Approximate patent retrieval method and device, electronic equipment and storage medium

Abstract

The invention provides an approximate patent retrieval method and device, electronic equipment and a storage medium, related technical keywords are extracted from to-be-retrieved patents to serve as retrieval keywords, approximate patents are retrieved in a patent database according to the retrieval keywords and retrieval strategies, and the retrieved approximate patents are recommended to a user after being sorted according to the approximation degree. According to the patent retrieval method and device, related technical features are screened from the to-be-retrieved patents for retrieval in an artificial intelligence mode according to the accurate retrieval requirement of the single to-be-retrieved patent file, the patent retrieval difficulty is lowered, the patent retrieval efficiency is improved, and the retrieval accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing and retrieval, in particular to a similar patent retrieval method and device, electronic equipment and storage medium. BACKGROUND

[0002] In the existing various patent retrieval platforms, the retrieval method is mainly that the user inputs the retrieval keywords, uses and, or and other logical connectors to form retrieval elements, and then selects time and date to perform patent literature retrieval. The above-mentioned retrieval method is biased towards keyword retrieval, and the relevance of the retrieved patents is general. In the later stage, the agent needs to spend a lot of time to perform relevant interpretation, and the threshold is high, and ordinary technical personnel have certain retrieval difficulty.

[0003] With the development of neural network and other technical means in recent years, more and more technologies such as patent evaluation and literature duplication checking are performed with the help of artificial intelligence, such as CN115841269A, a periodical dynamic evaluation method based on multi-dimensional index analysis, CN120104718A, a text retrieval enhancement generation method, device, medium and equipment, CN117033550A, a scientific and technological report achievement duplication checking method based on neural network feature recognition, and CN109726299A, an incomplete patent automatic indexing method. The above-mentioned patents have made improvements on literature duplication checking, patent indexing, periodical evaluation, etc. based on artificial intelligence text extraction technology, but there is no related technology disclosed in the field of patent duplication checking. SUMMARY

[0004] In order to make up for the defects in the above-mentioned patent duplication checking and retrieval, the present application provides a similar patent retrieval method, The steps include: The claims and the specification of the patent to be retrieved are respectively converted into a first technical file and a second technical file through a modification model; The technical keywords in the first technical file are extracted to form a first keyword group, and the technical keywords in the second technical file are extracted to form a second keyword group; Map all technical keywords in the first keyword group to the second keyword group according to the technical correspondence relationship; Determine whether the number of technical keywords in the first keyword group that have a mapping relationship with the second keyword group reaches a threshold value, if the threshold value is reached, the union of the first keyword group and the second keyword group is used as the retrieval keyword; If the threshold value is not reached, adjust the modification model to reconvert and extract the first keyword group and the second keyword group until the number of technical keywords in the first keyword group that have a mapping relationship meets the proportion requirement, and then obtain the retrieval keyword; According to the search keywords and the search strategy, approximate patents are searched in a patent database, and the searched approximate patents are ranked according to the degree of approximation and recommended to a user.

[0005] Further, the ranking process according to the degree of approximation includes: relevance ranking according to the coverage of the search keywords in the searched approximate patents from high to low, and sorting according to the frequency of the search keywords from high to low in the case of the same coverage.

[0006] Further, the approximate patent search device further comprises an identification process for the searched approximate patents, and the identification process comprises: adding the identification of the comprehensive approximate patent, the claim approximate patent and the specification approximate patent to each approximate patent according to the attribution of the search keywords in the first keyword group and the second keyword group.

[0007] Further, in the process of converting the claims and the specification of the patent to be searched into the first technical file and the second technical file through the modification model, the approximate patent search device further comprises the step of converting the patent to be searched into a technical recompilation and then converting it into the first technical file and the second technical file, and the technical recompilation process comprises: substituting the patent to be searched into a conversion model, converting the patent to be searched into a technical disclosure text, substituting the technical disclosure text into a generative artificial intelligence, generating a new patent application file from the generative artificial intelligence according to the technical disclosure text, and converting the claims and the specification of the new patent application file into the first technical file and the second technical file through the modification model.

[0008] Further, in the process of judging whether the proportion of the number of technical keywords in the first keyword group that have a mapping relationship with the second keyword group reaches a threshold value, the threshold value is 80%.

[0009] Further, the search strategy comprises a search time range, a classification number and a search region.

[0010] The application also provides an approximate patent search device, comprising: A conversion module is configured to convert the claims and the specification of the patent to be searched into the first technical file and the second technical file through the modification model. An extraction module is configured to extract the technical keywords of the first technical file to form the first keyword group, and extract the technical keywords of the second technical file to form the second keyword group. A mapping module is configured to map all the technical keywords in the first keyword group to the second keyword group according to the technical correspondence relationship. The search keyword generation module is used for judging whether the proportion of the number of technical keywords in the first keyword group that have a mapping relationship with the second keyword group reaches a threshold value, if the threshold value is reached, the union of the first keyword group and the second keyword group is taken as the search keyword, if the threshold value is not reached, the first keyword group and the second keyword group are marked as invalid this time, and an instruction is sent to the conversion module, the conversion module adjusts and modifies the model to reconvert and extract the first keyword group and the second keyword group; The search module is used for searching approximate patents in a patent database according to the search keyword and the search strategy, and recommending the searched approximate patents to the user after sorting the approximate patents according to the approximation degree.

[0011] The present application also provides a computer readable storage medium for storing instructions, when the stored instructions are run on a computer, the computer executes the steps of the approximate patent search method according to any one of the technical solutions.

[0012] The present application reduces the difficulty of patent search, improves the efficiency of patent search, and increases the accuracy of search by screening relevant technical features from the to-be-searched patent for search in an artificial intelligence manner and aiming at the accurate search requirement of a single to-be-searched patent file. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 The flowchart is for an embodiment of the method of the present application; Figure 2 The flowchart is for the technical recompilation of the to-be-searched file in an embodiment of the present application; Figure 3 The block diagram is for an embodiment of the system of the present application. DETAILED DESCRIPTION

[0014] The technical solutions of the present application will be described clearly and completely in combination with the drawings, obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0015] In the description of the application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings based on the orientation or positional relationship shown in the drawings, and are only for the purpose of facilitating the description of the application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application. In addition, the terms "first", "second", "third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0016] In the description of the application, it should be noted that unless otherwise specifically specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the application can be understood according to the specific circumstances. In the following description, specific details such as specific system structures, techniques are presented for the purpose of illustration, not for the purpose of limitation, so that the embodiments of the present application can be thoroughly understood. However, it should be clear to those skilled in the art that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits and methods are omitted to avoid unnecessary details that hinder the description of the present application. It should be understood that when used in the specification and the appended claims, the term "comprising" indicates the presence of the described features, whole, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, whole, steps, operations, elements, components and / or their collections. It should also be understood that the term "and / or" used in the specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0017] In the description of the present application, the reference to "one embodiment" or "some embodiments" and the like means that the specific features, structures or characteristics described in connection with the embodiment are included in one or more embodiments of the present application. Therefore, the statements "in one embodiment", "in some embodiments", "in other some embodiments", "in other some embodiments" and the like appearing in different places in the specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized. The terms "include", "contain", "have" and their variants mean "include but not limited to", unless otherwise specifically emphasized.

[0018] As a domestic professional patent retrieval platform provider, the applicant has been paying attention to optimizing the retrieval method of the retrieval platform during the retrieval process to improve the retrieval efficiency and accuracy of the user. At the present stage, more and more retrieval users on the applicant's platform are enterprise technical personnel, and in the process of using the retrieval platform, the summary of the patent scheme and the reading of the claims cannot reach the level of the agent, therefore, the applicant will embed the language large model into the new version of the retrieval platform on the basis of iterating the original retrieval platform, and through the platform, the patent technical scheme is analyzed and extracted, and the automatic retrieval for the input patent number or uploaded patent file is initially realized. In the latest retrieval platform developed by the applicant, the DeepSeek model has been nested, and on the basis of retaining the original self-provided expanded retrieval synonyms, a new retrieval scheme for patent text has been developed.

[0019] The method technical scheme combines Figure 1The application provides an approximate patent retrieval method. In the method, the approximate patent refers to a patent publication document that is close to a technical scheme of a searched patent in the patent field, which is similar to the definition of the comparison document in the examination guidelines but is limited to the published patent document. The steps include: converting the claims and the specification of the patent to be searched into a first technical document and a second technical document through a modification model respectively; in the embodiment, the first technical document and the second technical document are extracted as two separate technical documents to extract technical keywords, and the two technical documents are formed by using the same technical theme under two different description mode texts as the basis, the same modification model for extraction, the same technical idea but different technical details, especially the second technical document may have the same technical idea but different technical schemes, further expanding the extraction range of the technical keywords of the second technical document, and the technical keywords of the first technical document form a first keyword group, and the technical keywords of the second technical document form a second keyword group; in the extraction process of the technical keywords, the technical text composed of the content of the claims is used to extract the protected technical scheme, and the technical text composed of the specification is used to extract the technical scheme of the extended embodiment, and generally, the range recorded in the specification is greater than the protection range of the claims, so according to the extraction mode of the embodiment, the number of technical keywords of the second keyword group is greater than that of the first keyword group, and the second keyword group can improve the comprehensiveness of different word expression modes of related technical schemes. All the technical keywords in the first keyword group are mapped with the technical keywords of the second keyword group according to the technical correspondence; in the embodiment, the step is used to detect whether the extraction of the first keyword group and the second keyword group is accurate, and the first keyword group extracted from the technical scheme of the claims generally has consistent or approximate corresponding reference words in the specification. Since the above claims and the specification are independently converted into the first technical document and the second technical document respectively, and the writing mode of the claims and the specification is different, the first keyword group and the second keyword group may deviate from the technical keywords, so the coverage of the second keyword group to the first keyword group is determined by the mapping mode of the first keyword group, and then whether the extraction of the first keyword group and the second keyword group is accurate is judged. Whether the proportion of the number of technical keywords with the mapping relationship between the first keyword group and the second keyword group reaches a threshold value, if the threshold value is reached, the union of the first keyword group and the second keyword group is used as the retrieval keyword; if the threshold value is not reached, the modification model is adjusted to reconvert and extract the first keyword group and the second keyword group, until the number of retrieval keywords with the mapping relationship in the first keyword group meets the proportion requirement, and then the retrieval keywords are obtained; the approximate patents are retrieved in the patent database according to the retrieval keywords and the retrieval strategy, and the approximate patents retrieved are sorted according to the approximation degree and recommended to the user.In the present application, the modification model is a neural network model optimized for patent documents. It has the ability to recognize and convert specific subject-predicate-object and noun and transition words in Chinese and English patent documents. It also introduces professional vocabulary and synonymous vocabulary in various technical fields. It can effectively preserve the original technical solution connotation during the conversion of the first technical document and the second technical document. The modification model can be trained using a database of patent and technical fields, such as open-source deepseek, or using convolutional neural networks, recurrent neural networks, and Transformer models.

[0020] Based on the above embodiment, the sorting process according to the degree of approximation includes: sorting the relevance according to the coverage rate of the search keywords in the retrieved approximate patents from high to low, and sorting the search keywords according to the frequency of occurrence from high to low in the case of the same coverage rate.

[0021] Based on the above embodiment, the present embodiment is further improved. In the retrieval process, the present embodiment further includes a process of adding an identification to the retrieved approximate patents. The identification process includes: adding the identification of comprehensive approximate patents, claim approximate patents, and specification approximate patents to each approximate patent according to the attribution of the search keywords in the first keyword group and the second keyword group. In the present embodiment, the approximate patents are retrieved according to the first keyword group and the second keyword group, and the first keyword group and the second keyword group are obtained according to the claims and the specification of the patent to be searched. The marking method of this step can intuitively show that the approximate patents mainly disclose the claims and / or the specification of the patent to be searched, thereby facilitating users to quickly select and browse the disclosed documents according to the claims or the specification. For example, in patent infringement retrieval or invalidity retrieval, the patent documents with approximate claims can be quickly identified; and in patent technology association retrieval, the schemes with comprehensive approximation and specification approximation can also be quickly screened out, which is convenient and improves efficiency.

[0022] On the basis of one of the above embodiments, the present embodiment is further improved. In the process of converting the claims and the specification of the patent to be searched into the first technical file and the second technical file through the modification model respectively, it further includes the steps of recompiling the technology of the patent to be searched and then converting it into the first technical file and the second technical file. The technology recompilation process includes substituting the patent to be searched into the conversion model, converting the patent to be searched into a technical disclosure text, substituting the technical disclosure text into a generative artificial intelligence, generating a new patent application file from the generative artificial intelligence according to the technical disclosure text, and then converting the claims and the specification of the new patent application file into the first technical file and the second technical file through the modification model. In this embodiment, due to the differences in the writing habits and levels of agents during the writing process of the patent file, especially in some foreign language translation documents, the relevant description of the patent claims cannot be directly extracted into technical terms. Therefore, this step adopts a technology recompilation method to normalize the technology of the patent to be searched, so as to "dehydrate" the patent file and convert the first technical file and the second technical file into technical solutions as much as possible, facilitating the extraction of the first keyword group and the second keyword group. In this embodiment, since the large language model is now quite mature, different files can be generated through training of different databases. The conversion model of this embodiment is trained by a locally deployed large model, such as the open source DeepSeek-V3, in cooperation with relevant patent and disclosure databases, so that it can generate patent disclosure texts according to patent application files. The generative artificial intelligence of this embodiment can also fine-tune the database and production algorithm based on the conversion model to generate new patent application files from the disclosure texts. During the training process, technical terms and conventional terms in various fields are additionally introduced by the applicant, so that the patent application file is expressed in technical solutions as much as possible when it is converted into a disclosure text and a new patent application file, thereby maximizing the exclusion of different description methods caused by agents or translators, removing their excessive writing skills without changing the technical essence and hierarchical logic, and returning to a patent file with technical description. In this embodiment, the search of the patent to be searched is generally performed on the claims, mainly serving operations such as invalidation, FTO (Freedom to Operate, which mainly refers to the freedom of patent implementation), or other patent infringement, right confirmation, etc. Therefore, when the disclosure text generates a patent application file, the original claims of the patent to be searched are taken as control variables and brought into the generative artificial intelligence, so that the protection scope of the claims of the generated new patent application file is maximized to correspond to the original patent to be searched, facilitating the comparison of the claims in the later search and comparison process and the identification of the similarity of the claims and the specification in the display identification process.

[0023] On the basis of one of the above embodiments, the application is further improved, and in this embodiment, in the process of judging whether the proportion of the number of technical keywords in the first keyword group that have a mapping relationship with the second keyword group reaches a threshold value, the threshold value is 80%. In this step, through multiple actual tests, the threshold value of 80% is a relatively reasonable range. Under this threshold value, the search keywords can basically cover the technical solutions of the patents to be searched, and the expression of the first keyword group and the second keyword group is relatively accurate, and there is no situation that too many search words are introduced to cause that the patents are not strong in pertinence.

[0024] On the basis of one of the above embodiments, the application is further improved, and in this embodiment, the search strategy includes a search time range, a classification number and a search region.

[0025] The application also discloses a similar patent search device, which comprises: a conversion module, which is used for converting the claims and the specification of a patent to be searched into a first technical file and a second technical file through a modification model respectively; an extraction module, which is used for extracting technical keywords of the first technical file to form a first keyword group and extracting technical keywords of the second technical file to form a second keyword group; a mapping module, which is used for mapping all the technical keywords in the first keyword group to the second keyword group according to a technical correspondence relationship; a search keyword generation module, which is used for judging whether the proportion of the number of technical keywords in the first keyword group that have a mapping relationship with the second keyword group reaches a threshold value, if the threshold value is reached, taking the union set of the first keyword group and the second keyword group as search keywords, and if the threshold value is not reached, marking the first keyword group and the second keyword group as invalid this time and issuing an instruction to the conversion module to adjust the modification model to reconvert and extract the first keyword group and the second keyword group; and a search module, which is used for searching similar patents in a patent database according to the search keywords and a search strategy, and recommending the similar patents searched to a user after sorting the similar patents according to the degree of approximation.

[0026] In combination Figure 2 The application also provides an electronic device, which comprises a memory storing computer program instructions and a processor, wherein the computer program instructions are executed by the processor to implement the steps of the similar patent search method in any one of the above embodiments.

[0027] As Figure 3 The application also provides a computer readable storage medium, which is used for storing instructions, and when the stored instructions are run on a computer, the computer is caused to execute the steps of the similar patent search method in any one of the above embodiments.

[0028] The computer program includes computer program code, which can be in the form of source code, object code, executable code, or some intermediate form. The computer readable medium can include at least one of any entity or device capable of carrying the computer program code, a storage medium, a computer memory, a Read-Only Memory (ROM), a Random Access Memory (RAM), an electric carrier signal, a telecommunications signal, and a software distribution package, such as a disk or a compact disk etc. In some jurisdictions, depending on the legislation and patent practice, a computer readable medium can not be an electric carrier signal and a telecommunications signal.

[0029] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0030] Those skilled in the art can understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0031] In the embodiments provided in the present application, it should be understood that the disclosed apparatus / equipment and method can be implemented in other ways. For example, the apparatus / equipment controller embodiments described above are only schematic, for example, the division of the modules or units is only a logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0032] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e. they can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the present embodiment.

[0033] The above-described embodiments are only used to illustrate the technical solutions of the present application, but not limit them; although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A similar patent search method, characterized in that the steps include: Convert the claims and description of the patent to be searched into the first technical document and the second technical document respectively through the modification model; Extracting technical keywords from the first technical document to form a first keyword group, and extracting technical keywords from the second technical document to form a second keyword group; Mapping all technical keywords in the first keyword group with technical keywords in the second keyword group according to technical correspondences; Determine whether the proportion of technical keywords in the first keyword group that have a mapping relationship with the second keyword group reaches a threshold, and if so, use the union of the first keyword group and the second keyword group as the search keyword; If the threshold is not reached, the modification model is adjusted to reconvert and extract the first keyword group and the second keyword group until the number of technical keywords with a mapping relationship in the first keyword group meets the proportion requirement and the search keyword is obtained; Search for similar patents in the patent database based on search keywords and search strategies, sort the retrieved similar patents by similarity and recommend them to users.

2. The method according to claim 1, characterized in that The process of sorting by similarity includes: sorting the relevance from high to low based on the coverage of the search keywords in the retrieved similar patents, and sorting from high to low based on the frequency of occurrence of the search keywords under the same coverage.

3. The method according to claim 1, characterized in that The method also includes a process of adding identifications to the retrieved similar patents, wherein the identification process includes: adding identifications of comprehensive similar patents, claim similar patents and specification similar patents to each similar patent according to the attribution of the search keywords in the first keyword group and the second keyword.

4. The method according to claim 1, wherein The process of converting the claims and description of the patent to be searched into the first technical document and the second technical document respectively through the modification model also includes the step of technically recompiling the patent to be searched and then converting it into the first technical document and the second technical document. The technical recompilation process includes substituting the patent to be searched into the conversion model, converting the patent to be searched into a technical briefing text, substituting the technical briefing text into generative artificial intelligence, generating a new patent application document based on the technical briefing text by the generative artificial intelligence, and then converting the claims and description of the new patent application document into the first technical document and the second technical document through the modification model.

5. The method according to claim 1, wherein In the process of determining whether the proportion of technical keywords in the first keyword group that have a mapping relationship with the second keyword group reaches a threshold, the threshold is 80%.

6. The method according to claim 1, characterized in that The search strategy includes search time range, classification number and search area.

7. A similar patent search device, characterized in that: include: A conversion module, the conversion module is used to convert the claims and description of the patent to be searched into a first technical document and a second technical document respectively through the modification model; An extraction module, configured to extract technical keywords from a first technical document to form a first keyword group, and extract technical keywords from a second technical document to form a second keyword group; A mapping module, configured to map all technical keywords in the first keyword group to the second keyword group according to a technical correspondence relationship; A search keyword generation module is used to determine whether the proportion of technical keywords in the first keyword group that have a mapping relationship with the second keyword group reaches a threshold. If the threshold is reached, the union of the first keyword group and the second keyword group is used as the search keyword. If the threshold is not reached, the first keyword group and the second keyword group are marked as invalid and an instruction is issued to the conversion module, which adjusts the modification model to reconvert and extract the first keyword group and the second keyword group; The retrieval module is used to search for similar patents in the patent database according to the search keywords and the search strategy, and recommend the retrieved similar patents to the user after sorting them according to their similarity.

8. An electronic device comprising: a memory storing computer program instructions; A processor, when the computer program instructions are executed by the processor, implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store instructions. When the stored instructions are executed on a computer, the computer is caused to execute the steps of the method according to any one of claims 1 to 6.

Images