Information retrieval system

Abstract

An information retrieval system (1000) comprises a technical information acquisition unit (112a1), a retrieval unit (112c), a document set generation unit (112d), a solution extraction unit (112f), a problem extraction unit (112e), and a provision unit (112h). The technical information acquisition unit (112a1) acquires technical information including a technical problem. The retrieval unit (112c) retrieves patent documents having disclosures that are the same as or similar to the technical problem. The document set generation unit (112d) generates a set comprising a plurality of retrieved patent documents. The solution means extraction unit (112f) extracts the means for solving the problems of the inventions according to the patent documents included in the set. The problem extraction unit (112e) extracts the problems to be solved by the inventions according to the patent documents included in the set. The provision unit (112h) provides the extracted means for solving the problems and the problems.

Description

Information Retrieval System

[0001] The present invention relates to an information retrieval system.

[0002] Conventionally, it is known to search for patent information that meets the patent information search conditions set based on a user's input and display it on the display unit of a client device (Patent Document 1). In Patent Document 1, gazettes issued in the past regarding industrial property rights such as patents are used as patent information. Also, in Patent Document 1, the patent information search conditions are, for example, IPC classification, F-term, keyword, applicant or patentee, etc.

[0003] Japanese Patent Application Laid-Open No. 2019-067330

[0004] By the way, when developing a solution to a technical problem faced by a certain company, it is conceivable to search for patent documents based on technical information and use the information described in the patent documents as a solution to the problem.

[0005] However, in the technology described in the above patent document, it merely searches for patent information that meets the patent information search conditions.

[0006] An object of the present disclosure is to provide an information retrieval system capable of accurately retrieving useful information for those who are trying to develop a solution to a problem they are facing.

[0007] The gist of the present disclosure is as follows.

[0008] (1) The information retrieval system includes a technical information acquisition unit, a search unit, a document set generation unit, a solution extraction unit, a problem extraction unit, and a provision unit. The technical information acquisition unit acquires technical information including technical problems. The search unit searches for patent documents having a description that matches or is similar to the technical problem. The document set generation unit generates a set consisting of a plurality of the searched patent documents. The solution extraction unit extracts the problem-solving means of the invention related to the patent documents included in the set. The problem extraction unit extracts the problems that the invention related to the patent documents included in the set is trying to solve. The provision unit provides the extracted problem-solving means and the problems.

[0009] (2) The information retrieval system, in (1) above, identifies a technical field from the above technical problem and further searches for patent documents that are included in the above technical field and describe a problem different from the above technical problem.

[0010] (3) The information retrieval system further comprises a problem-level conceptualization unit and a solution-level conceptualization unit, as described in (1) or (2) above. The problem-level conceptualization unit groups the problems extracted by the problem extraction unit. The solution-level conceptualization unit groups the problem-solving means extracted by the solution-solving means extraction unit.

[0011] (4) In the information retrieval system described in (3) above, the providing unit provides a matrix-like map that associates the grouped problem-solving means with the grouped problems.

[0012] (5) In the information retrieval system, in (3) or (4) above, the above-mentioned problem-level conceptualization unit further conceptualizes the grouped problems, and the above-mentioned solution-level conceptualization unit further conceptualizes the grouped problem-solving means.

[0013] (6) In any of (1) to (5) above, the information retrieval system, the search unit searches for patent documents based on a new technical problem entered by a user who has recognized the provided problem-solving means. The search unit searches for patent documents that contain descriptions that match or are similar to the new technical problem. The document set generation unit generates a new set consisting of the newly retrieved patent documents. The solution extraction unit extracts the problem-solving means of the invention related to the patent documents included in the newly generated set.

[0014] (7) The information retrieval system further comprises a learning unit in any of (1) to (6) above. The learning unit learns an algorithm for when the search unit searches for the patent documents based on the user's evaluation of the set of patent documents.

[0015] This disclosure provides an information retrieval system that enables those seeking to develop solutions to challenges they face to search for useful information with high accuracy.

[0016] This figure shows an information retrieval system according to one embodiment of the present disclosure. This figure shows the configuration of the search server. This figure shows the configuration of the user's terminal. This figure shows the functional configuration of the search server. This figure shows the functional configuration of the user's terminal. This figure shows how a list of patent documents generated by the document collection generation unit is displayed on the screen of the display device. This figure shows an example of a user's evaluation of search results. This figure shows an example of a user's evaluation of a problem and a means of solving the problem. This figure shows an example of displaying multiple different problems A to C and multiple different means of solving the problem a to d in a bubble diagram. This figure shows an example of a user clicking and selecting a problem on the display device screen to discover a new means of solving the problem. This figure shows an example of a bubble diagram being displayed in a matrix including a new means of solving the problem e in Figure 9. This figure shows the sequence of processing performed by the search system.

[0017] Several embodiments relating to this disclosure will be described below with reference to the figures. However, these descriptions are intended to be merely illustrative of preferred embodiments of this disclosure and are not intended to limit this disclosure to such specific embodiments. In the following descriptions, similar components will be given the same reference numerals, and redundant descriptions will be omitted where appropriate.

[0018] An information retrieval system 1000 according to one embodiment of the present disclosure includes a search server 100 for retrieving information, as shown in Figure 1. The information retrieval system 1000 may further include a user terminal 200 and a generation AI server 300. The functions of the information retrieval system 1000 may be realized by the search server 100, the terminal 200, and the generation AI server 300. On the other hand, the information retrieval system 1000 may consist only of the search server 100. In that case, the search server 100 may also have the functions of the terminal 200 or the generation AI server 300. If the information retrieval system 1000 further includes a user terminal 200 and a generation AI server in addition to the search server 100, the search server 100, the terminal 200, and the generation AI server 300 may be able to communicate with each other via a communication network 500. The communication network 500 consists of communication lines. In other words, the communication network 500 relays communication between the search server 100, the terminal 200, and the generation AI server 300. The double arrows shown in Figure 1 indicate that information is sent and received via communication. The generation AI server 300 is configured to execute generation AI processing that generates content by utilizing relationships learned from data. The generation AI can generate content based on prompts given by the user. For example, the generation AI can learn patterns and relationships using a set of conditions and correct answers from previously collected data as training data, and then generate content. A commercial generation AI can be used as the generation AI server 300. Such a generation AI may generate program code, images, videos, audio, etc. For example, the use of generation AI that generates text data such as ChatGPT is becoming increasingly common. Examples of commercial text generation AIs include GPT-4 (registered trademark), Llama2, or PaLM2. By using a commercial generation AI, it is not necessary to collect training data individually, and the system can be easily constructed. The generation AI server 300 may be provided separately from the search server 100, or the functions of the generation AI server 300 may be integrated into the search server 100.

[0019] For example, when a company tries to develop a solution to a problem it faces, it may search for patent documents based on technical information. The company may then use the information described in the patent documents as a means of solving the problem. In such a case, when the company transmits technical information from terminal 200 as a user, the search server 100 receives the technical information via the communication network 500. The search server 100 searches the literature database to find patent documents that match the technical information. In other words, the search server 100 can use the literature database to search for information on solutions related to the technical information that represents the problem. Patent documents include patent gazettes, patent publications, or equivalent documents issued by the regulatory authorities of each country, or processed versions thereof. Patent documents may be digitized information. Processed documents may include summarized or translated documents. The search server 100 may also search for non-patent information that matches the technical information. In this case, the search server 100 may, for example, communicate with other data servers (not shown) to obtain non-patent information stored in those data servers. Non-patent information may include non-patent literature or information published on the internet. Specifically, information published on the internet may include information on the websites of individual companies. Non-patent literature may also include technical papers.

[0020] In particular, in this embodiment, the technical information includes information on one or more attributes. One or more attributes include a technical problem. One or more attributes may further include, for example, the name of the substance constituting the product as a means of solving the problem, the structural characteristics of the product, and the functions that the product can provide. One or more attributes may further include, for example, the use of the product as a means of solving the problem. For example, the user inputs "prevent the generation of die buildup at the extruder outlet when extruding XX resin" as a technical problem into terminal 200. Here, XX resin may be, for example, a thermoplastic resin, or for example, polyvinyl chloride. The phenomenon called "die buildup" in the example given here is the accumulation of thermally degraded resin near the die outlet of the extruder, causing streaks or scratches on the surface of the product.

[0021] For example, when technical information containing these attributes is input to terminal 200, it transmits the technical information to search server 100. Search server 100 searches for patent documents containing descriptions that match or are similar to the technical information. Search server 100 can search for descriptions that match or are similar to the technical information by utilizing highly co-occurring words, which indicate how frequently a certain word appears together with other words. More specifically, search server 100 performs morphological analysis to divide each sentence in the technical information and patent documents into the smallest meaningful morphemes. At this time, text normalization may be performed after deleting unnecessary words. Examples of unnecessary words include stop words and HTML tags. Next, search server 100 vectorizes the words that have undergone morphological analysis and stores them in a memory device. Search server 100 determines the cosine similarity between the vectors of sentences in the technical information and patent documents and extracts sentences with higher similarity from the patent documents. In this way, search server 100 can search for descriptions that match or are similar to the technical information from the patent documents.

[0022] For example, when the search server 100 receives technical information, it may focus its search on predetermined sections of patent documents determined according to the individual attributes of the technical information. Here, focusing the search on predetermined sections means giving more weight to the search when searching predetermined sections than when searching other sections. Also, for example, focusing the search on predetermined sections means achieving higher search accuracy when searching predetermined sections than when searching other sections. As an example, for example, the search server 100 searches for multiple keywords included in the attributes of the technical information using AND conditions for predetermined sections. Then, for other sections not specified in the search server 100 searches for multiple keywords included in the attributes of the technical information using OR conditions. Also, as an example, for example, the search server 100 broadens the range of similar words to the keywords in the attributes of the technical information when searching predetermined sections than when searching other sections. Also, as an example, when the search server 100 quantifies the attributes of the technical information and the descriptions in patent documents to determine the degree of match and similarity, it sets a higher threshold for predetermined sections than for other sections. For example, the search server 100 may focus its search on the sections within the full text of individual patent documents that describe the problem the invention aims to solve. The search server 100 may also perform a focused search by weighting each item described in the patent document that is necessary for a person skilled in the art to understand the technical significance of the invention.

[0023] The search server 100 then sends a set of patent documents containing descriptions that match or are similar to the technical information to the terminal 200 as search results. The search results are displayed on the screen of the terminal 200.

[0024] In particular, when searching for hints for developing solutions to a problem from a vast number of patent documents, it is preferable to perform the search based on the technical problem. The solution to the problem may be a specific product. In other words, in this embodiment, the search server 100 basically searches for technical information based on the technical problem that the user wants to solve. In addition, it is preferable to perform the search based on the names of the materials constituting the product as a solution to the problem, the structural characteristics, and the function as attributes of the technical information. In particular, if only the names of the materials constituting the product are used, the set of patent documents in the search results may become large. In that case, it is preferable to perform the search based on the structural characteristics of the product in addition to the names of the materials constituting the product as attributes of the technical information.

[0025] In particular, in this embodiment, the search server 100 may focus its search on predetermined locations in patent documents that are determined according to the individual attributes of the technical information. This allows the search server 100 to improve the accuracy of searching for predetermined locations corresponding to individual attributes, thereby suppressing the inclusion of noise in the search results.

[0026] As shown in Figure 2, the search server 100 includes a first control device 110 and a storage device 120. In other words, the search server 100 functions as an information retrieval device.

[0027] The first control device 110 includes a first processor 112, a first memory 114, and a first communication interface 116. The first processor 112, the first memory 114, and the first communication interface 116 are connected via a first communication bus 118. The first processor 112 has one or more CPUs (Central Processing Units) and their peripheral circuits. The first processor 112 may further have other arithmetic circuits such as a logic unit, a numerical unit, or a graphics processing unit. The first memory 114 includes, for example, a volatile semiconductor memory and a non-volatile semiconductor memory. The first communication interface 116 corresponds to the first communication I / F shown in Figure 2 and has an interface circuit for connecting the first control device 110 to the network in the search server 100. The first communication interface 116 also has an interface circuit for connecting to the communication network 500. The first communication interface 116 is configured to communicate with, for example, a terminal 200 and a generation AI server 300 via a communication network 500.

[0028] The storage device 120 includes, for example, a hard disk drive or an optical recording medium and its access device. In other words, the storage device 120 functions as a storage device. The storage device 120 may function as a patent database. The storage device 120 stores information from patent documents. The storage device 120 may also store synonyms of keywords that are associated with keywords constituting the technical information. The storage device 120 may also store computer programs for executing processes performed on the first processor 112.

[0029] As shown in Figure 3, terminal 200 includes a second control device 210, a display device 220, an input device 230, and a speaker 240. Terminal 200 may be a personal computer or a mobile device. The second control device 210 includes a second processor 212, a second memory 214, and a second communication interface 216. The second processor 212, the second memory 214, and the second communication interface 216 are connected via a second communication bus 218. The second communication interface 216 corresponds to the second communication I / F shown in Figure 3. Terminal 200 may be configured similarly to the configurations of the first processor 112, the first memory 114, and the first communication interface 116 of the search server 100.

[0030] Each function of the information retrieval system 1000 is realized by a functional block executed by the processor. As an example, the first processor 112, as shown in Figure 4, includes an information acquisition unit 112a, a search range specification unit 112b, a search unit 112c, a literature collection generation unit 112d, and a problem extraction unit 112e. The first processor 112 further includes a solution means extraction unit 112f, a priority setting unit 112g, a provision unit 112h, a prompt creation unit 112i, a prompt transmission unit 112j, and a learning unit 112k. The first processor 112 further includes a problem higher-level conceptualization unit 112m and a solution higher-level conceptualization unit 112n.

[0031] Each of these components of the first processor 112 is, for example, a functional module realized by a computer program running on the first processor 112. In other words, each of these components of the first processor 112 consists of the first processor 112 and a program, i.e., software, to make it function. The program may also be recorded in the first memory 114 of the first control device 110 or on a recording medium connected to the outside. Alternatively, each of these components of the first processor 112 may consist of a dedicated arithmetic circuit provided on the first processor 112. The second processor 212 of the terminal 200 shown in Figure 5 is realized by functional blocks executed by the second processor 212, similar to the first processor 112.

[0032] Note that the processor configurations shown in Figures 4 and 5 are examples, and individual components of one processor may be provided in other processors. For example, at least a portion of the functional blocks of the search server 100's processor may be provided in the terminal 200's processor. Also, individual components of one processor may be redundantly provided in other processors.

[0033] The following describes the functions of the information retrieval system 1000, which are realized by the functional blocks provided by the processor.

[0034] First, the functional blocks of the first processor 112 of the search server 100 will be described. The information acquisition unit 112a shown in Figure 4 acquires information transmitted from the terminal 200 or the generation AI server 300. Specifically, the information acquisition unit 112a includes a technical information acquisition unit 112a1. The technical information acquisition unit 112a1 acquires technical information transmitted from the terminal 200. The technical information includes technical issues. The information acquisition unit 112a may acquire identification information indicating which of one or more attributes the technical information corresponds to. The information acquisition unit 112a may acquire a response from the generation AI server 300 to a prompt sent to the generation AI server 300. The information acquisition unit 112a may acquire a user's evaluation of the search results transmitted from the terminal 200. Furthermore, the information acquisition unit 112a may acquire an issue for discovering a new solution, which has been entered by the user based on the search results. The issue for discovering a new solution corresponds, for example, to issue C, which will be described later in Figure 10.

[0035] The search range specification unit 112b specifies a predetermined location that is the search range for patent documents based on the attributes of the technical information. The search range specification unit 112b may specify the "problem to be solved by the invention" of the patent document as the predetermined location. The search range specification unit 112b may also specify a predetermined location that is the search range for patent documents based on identification information indicating which of one or more attributes the technical information corresponds to.

[0036] The search unit 112c uses a patent database to search for patent documents that contain descriptions matching or similar to the technical problem described in the technical information. The search unit 112c searches for patent documents that contain descriptions matching or similar to the technical information by determining whether the patent documents contain descriptions matching or similar to the technical information.

[0037] The search unit 112c may search predetermined locations in patent documents corresponding to individual attributes of the technical information with greater emphasis than other locations, and determine whether there are any descriptions in the patent documents that match or are similar to the technical information. Specifically, the search unit 112c accesses each of the patent documents stored in the storage device 120 and searches predetermined locations within each patent document specified by the search range specification unit 112b with greater emphasis. The search unit 112c then determines whether there are any descriptions in the patent documents that match or are similar to the technical information. Alternatively, the search unit 112c may search only predetermined locations and determine whether there are any descriptions in those locations that match or are similar to the technical information. An external patent database connected via the communication network 500 may be used as the patent database. Examples of patent databases include databases containing patent information from publications issued by regulatory authorities of each country, and databases that compile patent information from published publications. The patent database may be either paid or free.

[0038] For example, the search unit 112c determines whether the technical problem and the description in the patent document match or are similar. In this case, the search unit 112c gives more weight to the search for the description of "the problem the invention aims to solve" in the patent document designated as a predetermined location, compared to other descriptions in that patent document, and makes its determination. As a result, patent documents in which the problem of the technical information and the description of "the problem the invention aims to solve" match or are similar are determined with a higher degree of certainty to contain descriptions that match or are similar to the technical information.

[0039] The search unit 112c may select a patent classification corresponding to the technical information and search for patent documents using a search formula that utilizes the patent classification. In this case, the search unit 112c may search for patent classifications based on the technical information, create a search formula that includes the searched patent classifications and the technical information, and then search for patent documents using the created search formula. This identifies the technical field that constitutes the search scope. The patent classification may include the International Patent Classification, FI, and F-terms. Note that the patent classification may also be included in the technical information entered by the user. The search unit 112c may limit the search scope to technical fields directly related to the technical problem entered by the user. To limit the search scope in advance to technical fields directly related to the problem in the technical information, the search unit 112c may use IPC or F-terms corresponding to those technical fields.

[0040] The search unit 112c determines whether there are descriptions in the patent document that match or are similar to the technical information, for example, by using a full-text search method. Alternatively, the search unit 112c may decompose both the technical information and the descriptions in the patent document into phrases using natural language processing, and then determine whether there are descriptions in the patent document that match or are similar to the technical information. Specifically, the search unit 112c analyzes the morphological elements of both the technical information and the descriptions in the patent document. Then, the search unit 112c may determine whether there are descriptions in the patent document that match or are similar to the technical information by vectorizing each of the morphological elements and determining the cosine similarity. In this case, the search unit 112c may determine whether there are descriptions in the patent document that match or are similar to the technical information by setting a threshold for determining the similarity. Alternatively, for example, the search unit 112c may determine whether there are descriptions in the patent document that are similar to the technical information based on synonyms of individual keywords that are associated with each keyword constituting the technical information. Synonyms for each keyword that constitutes the technical information may be stored in the storage device 120.

[0041] The search unit 112c may search for patent documents having descriptions that match or are similar to the technical information using the generation AI server 300. In this case, the search unit 112c acquires the patent documents transmitted from the generation AI server 300 as an answer to the prompt. Further, the search unit 112c may search for patent classifications corresponding to the technical information using the generation AI server 300. In this case, the search unit 112c acquires the patent classifications transmitted from the generation AI server 300 as an answer to the prompt.

[0042] Furthermore, the search unit 112c may specify the technical field from the technical problem, and further search for patent documents included in the specified technical field and having problems described therein that are different from the technical problems included in the technical information. As described above, the search unit 112c may search for patent classifications based on the technical information, create a search formula including the searched patent classifications and the technical information, and search for patent documents using the created search formula. In this case, in order to further search for patent documents having problems described therein that are different from the technical problems included in the technical information, the search unit 112c searches in the technical field specified from the technical problem using a new search formula obtained by removing the search conditions related to the technical problem from the search formula. Thereby, the search unit 112c can search for patent documents included in the technical field specified from the technical problem and having problems described therein that are different from the technical problems included in the technical information.

[0043] The search unit 112c may narrow down the search results by searching for patent documents having a high degree of match between the problems extracted by the problem extraction unit 112e described later, or the problems summarized, upper-conceptualized, or grouped by the problem upper-conceptualization unit 112m and the technical problems initially input by the user.

[0044] The search unit 112c may acquire from the generation AI server 300 new technical problems generated from the technical problems specified by the user. Then, the search unit 112c may search for patent documents included in the specified technical field and having problems described therein that are different from the technical problems included in the technical information based on the acquired new technical problems.

[0045] Further, for example, the search unit 112c may obtain a new technical problem based on the technical problem specified by the user from the learned model. In this case, the learned model may be learned based on teacher data that takes a technical problem as an input and outputs a similar or different technical problem. Thus, the search unit 112c may perform a search not only based on the technical problem input by the user, but also based on a new technical problem obtained from the technical problem input by the user. Further, the search unit 112c may search for patent documents in which problems similar to those not included in the specified technical field and not included in the technical information are described, based on the obtained new technical problem.

[0046] For example, the search unit 112c searches for patent documents within the range of the technical field specified by the search formula created based on the technical problem input by the user. In so doing, the search unit 112c may search for patent documents based on a new technical problem other than the technical problem input by the user.

[0047] The document set generation unit 112d generates a set consisting of a plurality of patent documents searched by the search unit 112c. The document set generation unit 112d generates a set of patent documents having descriptions that match or are similar to the technical information, based on the search result of the search unit 112c. Specifically, the document set generation unit 112d generates a set of patent documents determined to have descriptions that match or are similar to the attributes of the technical information. That is, a patent document that matches the technical information may be a patent document determined to have a description that matches or is similar to the attributes of the technical information. Each individual patent document included in the set of patent documents may include the claims, the specification, the abstract, the drawings, bibliographic information, and the abstract.

[0048] Further, the document set generation unit 112d may generate a patent map from the set of patent documents. The generated patent map is included in the set of patent documents. The patent map may be generated according to the relationship between the problem that the invention according to the patent document seeks to solve and the problem-solving means of the invention according to the patent document.

[0049] The problem extraction unit 112e extracts the problem that the invention related to that patent document aims to solve from each patent document retrieved by the search unit 112c. Specifically, the problem extraction unit 112e may extract the description of the problem from the "Problem to be solved by the invention" column of the patent document. The extracted problem is included in the information of each individual patent document in the set of patent documents and provided to the user. For example, the problem extraction unit 112e may extract the content of the "Problem to be solved by the invention" column of the patent document as is.

[0050] The problem extraction unit 112e may use the generation AI server 300 to extract, summarize, and list the problems described in each patent publication.

[0051] The problem-level conceptualization unit 112m summarizes and conceptualizes the problems extracted by the problem extraction unit 112e. Summarization may be performed by so-called extractive summarization, abstractive summarization, or generative summarization. The problem-level conceptualization unit 112m also groups similar problems. For example, the problem-level conceptualization unit 112m analyzes the morphological elements of the extracted problem description, or the description of the problem summarized or conceptualized above. Then, the problem-level conceptualization unit 112m vectorizes each of the morphological elements and determines the cosine similarity to group similar problems. The problem-level conceptualization unit 112m may also use the generation AI server 300 to summarize, conceptualize above, and group problems. The problem-level conceptualization unit 112m may further conceptualize the grouped problems. The extracted problems, or the summarized, conceptualized above, or grouped problems, are included in the information of individual patent documents included in the set of patent documents and provided to the user.

[0052] The solution extraction unit 112f extracts the means for solving the problem of an invention related to an individual patent document included in the set of patent documents from that patent document. The solution extraction unit 112f extracts the means for solving the problem from each of the retrieved patent documents. Specifically, the problem extraction unit 112e may extract the means for solving the problem from the "claims" section of the patent document. Alternatively, the problem extraction unit 112e may extract the means for solving the problem from the "means for solving the problem," "modes for carrying out the invention," and "examples" sections. For example, the solution extraction unit 112f may extract the contents of the "claims" section of the patent document as is. In particular, in patent documents relating to material patents and substance patents, the "examples" section contains specific names of substances as means for solving the problem. For this reason, in patent documents relating to material patents and substance patents, it is preferable for the solution extraction unit 112f to extract substances and materials as means for solving the problem from the "examples" section.

[0053] The solution extraction unit 112f may extract, summarize, and list the problem-solving means described in each patent publication, for example, using the generation AI server 300.

[0054] The solution means higher conceptualization unit 112n summarizes and conceptualizes the solution means extracted by the solution means extraction unit 112f. The summarization may be performed by so-called extractive summarization, abstractive summarization, or generative summarization. The solution means higher conceptualization unit 112n also groups similar solution means. For example, the solution means higher conceptualization unit 112n analyzes the morphological elements of the extracted solution means description, or the description of the solution means summarized or conceptualized higher. The solution means higher conceptualization unit 112n then vectorizes each of the morphological elements and determines the cosine similarity to group similar solution means. For example, the solution means extraction unit 112f extracts multiple methods for preventing eye discharge from patent documents as solution means corresponding to the above-mentioned problem of "preventing the occurrence of eye discharge." The solution means higher conceptualization unit 112n may then group the solution means "containing an antiblocking agent" from among the multiple extracted solution means. The solution means higher conceptualization unit 112n may use the generation AI server 300 to summarize, conceptualize, and group the solution means. The solution means conceptualization unit 112n may further conceptualize the grouped problem-solving means. The extracted problem-solving means, or the summarized, conceptualized, or grouped solution means, are included in the information of individual patent documents included in the collection of patent documents and provided to the user.

[0055] As described later, the multiple issues extracted and classified by the issue extraction unit 112e and the multiple solutions extracted and classified by the solution extraction unit 112f may be displayed as a bubble diagram in a matrix on the terminal 200. The issues and solutions may be summarized, conceptualized, or grouped before being displayed as a bubble diagram. These patent maps are included in a collection of patent documents and provided to the user. The document collection generation unit 112d generates a matrix-like map that associates the issues with the solution solutions. Specifically, the document collection generation unit 112d automatically generates the bubble diagram in Figure 9, described later, using the multiple issues extracted and classified by the issue extraction unit 112e and the multiple solutions extracted and classified by the solution extraction unit 112f. The document collection generation unit 112d may also automatically generate the bubble diagram in Figure 9, described later, using the summarized, conceptualized, or grouped issues and solutions.

[0056] The priority setting unit 112g sets priorities for patent documents included in a set of multiple patent documents. For example, the priority setting unit 112g sets a higher priority for providing a patent document to the user if the degree of agreement with the technical information is high. For example, the priority setting unit 112g sets a higher priority for a patent document if the degree of agreement between the technical information and the description in the patent document is high. For example, the priority setting unit 112g sets a higher priority for a document in which the attribute of the technical information has a high degree of agreement with the description in a predetermined location corresponding to that attribute, than for a document in which the attribute of the technical information has a high degree of agreement with other descriptions other than the predetermined location. Priorities may be included in a set of patent documents.

[0057] The provisioning unit 112h provides the user with a set of patent documents generated by the document set generation unit 112d. Specifically, the provisioning unit 112h transmits the set of patent documents generated by the document set generation unit 112d to the terminal 200 as a search result. If a priority is set for the patent documents by the priority setting unit 112g, the provisioning unit 112h transmits the priority along with the set of patent documents. Furthermore, if the set of patent documents includes a patent map, the provisioning unit 112h transmits the patent map to the terminal 200 in order to provide it to the user.

[0058] The prompt creation unit 112i creates various prompts to send to the generating AI server 300. A prompt refers to instructions or information to be input to the generating AI. One example of a prompt is a prompt for searching for patent documents that contain descriptions matching or similar to the technical information. For example, if the technical problem expressed in the technical information is problem A, the prompt creation unit 112i creates a prompt such as, "Please tell me the patent documents that contain descriptions matching or similar to technical problem A." Another example of a prompt is a prompt for obtaining a new technical problem from a technical problem specified by the user. For example, if the technical problem expressed in the technical information is problem A, the prompt creation unit 112i creates a prompt such as, "Please tell me a new technical problem similar to technical problem A." Furthermore, another example of a prompt is a prompt for extracting problems or solutions from patent documents, summarizing or conceptualizing them, and grouping them. For example, the prompt creation unit 112i creates a prompt for extracting problems from patent documents such as, "Please tell me the problems described in this patent document." Furthermore, for example, the prompt generation unit 112i creates a prompt for extracting problem-solving means from patent documents, such as, "Please tell me the problem-solving means described in this patent document." Furthermore, for example, the prompt generation unit 112i creates a prompt for summarizing a problem, such as, "Please summarize the problem described in this patent document." Furthermore, for example, the prompt generation unit 112i creates a prompt for summarizing problem-solving means, such as, "Please summarize the problem-solving means described in this patent document." Furthermore, for example, the prompt generation unit 112i creates a prompt for raising the level of a problem, such as, "Please generalize and raise the level of a problem described in this patent document." Furthermore, for example, the prompt generation unit 112i creates a prompt for raising the level of a problem-solving means, such as, "Please redefine and raise the level of a problem-solving means described in this patent document."Furthermore, for example, the prompt generation unit 112i creates a prompt for grouping problems, such as, "Group together similar problems from among those described in these patent documents." Also, for example, the prompt generation unit 112i creates a prompt for grouping problem-solving means, such as, "Group together similar problem-solving means from among those described in these patent documents."

[0059] The prompt sending unit 112j sends the prompt created by the prompt creation unit 112i to the generation AI server 300.

[0060] The learning unit 112k learns an algorithm for when the search unit 112c searches for patent documents, based on the user's evaluation of the set of patent documents. As an example, the learning unit 112k uses training data that takes technical information as input and outputs a set of patent documents from which patent documents evaluated as noise by the user have been removed. Based on this training data, the learning unit 112k trains a neural network to create a trained model. In subsequent searches, by using the trained model, patent documents evaluated as noise by the user will not be included in the set, and the user will obtain search results that are more desirable. However, if the learning is biased, the search results desired by the user may not be obtained. For this reason, the user may be able to choose whether or not to perform evaluation-based learning by operating the input device 230 and making a predetermined input. In this case, if the user chooses to perform evaluation-based learning, the learning unit 112k will perform learning, and if the user chooses not to perform evaluation-based learning, the learning unit 112k will not perform learning.

[0061] Furthermore, the learning unit 112k may learn an algorithm for when the search unit 112c searches for patent documents based on the user's evaluation of the problem-solving means. If the user inputs information to modify the problem-solving means extracted as a search result, the learning unit 112k may learn an algorithm for when the search unit 112c searches for patent documents based on that information. An example of modifying a problem-solving means is, for example, if the problem-solving means extracted is "controlling particle shape," the user may modify it to "controlling specific surface area."

[0062] Furthermore, the learning unit 112k may learn an algorithm for when the search unit 112c searches for patent documents based on the user's evaluation of the problem. If the user inputs information to modify a problem extracted as a search result, the learning unit 112k may learn an algorithm for when the search unit 112c searches for patent documents based on that information. An example of modifying a problem is when the extracted problem is "improve heat resistance" and the user modifies it to "improve flame retardancy".

[0063] The learning unit 112k may use training data that takes technical information as input and outputs a set of patent documents from which patent documents in which the user has evaluated both the problem-solving means and the problem as noise have been removed. Alternatively, the learning unit 112k may use training data that takes technical information as input and outputs a set of patent documents from which patent documents in which the user has evaluated either the problem-solving means or the problem as noise have been removed.

[0064] Furthermore, the learning unit 112k may learn an algorithm for when the search unit 112c searches for patent documents based on the user's evaluation of grouped problem-solving means or problems. In this case, as an example, the learning unit 112k uses training data in which technical information is taken as input and a set of patent documents is output, which is obtained by removing the patent documents from the set of patent documents that the user has evaluated as noise.

[0065] Next, the functional blocks of the second processor 212 of the terminal 200 will be described. As an example, the second processor 212, as shown in Figure 5, includes an input information acquisition unit 212a, a transmission unit 212b, a search result receiving unit 212c, and a display processing unit 212d.

[0066] The input information acquisition unit 212a acquires information entered by the user of the terminal 200 by operating the input device 230. Specifically, the input information acquisition unit 212a acquires technical information entered by the user, for example, by referring to the screen of the display device 220. The terminal 200 displays the screen on the display device 220 where the user enters technical information.

[0067] Furthermore, the input information acquisition unit 212a acquires evaluations of the search results entered by the user. In addition, the input information acquisition unit 212a acquires problems entered by the user for discovering new solutions. Furthermore, the input information acquisition unit 212a acquires problem-solving means or information for modifying problems that have been extracted as search results and entered by the user.

[0068] The transmission unit 212b transmits the information acquired by the input information acquisition unit 212a to the search server 100. The transmission unit 212b transmits technical information consisting of one or more attributes to the search server 100. The transmission unit 212b also transmits identification information to the search server 100 indicating which of the one or more attributes the technical information corresponds to. Therefore, the search server 100 can specify a predetermined area, which is the search range for patent documents, based on the identification information. The identification information is set, for example, according to which of the multiple input fields corresponding to one or more technical pieces of information displayed on the display device 220 was entered.

[0069] The search result receiving unit 212c receives the search results transmitted from the search server 100. Specifically, the search result receiving unit 212c receives a set of patent documents transmitted from the search server 100. If a priority is set for a patent document by the priority setting unit 112g of the search server 100, the search result receiving unit 212c receives the priority.

[0070] The display processing unit 212d performs processing to display the search results received from the search server 100 on the display device 220 of the terminal 200. When the search result receiving unit 212c receives a priority, the display processing unit 212d displays the search results on the display device 220 according to the priority. For example, patent documents are displayed from top to bottom in order of priority. Alternatively, for example, patent documents may be displayed with labels indicating their priority. For example, the terminal 200 may cause the display device 220 to display labels indicating priority such as "High Priority," "Medium Priority," and "Low Priority." This allows the user to browse patent documents included in the set of patent documents according to their priority, and to easily identify the desired patent documents. The search results may also be output as audio by the speaker 240.

[0071] Furthermore, the screen of the display device 220 may display the contents of the patent document along with the patent document number, for example, the claims or abstract may be displayed. Also, when the user clicks on the patent document number, the screen may transition to a link in the patent document database, and detailed information about that patent document may be displayed. The information or URL of the link displayed on the display device 220 is included in the set of patent documents transmitted from the search server 100 to the terminal 200. The screen of the display device 220 can be scrolled by the user of the terminal 200 operating the input device 230.

[0072] Figure 6 shows how a list representing the collection of patent documents generated by the document collection generation unit 112d of the search server 100 is displayed on the screen of the display device 220. The list displays the problem that the invention aims to solve, which has been extracted by the problem extraction unit 112e of the search server 100. The list also displays the means for solving the problem, which has been extracted by the solution means extraction unit 112f of the search server 100. As an example, Figure 6 shows the search results when a user inputs "prevention of lint generation at the extruder outlet when extruding XX resin" as a technical problem into the terminal 200. Specifically, Figure 6 shows patent documents as search results where the problem that the invention aims to solve is "prevention of lint generation in XX resin," "suppression of lint generation in □□ resin," and "reduction of lint generation in △△ resin." Also, Figure 6 shows patent documents as search results that are included in the technical field identified from the technical problem entered by the user, but which describe a different problem from the technical problem entered by the user. In Figure 6, the patent documents of applicant d and applicant e describe problems different from the technical problem entered by the user. Also, in Figure 6, as an example, the status of whether the rights remain valid is shown as "valid" and "expired". Patent documents with "valid rights" may be displayed with shading as shown in Figure 6. This allows the user to easily distinguish between patent documents with "expired rights" that offer a means of solving the problem and patent documents with "valid rights". Note that for published gazettes where a patent has not been granted, the column indicating whether the rights remain valid may be displayed blank.

[0073] A user who views the search results in Figure 6 can evaluate the search results. Figure 7 shows an example of an evaluation when a patent document whose problem the invention aims to solve is "prevention of the generation of grease buildup in XX resin" is useful to the user. Figure 7 also shows an example of an evaluation when a patent document whose problem the invention aims to solve is "suppression of the generation of grease buildup in □□ resin" is useful to the user. Furthermore, Figure 7 shows an example of an evaluation when a patent document whose problem the invention aims to solve is "reduction of the generation of grease buildup in △△ resin" is not useful to the user. In the example in Figure 7, the user evaluates the patent document whose problem the invention aims to solve is "prevention of the generation of grease buildup in XX resin," i.e., the document from applicant company a, as "○ (useful)." The user also evaluates the patent document whose problem the invention aims to solve is "suppression of the generation of grease buildup in □□ resin," i.e., the document from applicant company b, as "○ (useful)." Furthermore, the user rated the patent document in which the problem the invention aims to solve is "reducing the generation of grease in △△ resin," i.e., the document from applicant company C, as "× (not helpful, noise)."

[0074] Furthermore, in the example in Figure 7, the user rated a patent document from applicant company d as "○ (useful)" among patent documents describing a different technical problem than the one entered by the user. The user also rated a patent document from applicant company e as "× (not useful, noise)" among patent documents describing a different technical problem than the one entered by the user.

[0075] Similarly, Figure 8 shows an example of how users evaluated problems and means of solving them. In the example in Figure 8, the user rated the patent document in which the problem the invention aims to solve is "prevention of the generation of grease in XX resin," i.e., the invention in the document submitted by applicant A, as "○ (useful)." On the other hand, the user rated the problem that the inventions in the documents submitted by applicants B and C aim to solve as "× (not useful, noise)." Furthermore, the user rated the means for solving the problem in the documents submitted by applicants A and B as "○ (useful)." On the other hand, the user rated the means for solving the problem in the document submitted by applicant C as "× (not useful, noise)."

[0076] Furthermore, in the example in Figure 8, the user rated the patent documents describing problems different from the technical problems entered by the user as "○ (useful)" for the problems that the inventions in the documents of applicants D and E aim to solve. The user also rated the patent documents describing problems different from the technical problems entered by the user as "○ (useful)" for the means to solve the problems in the documents of applicant D. On the other hand, the user rated the patent documents describing problems different from the technical problems entered by the user as "× (not useful, noise)" for the means to solve the problems in the documents of applicant E.

[0077] These evaluations are input by operating the input device 230 and acquired by the input information acquisition unit 212a. The evaluations are then transmitted to the search server 100 by the transmission unit 212b.

[0078] As an example, Figure 9 shows a bubble diagram displaying multiple different problems A to C and multiple different solutions a to d extracted from individual patent documents included in the set of patent documents generated by the search server 100. Problems A to C are problems extracted by the problem extraction unit 112e of the search server 100, or problems grouped by the problem higher conceptualization unit 112m. Solutions a to d are problems extracted by the solution extraction unit 112f of the search server 100, or solutions grouped by the solution higher conceptualization unit 112n. At least one of problems A to C may be included in the technical field identified from the technical problem entered by the user, and may be a different problem from the technical problem entered by the user. In Figure 9, the size of each first bubble 700 indicates the number of patent documents. By displaying a bubble diagram of problems A to C and solutions a to d as shown in Figure 9, the user can associate related solutions and come up with new solutions.

[0079] In the bubble diagram of Figure 9, the user can click on individual first bubbles 700 by operating the input device 230 and making predetermined inputs. This displays a list of individual patent documents narrowed down by the problem and solution corresponding to each first bubble 700, similar to Figure 8. Furthermore, for each patent document narrowed down by problem and solution, the following may be shown: "Bibliographic information," "Applicant," "Patent classification," and "Use of the invention." In addition, the following may be shown: "Claims," ​​"Problem to be solved by the invention," and "Means for solving the problem." Furthermore, the entire "Abstract" or "Specification" may be shown. The bibliographic information may include information on the applicant, inventor, publication number, filing date, publication date, validity, and expiration of rights. Based on the information regarding expiration of rights, the user can recognize that the problem-solving means of expired patents are freely available.

[0080] The user can return from the list display in Figure 8 to the bubble diagram in Figure 9 by operating the input device 230 and making a predetermined input. Therefore, for example, the user can find the desired patent document by alternately referring to the screen in Figure 9 and the screen in Figure 8. Also, for example, the user can train the search server 100 to learn an algorithm for finding the desired patent document by alternately referring to the screen in Figure 9 and the screen in Figure 8 and evaluating the search results.

[0081] Furthermore, in Figure 9, the user can evaluate each of the grouped problems A to C and the grouped solutions a to d. The evaluation may be performed by inputting "○ (helpful)" or "× (not helpful, noise)" for each of the problems A to C and solutions a to d, as in Figure 6.

[0082] Furthermore, in Figure 9, the user can select a problem for discovering a new solution. In this case, the user selects the problem for discovering a new solution by clicking on the screen of the display device 220. Figure 10 shows the user selecting problem C in the bubble diagram of Figure 9. Although Figure 9 shows only one problem C selected, the user can select multiple problems simultaneously. In addition, as a problem for discovering a new solution, the user can select a technical problem that they initially entered, or a problem different from the technical problem they entered.

[0083] When problem C is selected as shown in Figure 10, the transmission unit 212b transmits problem C to the search server 100 to discover a new solution. The search unit 112c of the search server 100 searches for patent documents based on the technical problem C in a technical field different from the technical field specified in the search formula used to generate the set of patent documents. The document set generation unit 112d then generates a set of newly found patent documents based on the search results of the search unit 112c. The problem extraction unit 112e extracts problems from the patent documents included in the new set of patent documents. The solution extraction unit 112f extracts the solution means of the invention related to the patent documents included in the newly generated set of patent documents. The solution means extracted in this way may include new solution means not shown in Figure 9. The new set of patent documents is integrated with the previously generated set of patent documents. The search results may be displayed on the display device 220 in the same manner as in Figure 6. The search results are also displayed in a bubble diagram of a matrix including the new solution means. As an example, Figure 11 shows an example in which a bubble diagram of a matrix including a new solution means e is displayed for Figure 9. In Figure 11, the new solution e is displayed in a way that distinguishes it from the previously displayed solutions a to d. This allows for the extraction of new solutions in fields unrelated to the technical problem entered by the user. The newly extracted solutions can then be applied to solve the technical problem initially entered by the user. For example, if the technical problem entered by the user is related to the pharmaceutical field, solutions in the supplement field will be extracted, and these solutions can then be applied to solve the technical problem entered by the user.

[0084] In the bubble diagram of Figure 11, when the user clicks on the second bubble 710, they can view detailed information, including "bibliographic information," about the patent document describing the new solution e. The user can further view detailed information, including "claims" and "means for solving the problem." The new solution displayed here could be a candidate for a new invention that can solve the technical problem initially entered by the user.

[0085] Next, the time-series flow of processing performed by the information retrieval system 1000 will be explained. The basic sequence of processing performed by the information retrieval system 1000 is shown in Figure 12. First, the input information acquisition unit 212a of the user's terminal 200 acquires technical information (step S200). Next, the transmission unit 212b of the user's terminal 200 transmits the technical information to the search server 100 (step S202).

[0086] Next, the technical information acquisition unit 112a1 of the search server 100 acquires the technical information transmitted from the terminal 200 (step S100). Next, the search range specification unit 112b of the search server 100 specifies a predetermined location in the patent document as the search range for the patent document based on the attributes of the technical information (step S102). Next, the prompt creation unit 112i of the search server 100 creates a prompt (step S104). Then, the prompt transmission unit 112j transmits the prompt to the generation AI server 300 (step S106).

[0087] The generating AI server 300 generates a response to the prompt (step S300). Then, the generating AI server 300 sends the response to the prompt (step S302). Note that the processes in steps S104, S106, S300, and S302 may be performed at any time. For example, the processes in steps S104, S106, S300, and S302 may be performed when the problem is extracted in step S112.

[0088] Next, the search unit 112c of the search server 100 focuses its search on predetermined sections of the patent documents based on the attributes of the technical information and determines whether there are any descriptions in the patent documents that match or are similar to the technical information (step S108). As mentioned above, the search unit 112c may also perform a search based on new technical information obtained from the technical information. Next, the document set generation unit 112d of the search server 100 generates a set consisting of the searched multiple patent documents based on the search results of the search unit 112c (step S110). Next, the problem extraction unit 112e of the search server 100 extracts the problem that the invention related to that patent document aims to solve from each patent document searched by the search unit 112c (step S112). Next, the solution means extraction unit 112f of the search server 100 extracts the problem-solving means of the invention related to each patent document included in the set of patent documents from that patent document (step S114).

[0089] Next, the provision unit 112h of the search server 100 provides the user with the set of patent documents generated by the document set generation unit 112d (step S116). The set of patent documents is then transmitted to the user's terminal 200.

[0090] Next, the search result receiving unit 212c of the user's terminal 200 receives the set of patent documents transmitted from the search server 100 as the search result. Then, the display processing unit 212d of the user's terminal 200 displays the search results on the screen of the display device 220 (step S204). Next, the transmitting unit 212b of the user's terminal 200 transmits the evaluation of the search results entered by the user in the input device 230 to the search server 100 (step S206).

[0091] Next, the information acquisition unit 112a of the search server 100 acquires the user's evaluation of the search results transmitted from the terminal 200 (step S118). Then, the learning unit 112k of the search server 100 learns the algorithm for when the search unit 112c searches for patent documents based on the evaluation of the set of patent documents transmitted from the user's terminal 200 (step S120).

[0092] Furthermore, the transmission unit 212b of the user's terminal 200 transmits the problem for discovering a new solution, which the user has entered into the terminal 200, to the search server 100 (step S208). The search server 100 receives the problem for discovering a new solution. Then, the search unit 112c searches for patent documents based on the new problem in a technical field different from the technical field specified by the search formula used in the search in step S108 (step S121). Then, the document set generation unit 112d generates a new set of patent documents based on the search results of the search unit 112c (step S122). Next, the solution extraction unit 112f extracts solutions from the patent documents included in the new set of patent documents (step S124).

[0093] Next, the provision unit 112h of the search server 100 provides the user with the set of patent documents generated by the document set generation unit 112d (step S126). The set of patent documents is then transmitted to the user's terminal 200.

[0094] Next, the search result receiving unit 212c of the user's terminal 200 receives the set of patent documents transmitted from the search server 100 as search results. Then, the display processing unit 212d of the user's terminal 200 displays the search results on the screen of the display device 220 (step S210). As a result, as shown in Figure 11, search results including new solutions to the new problem are displayed.

[0095] Next, the transmission unit 212b of the user's terminal 200 sends the evaluation of the search results entered by the user into the input device 230 to the search server 100 (step S212).

[0096] Next, the information acquisition unit 112a of the search server 100 acquires the user's evaluation of the search results transmitted from the terminal 200. Then, the learning unit 112k of the search server 100 learns the algorithm for when the search unit 112c searches for patent documents based on the evaluation of the set of patent documents transmitted from the user's terminal 200 (step S128).

[0097] As described above, this embodiment makes it possible for those seeking to develop solutions to technical problems to obtain useful problem-solving methods with high accuracy by utilizing patent documents. Furthermore, this embodiment eliminates the need for those seeking to develop problem-solving methods to repeatedly search for information that matches their needs. This eliminates extremely cumbersome and time-consuming work. Therefore, those seeking to develop problem-solving methods can easily perform searches of patent documents aimed at developing such methods. Consequently, the reinvention of publicly known technology and the resulting burdensome or costly waste in the research and development field are suppressed.

[0098] 100 Search server 110 First control device 112 First processor 112a Information acquisition unit 112a1 Technical information acquisition unit 112b Search range specification unit 112c Search unit 112d Literature collection generation unit 112e Problem extraction unit 112f Solution means extraction unit 112g Priority setting unit 112h Provision unit 112i Prompt creation unit 112j Prompt transmission unit 112k Learning unit 112m Problem higher-level conceptualization unit 112n Solution means higher-level conceptualization unit 114 First memory 116 First communication interface 118 First communication bus 120 Storage device 200 Terminal 210 Second control device 212 Second processor 212a Input information acquisition unit 212b Transmission unit 212c Search result receiving unit 212d Display processing unit 214 Second memory 216 Second communication interface 218 Second communication bus 220 Display device 230 Input device 240 Speaker 300 Generation AI server 500 Communication network 700 First bubble 710 Second bubble 1000 Information retrieval system

Claims

1. An information retrieval system comprising: a technical information acquisition unit that acquires technical information including a technical problem; a search unit that searches for patent documents containing descriptions that match or are similar to the technical problem; a document set generation unit that generates a set of a plurality of the searched patent documents; a solution means extraction unit that extracts the solution means of the invention related to the patent document included in the set; a problem extraction unit that extracts the problem that the invention related to the patent document included in the set aims to solve; and a provision unit that provides the extracted solution means and the problem.

2. The information retrieval system according to claim 1, wherein the search unit identifies a technical field from the technical problem and further searches for patent documents that are included in the technical field and describe a problem different from the technical problem.

3. The information retrieval system according to claim 1 or 2, further comprising: a problem-level conceptualization unit for grouping the problems extracted by the problem-extraction unit; and a solution-level conceptualization unit for grouping the problem-solving means extracted by the solution-extraction unit.

4. The information retrieval system according to claim 3, wherein the providing unit provides a matrix-like map that associates the grouped problem-solving means with the grouped problems.

5. The information retrieval system according to claim 3 or 4, wherein the problem-level conceptualization unit further conceptualizes the grouped problems, and the solution-level conceptualization unit further conceptualizes the grouped problem-solving means.

6. The information retrieval system according to any one of claims 1 to 5, wherein the search unit searches for patent documents containing descriptions that match or are similar to the new technical problem entered by the user based on the provided problem-solving means; the document set generation unit generates a new set consisting of the newly retrieved plurality of patent documents; and the solution means extraction unit extracts problem-solving means of inventions related to the patent documents included in the newly generated set.

7. The information retrieval system according to any one of claims 1 to 6, further comprising a learning unit that learns an algorithm for when the search unit searches the patent documents based on the user's evaluation of the set of patent documents.

Images