Information retrieval assistance system

Abstract

The present invention implements an information retrieval assistance system for retrieving or extracting many technical features by using products, drawings, and specifications. In order to protect a product of a host company, the appearance of the product or disassembled components thereof is imaged at the time of manufacture of the host company's undisclosed product, and inputs the image to a system. A user inputs the features of the product to the system in an interactive manner. The system lists candidates of the technical features on the basis of the input data, and creates a list.

Description

Information retrieval support system

[0001] One aspect of the present invention relates to an assistance system, an information processing system, and an information processing method.

[0002] Note that one embodiment of the present invention is not limited to the above technical field. The technical field of one embodiment of the invention disclosed in this specification and the like relates to an object, a method, or a manufacturing method. Alternatively, one embodiment of the present invention relates to a process, a machine, manufacture, or a composition of matter. Therefore, more specifically, examples of the technical field of one embodiment of the present invention disclosed in this specification include a semiconductor device, a display device, a light-emitting device, a power storage device, a memory device, a driving method thereof, or a manufacturing method thereof.

[0003] In recent years, the development of language models using neural networks has been actively pursued, with large-scale language models (LLMs) attracting particular attention. A large-scale language model is a natural language processing model trained using a large amount of data. A large-scale language model can realize, for example, a dialogue model that responds to user instructions. Non-Patent Document 1 discloses GPT-4 (Generative Pre-trained Transformer 4) (registered trademark) as a large-scale language model, and ChatGPT as a dialogue model.

[0004] The use of large-scale language models has significantly increased the capabilities of natural language processing models. However, as language models become larger, it is difficult to incorporate and operate language models in-house due to the equipment and cost involved. Therefore, one way to use language models is to use external services that provide language models.

[0005] Summary of ChatGPT / GPT-4 Research and Perspective Towards the Future of Large Language Models, Yiheng Liu et al. (Submitted on 4 Apr 2023, [online], Internet <URL: https: / / arxiv.org / abs / 2304.01852>

[0006] Patent laws vary from country to country. For example, the handling of double patenting, where two applications are filed by the same applicant on different dates, differs. Under Japanese patent law, if the claims of two applications are substantially identical, the later application will be rejected. On the other hand, U.S. patent law has a terminal disclaimer system, which allows both applications to be patented if documents are submitted to align the expiration date of the later application with the expiration date of the earlier application, even if the scope of protection of the two claims is substantially identical.

[0007] In this way, measures are taken to comply with the patent laws of each country, and the scope of patent claims that are valid in each country often differs depending on the examination standards of the country in which the examination is conducted. Therefore, it is advisable to prepare the scope of patent claims that are appropriate for each country.

[0008] It is also preferable for inventors to protect the products they invent and manufacture with patent rights.

[0009] Ideally, a specification would be drafted at the time of filing so that the applicant's desired scope of claims is permitted; however, due to time constraints or the judgment of the specification drafter, defects may arise in the specification or scope of claims at the time of filing.

[0010] When the specification at the time of filing contains multiple inventions (for example, feature A and feature B), if a patent is granted based on feature B only, which is stated in the claims without mentioning feature A, it is unclear whether the company will be able to adequately protect products that have feature A when manufactured.

[0011] Ideally, when a patent is filed, the claims would include feature A, feature B, or both.

[0012] The objective of this project is to realize an information search support system that searches or extracts many technical features using products, drawings, and specifications.

[0013] First, to protect their products, they take photos of the undisclosed exterior of their products or disassembled parts at the time of their manufacture and input the images into the system.

[0014] The user then interactively inputs product features into the system, for example, using a dialogue interface, which generates a textual description of the input image.

[0015] The system preferably has a table of technical features according to product classification, and can then refer to the table to proceed with a dialogue with the user.

[0016] The table can be created based on frequently occurring words that appear in the company's specifications for the above product classification. For example, if the product is a "secondary battery," frequently occurring words would be "positive electrode," "negative electrode," "separator," "electrolyte," etc.

[0017] If the system classifies a product as a battery from an image, it can prompt the user with questions like, "Are there any distinctive features about the positive electrode?" or, "Are there any distinctive features in the manufacturing method of that positive electrode?" to dig deeper into frequently occurring words.

[0018] The system refers to and compares the answer obtained from the user with the phrases described in the user's specification, and if phrase X and the answer are technically close, the system may prompt the user to explain, saying, "Is there any technical content related to phrase X shown in the user's specification?" The system can determine the closeness of the technical features by, for example, comparing the distributed representations of the phrases.

[0019] The system then lists candidate technical features based on this data.

[0020] Next, the user inputs the corresponding specification data (at the time of filing) into the system. Of course, the user can input the specification data (at the time of filing) into the system before taking an image. The system then searches whether the previously listed technical feature candidates are present in the specification. The system then outputs the search results, indicating whether the technical feature is present or not, to the user.

[0021] By comparing the system output with the photographed product, users can determine whether the technical features are sufficiently described in the claims at the time of filing. If they are not, users can use the system output as a reference to draft new claims. However, if users draft new claims, it is preferable to file a new divisional application using the terminal disclaimer system under U.S. patent law.

[0022] The invention disclosed in this specification is configured as an information retrieval support system that has one or more processors, accepts an image of a product structure and an explanatory text from a user, creates a prompt, inputs the prompt into a language model, obtains from the language model a list of technical features that identify the product structure, accepts text data of claims and patent specifications selected by the user, searches for whether the technical features are described in the text data, and displays a list of whether the technical features are present.

[0023] In the above configuration, the language model includes at least one of a large-scale language model, a generative language model, or a trained model. The large-scale language model (LLM) uses text data. The generative language model includes a large-scale visual language model (VLM) that processes image data in addition to text data.

[0024] In the above configuration, the explanatory sentence is text data obtained when the language model provides one or more question sentences to the user via the interactive interface, and the user provides the question sentence or answer sentence to the language model.

[0025] In each of the above configurations, the technical features include the external shape, the cross-sectional shape, or the parameter values.

[0026] The inventor can search for technical features, specifically new components, that are different from the scope of the patent claims at the time of filing from the description of the product (or prototype) invented and manufactured by the inventor at the time of filing the application, and suggest them to the user.The inventor can then use the terminal disclaimer system under U.S. patent law to file a new divisional application and protect his or her product through multiple applications.

[0027] Fig. 1 is a diagram illustrating a configuration of an information processing system according to one embodiment of the present invention. Fig. 2 is an example of a flow diagram illustrating one embodiment of the present invention. Fig. 3 is an example of a flow diagram illustrating one embodiment of the present invention. Fig. 4 is an example of a flow diagram illustrating one embodiment of the present invention. Fig. 5 is a diagram illustrating an example of an output from the information processing system.

[0028] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. However, the present invention is not limited to the following description, and it will be readily understood by those skilled in the art that various modifications can be made to the embodiments and details. Furthermore, the present invention should not be interpreted as being limited to the description of the embodiments shown below.

[0029] In this specification, the words "first" and "second" are used for the convenience of understanding the technical content or to identify each component. Therefore, the words "first" and "second" do not limit the number of each component. Furthermore, the words "first" and "second" do not limit the order of each component. Furthermore, the words "first" and "second" or identifying symbols used in this specification may not match the words or identifying symbols in the claims.

[0030] Embodiment 1 In this embodiment, an information processing system and an information processing method according to one embodiment of the present invention will be described with reference to FIGS. 1 to 3. An example of the information processing system is an information search support system.

[0031] <Configuration Example of Information Processing System> As shown in FIG. 1, the information processing system described in this embodiment includes information processing devices 10, 20, and 30. Each of the information processing devices 10, 20, and 30 includes at least one processor, i.e., hardware that executes machine language instructions within a computer, specifically a CPU (Central Processing Unit). Of course, this includes a multi-core processor, in which a single CPU includes multiple cores. The information processing system includes a system server that receives requests or instructions from other information processing devices, and one or both of the information processing devices 20 and 30 correspond to this system server. In this embodiment, the information processing device 20 corresponds to the system server. The information processing device 10 or 20 and the information processing device 30 are connected via a network 50, allowing for the transmission and reception of document data and the like. The information processing device 10 has a function of accepting information input by a user (information IN1 shown in FIG. 1) and transferring it to the information processing device 20. The information IN1 is image data or text data.

[0032] The information processing system described in this embodiment may also be configured so that a user can input documents or the like by directly operating the information processing device 11. As shown in Fig. 1, the information processing device 11 is connected to an information processing device 20 for transmission and reception. The information processing system may also be configured so that an image, document, or the like can be input using an information processing terminal having an imaging means 15 and a display unit 12.

[0033] The information processing device 20 also has a function of accepting information (information IN3 shown in FIG. 1) from the information processing device 30 and transferring it to the information processing device 10. The information IN3 is image data or text data.

[0034] The information processing device 30 also has a function of receiving information (information IN2 shown in FIG. 1) from the information processing device 20 and transferring information IN3 to the information processing device 20. The information IN2 is image data or text data.

[0035] An example of an information processing method used in the information processing system described in this embodiment will be described with reference to FIG.

[0036] [Information Processing Method] FIG. 2 is a flowchart illustrating an example of an information processing method according to one embodiment of the present invention.

[0037] The information processing method of one embodiment of the present invention is "started," and in step S101 of Fig. 2, an image of the exterior of a company's own product or a product under development is captured or photographed using the imaging means 15 or a scanner device of the information processing device 10 shown in Fig. 1. In step S102 of Fig. 2, image data is input into the system of the information processing device 20. The image data is not limited to the exterior of the product, and an electron microscope may be used to obtain a cross-sectional image.

[0038] Next, in step S103 of FIG. 2 , the user interactively inputs a description of the product. In step S104 of FIG. 2 , the user and the system exchange questions and answers. Then, in step S105 of FIG. 2 , candidate technical features are listed. For example, the information processing device 20 creates a prompt to explain the product structure, and the information processing device 30 outputs a list of the product's technical features, and the product features are displayed on the information processing device 20 or the information processing device 10 so that the user can confirm them. In this embodiment, the information processing device 30 uses LLM to list the technical features. LLM is a type of deep learning algorithm and a type of generative AI (Artificial Intelligence) specialized for natural language processing.

[0039] If the technical features listed in the LLM at this stage are few or insufficient for the user, steps S101 to S104 may be repeated, for example, by repeating questions or answers or adding images, to increase the number of technical features listed.

[0040] Then, in step S111 of Fig. 2, the information processing device 20 accepts the specification and claims. These specifications and claims use one or more post-filing data selected by the user. The target case, which is the post-filing data, is in a status where divisional filing is possible. Note that step S111 of Fig. 2 may be accepted in advance by the information processing device 20 or the information processing device 30 prior to step S101.

[0041] In step S121 of Fig. 2, a search is performed to determine whether the candidate technical features listed in step S105 are described in the specification and claims. The LLM may be queried for technical classification. The features may be vectorized in advance and a vector search may be performed. Then, in step S122 of Fig. 2, the results are presented to the user as list data.

[0042] The list data is displayed on the information processing device 20 or the information processing device 10, and the user creates a new claim. A technical feature is a character or character string of a predetermined unit. The user can create a claim by including one or more technical features as components of the claim. Therefore, the information processing system disclosed in this embodiment can also be said to be a support system for creating claims.

[0043] Then, in step S123 of Fig. 2, the information processing device 20 or 10 searches the specification to determine whether the basis for the claim created by the user is present. If the specification does not contain any basis, the process is restarted. If the process is restarted in step S152, steps S101 to S105 are repeated multiple times to increase the number of candidates for technical features. If the specification contains any basis, the basis is used as a claim for a divisional application, and preparations for the divisional application are made.

[0044] In this embodiment, technical features are extracted using image data of the exterior of a company's own products or products under development, and if the technical features are not included in the claims after filing, a list can be created that can be used as a reference when users prepare claims for a divisional application.Since the terminal disclaimer system can be used in U.S. patent law, it is useful to use this system when preparing a divisional application.

[0045] Second Embodiment While the first embodiment shows an example of a specification after filing, the present embodiment shows an example of using a specification before filing.

[0046] Figure 3 is a flow diagram illustrating the overall flow of an information processing method according to one embodiment of the present invention. Since Figure 3 is identical to Figure 2 up to step S105, detailed description thereof will be omitted here. After step S105, in which candidate technical features are listed in the information processing device 20, the user receives the specification and claims of the target case before filing via the information processing device 10 (step S111). The information processing device 20 then searches for the presence or absence of descriptions related to the candidate technical features (step S121). The results are output to the information processing device 10 and disclosed to the user.

[0047] In step S142, the user modifies or adds to the specification or claims if necessary. Since the result disclosed to the user is the presence or absence of statements in the specification before filing, the user can modify or add to the specification or claims if necessary. Therefore, the information processing system disclosed in this embodiment can also be said to be a specification drafting support system.

[0048] Through the above process, users can use image data of the exterior of their own manufactured or developed products to add claims, etc., and create more comprehensive specifications and claims. By using the system disclosed in this embodiment, users can complete specifications that can protect their own products or technologies developed in-house.

[0049] Also, in Figure 3, an example is shown in which step S111 is performed after step S105, but this is not particularly limited, and the specification and claims of the target case before filing may be received by the information processing device 20 before step S101.

[0050] This embodiment mode can be freely combined with other embodiment modes.

[0051] (Embodiment 3) In the first embodiment, an example of a specification (one application) after filing was shown, but in this embodiment, an example of using a company's own patent database (multiple applications) for which applications have already been filed will be shown.

[0052] 4 is a flowchart illustrating the overall flow of an information processing method according to one embodiment of the present invention. Note that since the steps up to step S105 in FIG. 4 are the same as those in FIG. 2, detailed description thereof will be omitted here.

[0053] After step S105 in which candidates for technical features are listed in the information processing device 20, the system searches the database of the information processing device 20 or the information processing device 30 for cases that describe the listed technical features (step S111).

[0054] Next, in step S121, the examination status of each country is searched to see whether the detected target case is under examination or not.

[0055] Then, each status or table of the searched target cases, i.e., the search results, are displayed on the display unit of the information processing device 10 so that the user can confirm them. An example of the search results output in step S141 is shown in FIG.

[0056] As shown in FIG. 5, the list output in step S141 can be displayed on the display unit 12 of the portable information terminal.

[0057] As shown in FIG. 5, the list 220 preferably has an area 222 indicating whether or not the technical feature is described in the specification, and an area 223 indicating whether or not the technical feature is described in the claim data.

[0058] In the example of table 220 shown in FIG. 5 , for technical feature A, technical feature B, technical feature C, technical feature D, technical feature E, technical feature F, and technical feature G possessed by the claim data (claims 1 to 5), the table shows that technical feature A is described only in the specification, technical feature B is described in claim 1, and technical feature C is not described in claim 1.

[0059] At the bottom of table 220 is a status column indicating whether the US application is pending.

[0060] The claim data described above may be displayed in table 220 for all target cases at once, but this may make table 220 difficult to read. Therefore, it is preferable to use a method of displaying the data for each target case, as shown in tabs 221a and 221b in Figure 5.

[0061] In step S143, the user adds claims or files a divisional application if necessary. The user looks at the output search results, and if the status is pending, the user can add claims, for example, when an amendment is possible. Of course, the claims are not new matter, but are assumed to be already stated in the specification. In this way, the user can add claims if necessary. Also, a divisional application can be filed, and the user can create the claims.

[0062] Your company's products or prototypes may be protected by a patent that has already been filed, so there may be no need to file a new application. Also, if the person in charge of writing the specification files an application thinking that it is different from your company's prior application, there is a risk that the application will be rejected due to double patenting with your company's prior application. If you can search in advance using this system, you can cut down on unnecessary application costs.

[0063] This embodiment can be called a first-in-one search system that searches using video data of a company's own products.

[0064] 10: Information processing device, 11: Information processing device, 12: Display unit, 15: Imaging means, 20: Information processing device, 30: Information processing device, 50: Network, 220: Table, 221a: Tab, 221b: Tab, 222: Area, 223: Area,

Claims

having one or more processors, It accepts images of the product's structure and a description from the user, creating a prompt and inputting the prompt into a language model; obtaining from the language model an enumeration of technical features that identify the structure of the product; Accepting text data of the claims and patent specification selected by the user; An information retrieval support system that searches for whether the technical features are described in the text data and displays the presence or absence of the technical features in a list.

2. The information retrieval support system of claim 1, wherein the language model includes at least one of a large-scale language model, a generative language model, or a trained model.

2. An information retrieval support system according to claim 1, wherein the explanatory text is text data obtained by the language model providing one or more question sentences to the user via an interactive interface, and the user providing a question sentence or an answer sentence to the language model.

2. The information retrieval support system according to claim 1, wherein the technical features include parameter values.

Images