When called from S214, “patents selected in advance as unrelated patents” shall be substituted. The following description is based on the assumption that it is called from step S212 in which “search patent” is assigned to Y. Step S213 and step S214 are each read as different.

  (Step S501) The selection unit 109 substitutes 1 for a counter q.

  (Step S502) The selection unit 109 determines whether there is a q-th search patent document. If there is a q-th related patent document, the process proceeds to step S503, and if not, the process returns to the upper process.

  (Step S503) The selection unit 109 calculates the similarity between the feature vector of the qth search patent and the average vector of the related patent feature vectors.

  (Step S504) The selection unit 109 calculates the similarity between the feature vector of the qth search patent and the average vector of unrelated patent feature vectors.

  (Step S505) The selection unit 109 determines whether the similarity with the average vector of the related patent feature vector acquired in step S503 is higher than the similarity with the average vector of the unrelated patent feature vector acquired in step S504. Judging. If it is higher, the process proceeds to step S506, and if it is lower, the process returns to step S502.

  (Step S506) The selection unit 109 determines whether the similarity with the average vector of the related patent feature vectors acquired in step S503 is less than a threshold value. If it is equal to or smaller than the threshold value, the process proceeds to step S507, and if it is equal to or larger than the threshold value, the process returns to step S502.

  (Step S507) The selection unit 109 accumulates information for specifying the qth search patent in a storage unit (not shown). Then, the process returns to step S502.

  Hereinafter, a specific operation of the information processing apparatus 1 in the present embodiment will be described. In addition, the information of each drawing shown in this specific example is prepared for convenience of explanation, and does not indicate actual data. Further, in this specific example, it is assumed that “patent search result.file” received by the receiving unit 101 is a file including related patent identification information and unrelated patent identification information. In this specific example, the information specifying the patent document is a public number. In this specific example, it is assumed that the search formula created by the search formula generation unit 104 is a search formula for only technical terms.

  In this specific example, it is assumed that the patent documents stored in the patent document storage unit 105 are those shown in FIG. The table in FIG. 7 includes information specifying patent documents and patent documents. The patent document has the title of the invention, international patent classification, abstract, claims, background art, and the like. For example, in the record of the patent document ID “1”, the name of the invention “search expression generation device,... ((Hereinafter omitted))”, the international patent classification “G06F 17/30”, and the publication number “JP2011-2011”. According to the patent search support apparatus according to the present invention, AAAAAA ", summary" GPS position information ... (hereinafter omitted) ", claim" Show first patent publication ... " ... (hereinafter omitted) "and other information are registered in association with each other.

  In this specific example, the similar elements stored in the similar element storage unit 107 are assumed to be those shown in FIG. The table in FIG. 8 has a unified element and similar elements unified by the unified element. For example, a similar element “GPS, global positioning,... (Hereinafter omitted)” is registered in association with the record of the unified element “world positioning system”.

  It is assumed that the user of the user terminal 2 operates a pointing device, a keyboard, etc., selects a file in which a patent search result is described, and then clicks an “execute” button. Then, the accepting unit 101 accepts “patent search result.file” via the network 100 (step S201). The receiving unit 101 acquires related patent identification information and unrelated patent identification information from “patent search result.file”, and related patent identification information is stored in the related patent identification information storage unit 102 and unrelated patent identification information is unrelated. The information is stored in the patent identification information storage unit 103 (steps S202 and S203). As a result, the related patent identification information is stored in the related patent identification information storage unit 102 as shown in FIG. Further, as shown in FIG. 9B, the unrelated patent identification information is similarly stored in the unrelated patent identification information storage unit 103.

  When the information specifying the patent document is stored in the related patent identification information storage unit 102 and the unrelated patent identification information storage unit 103, the search formula generation unit 104 generates a search formula (step S204). The search expression generation unit 104 uses the related patent identification information “Japanese Patent Laid-Open No. 2011-AAAAAA” stored in the related patent identification information storage unit 102 to retrieve the patent document with the patent document ID “1” from the patent document storage unit 105. Is acquired (step S301). The search expression generation unit 104 similarly acquires related patent documents using all the related patent identification information stored in the related patent identification information storage unit 102. Next, the search expression generation unit 104 uses the unrelated patent identification information “JP 2012-DDDDDD” stored in the unrelated patent identification information storage unit 103, from the patent document storage unit 105 to the patent document ID “ 4 "is acquired (step S302). The search expression generation unit 104 similarly acquires unrelated patent documents using all the unrelated patent identification information stored in the unrelated patent identification information storage unit 103. Then, the search expression generation unit 104 acquires the technical terms “GPS”, “location information”, “patent bulletin”, etc. described in the patent document with the patent document ID “1” which is the first related patent document. (Steps S303 to S306). The search expression generation unit 104 acquires technical terms from all related patent documents acquired in the same manner. Further, the search expression generation unit 104 uses the technical terms “patent search”, “search object”, “patent bulletin”, etc. described in the patent document with the patent document ID “4” which is the first unrelated patent document. Is acquired (steps S307 to S310). The search expression generation unit 104 acquires technical terms from all unrelated patent documents acquired in the same manner. Finally, the search expression generation unit 104 searches all related patent documents using the acquired technical terms and does not search unrelated patent documents “(GPS OR position information OR patent publication OR...) AND ( NOT (patent search OR search target OR...)) ”Is generated (step S311). When the generation of the search formula is completed, the search formula generation unit 104 notifies the search patent identification information acquisition unit 106 of the generated search formula.

  When the search patent identification information acquisition unit 106 acquires the search formula from the search formula generation unit 104, the search patent identification information acquisition unit 106 searches the patent document storage unit 105. Then, the search patent identification information acquisition unit 106 acquires search patent identification information “JP 2012-GGGGGG”, “JP 2011-HHHHHH”, etc., which is information for specifying a patent document searched by the search formula ( Step S205).

  The feature vector acquisition unit 108 creates a feature vector from the stored related patent documents (step S206). That is, the feature vector acquisition unit 108 uses the related patent identification information “Japanese Patent Application Laid-Open No. 2011-AAAAAA” stored in the related patent identification information storage unit 102, and stores the patent document ID “1” from the patent document storage unit 105. A patent document is acquired (step S401). The feature vector acquisition unit 108 similarly acquires related patent documents using all the related patent identification information stored in the related patent identification information storage unit 102. Next, the feature vector acquisition unit 108 displays the technical terms “GPS”, “position information”, “patent bulletin”, etc. described in the patent document with the patent document ID “1” which is the first related patent document. Obtain (steps S402 and S403). Then, the feature vector acquisition unit 108 replaces “GPS” with the “world positioning system” as a unified element (step S405). Further, the feature vector acquisition unit 108 sets the TF / IDF for each technical term acquired from the patent document with the patent document ID “1”, which is the first related patent document, such as “World positioning system: 0.0272”. Values are calculated (step S406), and related patent feature vectors having dimensions corresponding to the number of technical terms and having the calculated TF / IDF values as values are acquired (steps S407 and S408). And the feature vector acquisition part 108 acquires a related patent feature vector with respect to all the related patent documents similarly. When the feature vector acquisition unit 108 acquires the related patent feature vector, the feature vector acquisition unit 108 acquires the unrelated patent feature vector and the search patent feature vector by the same method (steps S207 and S208). As a result, a feature vector as shown in FIG. 10 is acquired. When the feature vector acquisition unit 108 has acquired all the vectors, the feature vector acquisition unit 108 passes the acquired feature vector to the selection unit 109.

  Upon receiving the feature vector from the feature vector acquisition unit 108, the selection unit 109 unifies the number of dimensions so that all the feature vectors become vectors having the same number of dimensions (step S209). Next, the selection unit 109 acquires an average vector of related patent feature vectors and an average vector of unrelated patent feature vectors (steps S210 and S211). At this time, the selection unit 109 has twice the similarity between the related patent feature vector and the average vector of the information “Japanese Patent Laid-Open No. 2011-AAAAAA” specifying the patent document, and the average of the similarities with other related patent feature vectors. It is assumed that the average vector of the related patent feature vectors excluding the related patent feature vector of the information “JP-A 2011-AAAAAA” specifying the patent document is acquired as higher. Similarly, when acquiring an average vector of unrelated patent feature vectors, the selection unit 109 similarly uses the average vector of unrelated patent feature vectors excluding information “JP-2011-EEEEEE” and the like for specifying patent documents. Is obtained. Then, the selection unit 109 selects a related patent among the search patents (step S212). In other words, the selection unit 109 calculates the COS scale between the search patent feature vector corresponding to the information “JP 2012-GGGGGG” specifying the patent document that is the first search patent document and the average vector of the related patent feature vectors. Calculate (steps S501 to S503). Then, the selection unit 109 calculates a COS measure between the search patent feature vector corresponding to the information “JP 2012-GGGGGG” specifying the patent document and the average vector of the unrelated patent feature vectors (step S504). At this time, the search patent feature vector corresponding to the information “JP 2012-GGGGGG” specifying the patent document has a higher similarity to the related patent feature vector and the highest similarity from the average vector of the related patent feature vectors. It is assumed that the degree of similarity is higher than that of a related patent feature vector having a lower Therefore, the selection unit 109 determines that the information “JP 2012-GGGGGG” specifying the patent document is a related patent and selects the information (steps S505 to S506). The selection unit 109 accumulates information for specifying the selected patent document in a storage unit (not shown). The selection unit 109 performs the same processing for the remaining search patent feature vectors, related patent feature vectors, and unrelated patent feature vectors, and determines how many items other than the information “JP 2012-GGGGGG” specifying the patent document. Search patent identification information, all related patent identification information except related patent identification information “Japanese Unexamined Patent Publication No. 2011-AAAAAA”, and unrelated patent identification information “Japanese Unexamined Patent Application Publication No. 2011-EEEEEE” are selected (step S213). , S214). The selection unit 109 notifies the evaluation unit 110 that selection of related patents has been completed. Note that FIG. 11 is two-dimensionally mapped in order to make it easy to understand the relationship between the patent documents when the selection unit 109 selects. In FIG. 11, information “JP 2012-GGGGGG” for specifying a patent document and unrelated patent identification information “JP 2011-EEEEEE” are included in a range similar to the stored related patent. The information “Japanese Unexamined Patent Application Publication No. 2011-AAAAAA” is included in a range similar to an unrelated patent. In FIG. 11, the information specifying the stored related patent document and the information specifying the search patent document are shown by a circle surrounded by a square.

  When the selection by the selection unit 109 is completed, the evaluation unit 110 calculates a relevance ratio of 0.85 and a recall ratio of 0.94 using information for specifying patent documents stored in a storage unit (not shown), and evaluates Assume that an F value of 0.89 is calculated (step S215). The evaluation unit 110 notifies the output unit 111 of the evaluation result.

  The output unit 111 outputs the evaluation result received from the evaluation unit 110 and the information specifying the patent document stored in the storage unit (not shown) (step S216). Then, information is displayed below the result of FIG.

  As described above, according to the present embodiment, a patent document that can be a related patent can be automatically selected by using the patent document as the related patent document selected in the patent search. Further, according to the present embodiment, it is possible to automatically select a patent document with a possibility of a related patent by using a related patent document selected in a patent search and an unrelated patent. Further, according to the present embodiment, it is possible to select a patent document having a high possibility of a similar related patent by creating a search formula from the related patent document. Further, according to the present embodiment, a search expression can be created from related patent documents and non-related patent documents to select patent documents having a higher possibility of similar related patents. In addition, according to the present embodiment, it is possible to evaluate the screening work of the patent search performed manually.

  Moreover, although the case where the evaluation unit 110 is provided has been described in the present embodiment, the information processing apparatus 1 may not include the evaluation unit 110. When the evaluation unit 110 is not provided, the output unit 111 may output only the result of the selection unit 109 without performing the evaluation.

  Further, although the case where the similar element storage unit 107 is provided has been described in the present embodiment, the information processing apparatus 1 may not include the similar element storage unit 107. When the similar element storage unit 107 is not provided, the search formula generation unit 104 and the feature vector acquisition unit 108 acquire and use similar elements from an external device that provides similar elements via the network 100. Alternatively, similar elements may not be used.

  Moreover, although the case where the patent document storage unit 105 is provided has been described in the present embodiment, the information processing apparatus 1 may not include the patent document storage unit 105. When the patent document storage unit 105 is not provided, the search expression generation unit 104, the search patent identification information acquisition unit 106, and the feature vector acquisition unit 108 are connected via an external device that provides patent documents via the network 100. Patent documents may be obtained and used.

  In the present embodiment, the case where the search expression generation unit 104 and the search patent identification information acquisition unit 106 are provided has been described. However, the information processing apparatus 1 includes the search expression generation unit 104, the search patent identification information acquisition unit 106, and the like. It is not necessary to have. When the search expression generation unit 104 and the search patent identification information acquisition unit 106 are not provided, the feature vector acquisition unit 108 may acquire feature vectors from all existing patent documents, and the reception unit 101 may The feature vector may be acquired from the patent document received via the system 100.

  Moreover, the software which implement | achieves the information processing apparatus 1 in this Embodiment is the following programs. In other words, the program can access a related patent identification information storage unit that stores one or more related patent identification information that is information for identifying related patent documents that are determined to be related as a result of screening work in patent search. One or more computers stored in the related patent identification information storage unit using elements included in the patent document identified by the one or more related patent identification information stored in the related patent identification information storage unit. A search expression generation unit that generates a search expression that can acquire a patent document identified by related patent identification information, and information that specifies a search patent document that is a patent document acquired using the search expression generated by the search expression generation unit Search patent identification information acquisition unit for acquiring search patent identification information, and characteristics of related patent documents identified by each related patent identification information stored in the related patent identification information storage unit One or more search patent feature vectors which are feature vectors of a search patent document identified by each search patent identification information acquired by the search patent identification information acquisition unit are acquired. Corresponds to a search patent feature vector similar to a feature of one or more related patent feature vectors acquired by the feature vector acquisition unit among one or more search patent feature vectors acquired by the feature vector acquisition unit and the feature vector acquisition unit The search patent identification information that is to be searched and the search patent identification information that does not match the related patent identification information stored in the related patent identification information storage unit is selected. This is a program for causing an output unit to function.

  In the present embodiment, each process (each function) may be realized by centralized processing by a single device (system), or by distributed processing by a plurality of devices. May be. In the present embodiment, it goes without saying that two or more communication means existing in one apparatus may be physically realized by one medium.

  In the present embodiment, each component may be configured by dedicated hardware, or a component that can be realized by software may be realized by executing a program. For example, each component can be realized by a program execution unit such as a CPU reading and executing a software program recorded on a recording medium such as a hard disk or a semiconductor memory.

  In the program, the functions realized by the program do not include functions that can be realized only by hardware. For example, functions that can be realized only by hardware such as a modem and an interface card in an acquisition unit that acquires information, an output unit that outputs information, and the like are not included in the functions realized by the program.

  FIG. 13 is a schematic diagram showing an example of the external appearance of a computer that executes the program and implements the present invention according to the embodiment. The embodiment described above can be realized by computer hardware and a computer program executed on the computer hardware.

  In FIG. 13, a computer system 1100 includes a computer 1101 including a CD-ROM drive 1105 and an FD drive 1106, a keyboard 1102, a mouse 1103, and a monitor 1104.

  FIG. 14 is a diagram showing an internal configuration of the computer system 1100. In FIG. 14, in addition to the CD-ROM drive 1105 and the FD drive 1106, a computer 1101 is connected to an MPU 1111, a ROM 1112 for storing programs such as a bootup program, and an MPU 1111. And a RAM 1113 that provides a temporary storage space, a hard disk 1114 that stores application programs, system programs, and data, an MPU 1111, a bus 1115 that interconnects the ROM 1112, and the like. The computer 1101 may include a network card (not shown) that provides connection to the LAN.

  A program for causing the computer system 1100 to execute the functions of the present invention according to the above embodiment is stored in the CD-ROM 1121 or the FD 1122, inserted into the CD-ROM drive 1105 or the FD drive 1106, and transferred to the hard disk 1114. May be. Instead, the program may be transmitted to the computer 1101 via a network (not shown) and stored in the hard disk 1114. The program is loaded into the RAM 1113 when executed. The program may be loaded directly from the CD-ROM 1121, the FD 1122, or the network.

  The program does not necessarily include an operating system (OS) or a third-party program that causes the computer 1101 to execute the functions of the present invention according to the above-described embodiment. The program may include only a part of an instruction that calls an appropriate function (module) in a controlled manner and obtains a desired result. How the computer system 1100 operates is well known and will not be described in detail.

  The present invention is not limited to the above-described embodiments, and various modifications are possible, and it goes without saying that these are also included in the scope of the present invention.

  As described above, the information processing apparatus and the like according to the present invention have an effect that related patents can be automatically selected from patent documents, and are useful as, for example, an apparatus for selecting related patent documents.

DESCRIPTION OF SYMBOLS 1 Information processing apparatus 101 Reception part 102 Related patent identification information storage part 103 Unrelated patent identification information storage part 104 Search formula production | generation part 105 Patent document storage part 106 Search patent identification information acquisition part 107 Similar element storage part 108 Feature vector acquisition part 109 Selection unit 110 Evaluation unit 111 Output unit

Claims (13)

A related patent identification information storage unit that stores one or more related patent identification information, which is information for specifying related patent documents, which are patent documents determined to be related as a result of screening work in the patent search;
One or more related patents stored in the related patent identification information storage unit using elements included in the patent document identified by the one or more related patent identification information stored in the related patent identification information storage unit A search expression generation unit that generates a search expression capable of acquiring a patent document identified by the identification information;
A search patent identification information acquisition unit that acquires search patent identification information that is information for specifying a search patent document that is a patent document acquired using the search formula generated by the search formula generation unit;
One or more related patent feature vectors that are feature vectors of related patent documents identified by each related patent identification information stored in the related patent identification information storage unit are acquired, and the search patent identification information acquisition unit is A feature vector acquisition unit that acquires one or more search patent feature vectors that are feature vectors of a search patent document identified by each acquired search patent identification information;
Search patent identification information corresponding to a search patent feature vector similar to a feature of one or more related patent feature vectors acquired by the feature vector acquisition unit among one or more search patent feature vectors acquired by the feature vector acquisition unit And a selection unit that selects at least search patent identification information that does not match the related patent identification information stored in the related patent identification information storage unit, and
An information processing apparatus comprising: an output unit that outputs information related to a result selected by the selection unit. It further includes an evaluation unit that performs a lower evaluation as more search patent identification information that does not match the related patent identification information stored in the related patent identification information storage unit among the search patent identification information selected by the selection unit. ,
The output unit is
The information processing apparatus according to claim 1, wherein a result of the evaluation by the evaluation unit is output. The selection unit includes:
Search patent identification information corresponding to a search patent feature vector similar to a feature of one or more related patent feature vectors acquired by the feature vector acquisition unit among one or more search patent feature vectors acquired by the feature vector acquisition unit The information processing apparatus according to claim 1 or 2, wherein search patent identification information corresponding to a patent document identified by the related patent identification information stored in the related patent identification information storage unit is also selected. . The selection unit includes:
Search patent identification information corresponding to a search patent feature vector similar to a feature of one or more related patent feature vectors acquired by the feature vector acquisition unit among one or more search patent feature vectors acquired by the feature vector acquisition unit And also search patent identification information corresponding to the patent document identified by the related patent identification information stored in the related patent identification information storage unit,
The evaluation unit is
Related patent identification information stored in the related patent identification information storage unit that is the search patent identification information acquired by the search patent identification information acquisition unit and does not match the search patent identification information selected by the selection unit The information processing apparatus according to claim 2, wherein the lower the evaluation, the lower the evaluation. And a non-related patent identification information storage unit for storing at least one non-related patent identification information, which is information for identifying an unrelated patent document that is determined to be irrelevant as a result of screening work in the patent search. ,
The search expression generation unit
1 further stored in the unrelated patent identification information storage unit using elements included in the patent document identified by one or more unrelated patent identification information stored in the unrelated patent identification information storage unit. The information processing apparatus according to any one of claims 1 to 4, wherein a search expression that does not acquire at least a part of the patent document identified by the above unrelated patent identification information is generated.

The feature vector acquisition unit
Further obtaining one or more unrelated patent feature vectors that are feature vectors of unrelated patent documents identified by each unrelated patent identification information stored in the unrelated patent identification information storage unit;
The selection unit includes:
Search patent identification corresponding to a search patent feature vector that is not similar to a feature of one or more unrelated patent feature vectors acquired by the feature vector acquisition unit among one or more search patent feature vectors acquired by the feature vector acquisition unit The information processing apparatus according to claim 5, wherein information is selected. And a non-related patent identification information storage unit for storing at least one non-related patent identification information, which is information for identifying an unrelated patent document that is determined to be irrelevant as a result of screening work in the patent search. ,
The feature vector acquisition unit
Further obtaining one or more unrelated patent feature vectors that are feature vectors of unrelated patent documents identified by each unrelated patent identification information stored in the unrelated patent identification information storage unit;
The selection unit includes:
Search patent identification corresponding to a search patent feature vector that is not similar to a feature of one or more unrelated patent feature vectors acquired by the feature vector acquisition unit among one or more search patent feature vectors acquired by the feature vector acquisition unit The information processing apparatus according to claim 1, wherein information is selected. The selection unit includes:
Among the one or more unrelated patent feature vectors acquired by the feature vector acquisition unit, the feature vector acquisition unit acquires 1 similar to the feature of the one or more related patent feature vectors acquired by the feature vector acquisition unit. The information processing apparatus according to claim 6 or 7, wherein non-related patent identification information corresponding to an unrelated patent feature vector that is not similar to the feature of the unrelated patent feature vector is also selected. And a non-related patent identification information storage unit for storing at least one non-related patent identification information, which is information for identifying an unrelated patent document that is determined to be irrelevant as a result of screening work in the patent search. ,
The feature vector acquisition unit
Further obtaining one or more unrelated patent feature vectors that are feature vectors of unrelated patent documents identified by each unrelated patent identification information stored in the unrelated patent identification information storage unit;
The selection unit includes:
Among the one or more unrelated patent feature vectors acquired by the feature vector acquisition unit, the feature vector acquisition unit acquires 1 similar to the feature of the one or more related patent feature vectors acquired by the feature vector acquisition unit. Also select unrelated patent identification information corresponding to the unrelated patent feature vector that is not similar to the features of the unrelated patent feature vector,
The evaluation unit is
The information processing apparatus according to claim 2 or 3, wherein the lower the evaluation is, the more unrelated patent identification information is selected by the selection unit. The feature vector acquisition unit
2. Each patent document identified by information identifying each patent document is acquired, one or more elements are acquired from each patent document, and each feature vector is acquired using the one or more elements. The information processing apparatus according to any one of claims 9 to 9. The feature vector acquisition unit
The information processing apparatus according to any one of claims 1 to 10, wherein each feature vector is obtained by using similar elements acquired from the respective patent documents as elements corresponding to the same vector elements. A related patent identification information storage unit that stores one or more related patent identification information, which is information for identifying related patent documents that are determined to be related as a result of screening work in the patent search, a search expression generation unit, An information processing method processed using a search patent identification information acquisition unit, a feature vector acquisition unit, a selection unit, and an output unit,
The search expression generation unit is stored in the related patent identification information storage unit using an element included in a patent document identified by one or more related patent identification information stored in the related patent identification information storage unit. A search expression generation step for generating a search expression capable of acquiring a patent document identified by the one or more related patent identification information;
The search patent identification information acquisition unit acquires search patent identification information that is information for specifying a search patent document that is a patent document acquired using the search formula generated in the search formula generation step. A step and the feature vector acquisition unit acquire one or more related patent feature vectors that are feature vectors of related patent documents identified by each related patent identification information stored in the related patent identification information storage unit; and A feature vector acquisition step of acquiring one or more search patent feature vectors that are feature vectors of the search patent document identified by each search patent identification information acquired in the search patent identification information acquisition step;
The selection unit corresponds to a search patent feature vector similar to a feature of one or more related patent feature vectors acquired in the feature vector acquisition step among one or more search patent feature vectors acquired in the feature vector acquisition step. A selection step of selecting at least search patent identification information that is search patent identification information that does not match the related patent identification information stored in the related patent identification information storage unit;
An information processing method comprising: an output step in which the output unit outputs information relating to a result selected in the selection step. A computer that can access a related patent identification information storage unit that stores one or more related patent identification information, which is information for identifying related patent documents that are determined to be related as a result of screening work in a patent search;
One or more related patents stored in the related patent identification information storage unit using elements included in the patent document identified by the one or more related patent identification information stored in the related patent identification information storage unit A search expression generation unit that generates a search expression capable of acquiring a patent document identified by the identification information;
A search patent identification information acquisition unit that acquires search patent identification information that is information for specifying a search patent document that is a patent document acquired using the search formula generated by the search formula generation unit,
One or more related patent feature vectors that are feature vectors of related patent documents identified by each related patent identification information stored in the related patent identification information storage unit are acquired, and the search patent identification information acquisition unit is A feature vector acquisition unit that acquires one or more search patent feature vectors that are feature vectors of a search patent document identified by each acquired search patent identification information;
Search patent identification information corresponding to a search patent feature vector similar to a feature of one or more related patent feature vectors acquired by the feature vector acquisition unit among one or more search patent feature vectors acquired by the feature vector acquisition unit And a selection unit that selects at least search patent identification information that does not match the related patent identification information stored in the related patent identification information storage unit,
The program for functioning as an output part which outputs the information regarding the result which the said selection part selected.

Images