Method for providing auxiliary text and related apparatus

Abstract

Provided in embodiments of the present application are a method for providing auxiliary text and a related apparatus. The method comprises: acquiring technical solution information inputted by a user; and calling a pre-trained large language model to generate auxiliary description text corresponding to the technical solution information, wherein the auxiliary description text is generated on the basis of the technical solution information and a retrieval result corresponding to the technical solution information, the retrieval result is used for characterizing a distinguishing feature of the technical solution information over a prior art document, and the auxiliary description text comprises auxiliary description content used for clarifying the distinguishing feature. Implementing the embodiments of the present application improves the usefulness of the auxiliary description text as a reference.

Description

Methods and related devices for providing auxiliary text

[0001] Cross-references to related applications

[0002] This application claims priority to Chinese patent application No. 202510377133.9, filed on March 27, 2025, entitled “Method and related apparatus for providing auxiliary text”, the entire contents of which are incorporated herein by reference. Technical Field

[0003] The embodiments described in this application relate to the field of computer technology, and in particular to a method and related apparatus for providing auxiliary text. Background Technology

[0004] Social development depends on technological progress, which generates a vast number of technological solutions. Those who possess these solutions often desire legal protection. To achieve this, they typically commission professionals to develop their solutions into legally compliant documents and submit them through legal channels. The goal is for these documents to pass review and secure legal protection.

[0005] Specifically, specialized documents include: a multi-layered scope of protection corresponding to the technical solution, and a specification for explaining the scope of protection. To clearly and explicitly explain the scope of protection, the specification is generally quite lengthy, requiring personnel with specialized knowledge to integrate multi-dimensional information related to the scope of protection (such as technical detail descriptions, product structure diagrams, etc.) based on their existing experience, thereby forming a specification that is appropriate to the scope of protection, logically consistent between its parts, and legally compliant. However, the explanatory text generated by these methods still suffers from low reference value. Summary of the Invention

[0006] In view of this, various embodiments of this application aim to provide a method and related apparatus for providing supplementary text, which can improve the referenceability of the supplementary explanatory text.

[0007] One embodiment of this application provides a method for providing auxiliary text, the method comprising: acquiring technical solution information input by a user; invoking a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information; wherein the auxiliary explanatory text is generated based on the technical solution information and search results corresponding to the technical solution information, the search results being used to characterize the differentiated features of the technical solution information from existing technical documents, and the auxiliary explanatory text containing auxiliary explanatory content for clarifying the differentiated features.

[0008] In some possible implementations, the technical solution information includes technical detail description information and feature clauses extracted from the technical solution information; after obtaining the technical solution information input by the user, the method further includes: extracting technical features from the feature clauses as first technical features; extracting technical features from the technical detail description information as second technical features; generating modification suggestions for the feature clauses based on the relevance between the first technical features and the second technical features; wherein the modification suggestions are used to display them through a visual interface to improve the relevance.

[0009] In some possible implementations, the technical solution information includes feature clauses extracted from the technical solution information; after obtaining the technical solution information input by the user, the method further includes: performing a formal check on the feature clauses according to a preset format specification corresponding to the feature clauses to obtain a formal check result; wherein the formal check result is used to display through a visual interface to indicate formal problems in the feature clauses that do not conform to the format specification.

[0010] In some possible implementations, the technical solution information includes feature clauses and technical detail description information extracted from the technical solution information; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: calling the pre-trained large language model to determine at least one relevant information among the classification number, technology type, technical problem, technical means, and technical efficacy corresponding to the technical detail description information; and generating auxiliary explanatory text corresponding to the technical solution information when at least one of the relevant information and the feature clauses is confirmed.

[0011] In some possible implementations, the technical solution information includes feature clauses extracted from the technical solution information; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: when an editing operation is detected on the feature clauses, calling the pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information based on the feature clauses after the editing operation and the technical solution information.

[0012] In some possible implementations, the method further includes: triggering text format monitoring of the specified text upon detecting an editing operation on the specified text, to obtain monitoring results; or triggering text format monitoring of the specified text after the specified text is generated, to obtain monitoring results; or triggering text format monitoring of the specified text after the specified text is updated, to obtain monitoring results; or triggering text format monitoring of the specified text upon detecting a user-triggered operation, to obtain monitoring results; wherein the monitoring results are used to display through a visual interface to characterize the text format problems existing in the specified text, and the specified text includes the auxiliary explanatory text.

[0013] In some possible implementations, the technical solution information includes feature clauses and technical detail descriptions extracted from the technical solution information; the auxiliary explanatory text includes multiple implementation descriptions; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: calling the pre-trained large language model to generate implementation descriptions based on the differentiating features between the technical solution information and existing technical documents indicated by the search results; or, calling the pre-trained large language model to generate implementation descriptions that conform to the technical solution layout based on the implementation schemes recorded in the technical detail description information; wherein, the technical solution layout is formed based on the calling relationship between the feature clauses.

[0014] In some possible implementations, the technical solution information includes feature clauses and technical detail description information extracted from the technical solution information; the auxiliary explanatory text includes specification drawings; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: calling a pre-trained large language model to generate specification drawings corresponding to the technical solution information; wherein, the specification drawings include at least one of the following: flowchart, structural diagram, formula diagram, chemical formula structural diagram, and circuit diagram; wherein, the flowchart is generated based on at least one of the feature relationships recorded in the feature clauses and the calling relationships between the feature clauses; the structural diagram is generated based on at least one of the structural information recorded in the feature clauses, the structural information recorded in the technical detail description information, and image information; the formula diagram, the chemical formula structural diagram, and the circuit diagram are generated based on the recording of the technical detail description information.

[0015] In some possible implementations, the method further includes: determining the drawing number corresponding to each of the specification drawings; adding corresponding labels to each part of the specification drawings according to the drawing number corresponding to the specification drawings; adjusting the auxiliary explanatory text according to the specification drawings containing the labels; wherein the adjusted auxiliary explanatory text contains textual description information and corresponding labels corresponding to each part of the specification drawings.

[0016] In some possible implementations, after calling the pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information, the method further includes: displaying the auxiliary explanatory text through a visual interface; and generating auxiliary explanatory text after the edit operation is detected when an edit operation is detected on the auxiliary explanatory text.

[0017] In some possible implementations, the process of generating the search results includes: calling a pre-trained large language model to generate a search expression corresponding to the technical solution information; performing a search process based on the search expression to recall existing technical documents related to the technical solution information; wherein the search process includes at least one of the following: a semantic search process, a classification number search process, and a keyword search process; and generating the search results based on the relevance between the existing technical documents and the technical solution information.

[0018] In some possible implementations, the technical solution information includes technical detail description information and feature clauses extracted from the technical solution information; the process of generating the search results includes: generating a search expression based on the feature clauses and the technical detail description information; and generating search results based on the prior art documents recalled by the search expression to characterize the differences between the technical solution information and the prior art documents.

[0019] In some possible implementations, after generating the search results, the method further includes: generating modification suggestions for the technical solution information based on the feature comparison results between the technical solution information represented by the search results and existing technical documents; wherein the modification suggestions are used to display them through a visual interface to broaden the differences between the technical solution information and existing technical documents.

[0020] In some possible implementations, the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: determining a target text review standard suitable for the technical solution information from multiple text review standards based on the language corresponding to the technical solution information; and calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information based on the template corresponding to the target text review standard.

[0021] In some possible implementations, the method further includes: exporting the auxiliary explanatory text to a specified file upon detecting a user-triggered export operation.

[0022] One embodiment of this application also provides an apparatus for providing auxiliary text, the apparatus comprising: an information acquisition module for acquiring technical solution information input by a user; and a text generation module for invoking a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information; wherein the auxiliary explanatory text is generated based on the technical solution information and search results corresponding to the technical solution information, the search results being used to characterize the differentiated features of the technical solution information from existing technical documents, and the auxiliary explanatory text containing auxiliary explanatory content for clarifying the differentiated features.

[0023] In some possible implementations, the technical solution information includes technical detail description information and feature clauses extracted from the technical solution information; the device further includes: a search expression generation module, used to generate a search expression based on the feature clauses and the technical detail description information; and a search module, used to generate search results based on existing technical documents recalled by the search expression to characterize the differences between the technical solution information and existing technical documents.

[0024] One embodiment of this application also provides a computer device, the computer device including a memory and a processor, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to implement the method as described above.

[0025] One embodiment of this application also provides a computer-readable storage medium storing at least one computer program that, when executed by a processor, can implement the method described above.

[0026] One embodiment of this application also provides a computer program product for implementing the method as described above.

[0027] In the various embodiments provided in this application, since technical solution information input by the user can be obtained, a pre-trained large language model can be invoked, and auxiliary explanatory text can be generated based on the technical solution information and its corresponding search results, and the search results are used to characterize the differentiated features of the technical solution information from existing technical documents, these differentiated features can be used as constraint inputs when generating the auxiliary explanatory text. This ensures that the auxiliary explanatory text includes auxiliary explanatory content to clarify the differentiated features, thereby increasing the explicit coverage of the differentiated features in the auxiliary explanatory text and reducing the risk of convergence with existing technical documents. In summary, compared to generating a specification solely based on user input, using search results as one of the bases for generating auxiliary explanatory text allows the auxiliary explanatory text to be as different from existing technical documents as possible, thus improving its referenceability. Attached Figure Description

[0028] Figure 1 is a schematic diagram of a system architecture for implementing a method for providing auxiliary text, according to one embodiment of this application.

[0029] Figure 2 is a flowchart of a method for providing auxiliary text according to one embodiment of this application.

[0030] Figure 3 is a schematic diagram of the technical solution information input interface provided in one embodiment of this application.

[0031] Figure 4 shows the auxiliary explanatory text display interface provided for one embodiment of this application.

[0032] Figure 5 is a flowchart of a method for providing auxiliary text according to another embodiment of this application.

[0033] Figure 6 is a schematic diagram of a device for providing auxiliary text according to one embodiment of this application.

[0034] Figure 7 is a schematic diagram of a device for providing auxiliary text according to another embodiment of this application.

[0035] Figure 8 is a schematic diagram of a computer device provided in one embodiment of this application. Detailed Implementation

[0036] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0037] In the description of the embodiments of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0038] Technological development drives social progress. To protect the legitimate rights and interests of technological innovation, and in accordance with legal requirements, technical solutions as technological achievements can seek legal protection in the form of specialized documents, such as patent documents and software copyright application documents. The following explanation uses patent documents as an example. Patent documents typically include the following parts: abstract, claims, description, and drawings. The description is usually the longest part. Generally, personnel with specialized knowledge (such as patent agents and intellectual property lawyers) use their experience, combined with one or more pieces of information such as technical disclosures, product structure diagrams, technical knowledge, and prior art documents, to draft a description that conforms to the multi-layered protection scope defined by the claims. At the same time, they must ensure that the description is logical, hierarchical, readable, and legally compliant.

[0039] Clearly, manually writing instruction manuals presents the following problems: it heavily relies on the experience of personnel with specialized knowledge, has low efficiency, and suffers from unavoidable formal errors (such as typos and inconsistencies). Existing solutions for automatically generating instruction manuals in related technologies all rely on user input. When the user input is existing content, the generated manual is not easily distinguishable from existing technical documents and has low reference value.

[0040] [Amended according to Rule 26, April 17, 2026] Patent document 1 provides a system and method for automatically generating patent files through semantic analysis, disclosing: an input module for receiving descriptive text input by a user, converting the descriptive text into a string for tagging, sending the string information, and recording the input language of the string information in temporary memory; a semantic analysis module for receiving the string information, analyzing and segmenting it through a natural language database, generating and sending semantic analysis results; an industry classification module for analyzing industry category classification codes based on the semantic analysis results, connecting to a patent database module to determine and generate at least one set of patent classification codes; a content learning module, which is a large language model that learns the corresponding content in the patent specification from the patent database module for different patent classification codes; and a content generation module for receiving and generating technical content based on the semantic analysis results and the content learning module to form a patent file.

[0041] The patent file (including the specification) of Patent Document 1 is generated based on the semantic recognition results of user input. As a text explaining and supporting the technical solution, if the specification is similar to existing technical documents, it will be difficult to highlight the innovativeness of the technical solution as an intellectual achievement. For users who need to refer to the patent file when drafting the specification, the patent file generated by Patent Document 1 based solely on user input may be similar to existing technical documents, resulting in insufficient reference value.

[0042] Patent document 2 (KR1020040029117A) provides an automated system and method for preparing patent application specifications and evaluating technologies. It discloses software that automatically generates graphical representations associated with the patent or technology being evaluated and outputs them in a user-viewable and modifiable format, as well as a computer with an input device through which at least one user inputs information related to components in the patent document, categorized by hierarchy and relevance. Thus, the user can input additional, more detailed information that forms the basis for the textual representation of the patent specification being drafted or the technology being evaluated. This additional information is associated with or linked to the graphical representation, for example, through automatic links or hyperlinks, allowing the user to switch between the graphical representation and a text-based detailed description of the patent-related components. Therefore, the portion of the patent application being drafted is formatted to a text-based structure suitable for submission as a patent application and can be further modified, altered, and / or standardized before the patent application is delivered in print or electronically.

[0043] The patent specification in the aforementioned patent document 2 is based on information input by the user. For users who need to refer to the specification drafted by the automated system to write the specification to be submitted to the relevant authorities, patent document 2 also has the problem that the specification generated solely based on user input may be similar to existing technical documents. Such a specification is not reliable enough as a reference text for users to write their own specifications.

[0044] Patent document 3 (JP2020095716A) provides a creative support device and method, disclosing an apparatus for storing the results of parsing the claims of a prior application; a receiving device for receiving information; an extraction means for extracting all or part of information similar to past applications; and a means for generating application documents, including claims generated based on the results of parsing the claims of past applications and a specification or drawings generated based on all or part of the information. …Users can upload technical documents such as invention proposals and technical drawings, still images such as photographs, moving images, and other files (hereinafter referred to as files) to replace or supplement keywords. Users can enter the path of the target file in the image upload area by referring to documents on their user terminal. Alternatively, users can directly enter the path of the target file in the image upload area. Alternatively, users can upload image files of technical documents such as invention proposals, instead of files.

[0045] Although Patent Document 3 is used to extract all or part of the information similar to past applications, its purpose is to determine the generation model / template for the claims and specification, and to generate claims and specifications similar to those of past applications. Patent Document 3 generates the specification based on user input (i.e., technical documents such as invention proposals, technical drawings, photographs, still images, and moving images). Patent Document 3 also suffers from the problem that a specification generated solely based on user input may resemble prior art documents, making it insufficiently reliable as a reference text for users drafting their own specifications.

[0046] Therefore, it is necessary to provide a method for providing supplementary text. This method acquires technical solution information input by the user and dynamically generates supplementary explanatory text based on the technical solution information and its corresponding search results. This allows the supplementary text to automatically adapt to the multi-layered protection scope requirements of the technical solution information. Compared to the traditional method relying on manual writing, this application combines real-time search results with the correlation analysis of technical solution information to generate supplementary explanatory text. This reduces the workload of manually compiling technical solution information, and relevant personnel can refer to the supplementary explanatory text to form a specification, which helps improve the efficiency of specification formation. Furthermore, compared to generating a specification solely based on user input, using search results as one of the bases for generating supplementary explanatory text allows the supplementary explanatory text to be distinguished from related technologies as much as possible, thus improving its reference value.

[0047] Please refer to Figure 1. In several embodiments provided in this application, the method for providing auxiliary text can be applied to a device for providing auxiliary text. The device for providing auxiliary text can be an electronic device with certain computing power and network access capabilities. This electronic device can be a desktop computer, laptop computer, tablet computer, or a server. In some embodiments, the electronic device can connect to the server via a network. The server can be a distributed server, including multiple processors, memory, network communication modules, etc., working together to achieve various functions. Alternatively, the server can also be a server cluster formed by several servers, possessing higher computing and data processing capabilities. With the development of science and technology, the server can also be implemented using new forms of technology, such as a new type of "server" based on quantum computing. Of course, in some embodiments, the device for providing auxiliary text can also be a program module running in an electronic device.

[0048] Specifically, the electronic device includes a processor, a memory, a display module, and a network access module for network communication. The processor is used to acquire technical solution information input by the user; the memory is used to store the technical solution information; the network access module is used to read the technical solution information from the memory and send it to the server; the server is used to generate auxiliary explanatory text corresponding to the technical solution information based on the technical solution information and the search results corresponding to the technical solution information, and send the auxiliary explanatory text to the electronic device; the network access module is used to receive the auxiliary explanatory text returned by the server and store it in the memory; and the display module is used to display the auxiliary explanatory text in the memory to the user.

[0049] In some implementations, the electronic device can acquire technical solution information input by the user, and generate auxiliary explanatory text corresponding to the technical solution information based on the technical solution information and the search results corresponding to the technical solution information, and then display the auxiliary explanatory text.

[0050] Please refer to Figure 2. One embodiment of this application provides a method for providing auxiliary text. The method for providing auxiliary text can be applied to an apparatus for providing auxiliary text. The method for providing auxiliary text may include the following steps.

[0051] Step S110: Obtain the technical solution information input by the user.

[0052] Step S120: Call the pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information; wherein, the auxiliary explanatory text is generated based on the technical solution information and the search results corresponding to the technical solution information, the search results are used to characterize the differentiated features of the technical solution information from existing technical documents, and the auxiliary explanatory text contains auxiliary explanatory content for clarifying the differentiated features.

[0053] In this embodiment, the device providing auxiliary text can provide a technical solution information input interface as shown in FIG3. The technical solution information input interface includes any number of technical solution information input areas, which can be used to receive technical solution information input by the user.

[0054] In this context, "technical solution information" refers to information used to describe / characterize / embody a technical solution. The information types corresponding to technical solution information can be varied, such as text, images, videos, and links, and this application embodiment does not limit this. The same type of technical solution information can correspond to different properties, which may include, but are not limited to: feature clauses extracted from technical solution information, detailed technical description information, product structure schematic information, project plans, etc. It should be noted that feature clauses extracted from technical solution information can refer to compliance text clauses such as claims and contract terms. Detailed technical description information can be text used to comprehensively describe the technical solution, such as technical disclosure documents or technical documents corresponding to the application. In other words, detailed technical description information can refer to original technical disclosure documents, experimental records, technical documents, product specifications, or other original technical data used to comprehensively describe the technical solution. Detailed technical description information is not extracted from the aforementioned original data. Detailed technical description information is the original source, while feature clauses are expressions extracted from the original source and organized for compliance / representation purposes.

[0055] Taking the claims as an example, the claims are a type of technical solution information that is characterized by clauses. The corresponding information types can be text, images, or links.

[0056] It should be further noted that multiple technical solution information of the same type (e.g., text) can be entered by the user. For example, technical disclosure documents and claims, both of which are text, can be received as technical solution information in the same / different technical solution information input areas.

[0057] Furthermore, the number of technical solution information input areas depends on actual needs, and this application embodiment does not limit this. In some possible implementations, the technical solution information input interface shown in Figure 3 provides an area addition function. When this function is triggered, technical solution information input areas can be added to the interface, providing users with more areas to input technical solution information. Each technical solution information input area can receive multiple types of technical solution information, or it can be used only to receive a single type of information. At least two technical solution information input areas can be displayed simultaneously in the interface, allowing users to easily compare and view different technical solution information to check the correlation between them.

[0058] If the technical solution information input interface shown in Figure 3 provides multiple technical solution information input areas, in some cases, these multiple input areas correspond to the same information type, so as to receive technical solution information of the same information type but different substantive content respectively. For example, the technical solution information input interface may display two technical solution information input areas simultaneously, used to display the technical disclosure and the claims respectively. In other cases, multiple technical solution information input areas correspond to different information types, so as to receive technical solution information of different information types but whose substantive content may be the same / different respectively. For example, the technical solution information input interface may display two technical solution information input areas simultaneously, used to display the claims and the product structure diagram respectively, and both are used to represent the same technical solution.

[0059] It should be further noted that the technical solution information input areas for receiving different types of information can correspond to different information receiving methods. Specifically, in some possible implementations, the technical solution information input area for receiving text can directly read the text entered by the user as technical solution information; the technical solution information input area for receiving images or videos can acquire images as technical solution information based on drag-and-drop operations, insertion operations, etc.; and the technical solution information input area for receiving links automatically links to the corresponding webpage and reads the content of the webpage as technical solution information.

[0060] After obtaining the technical solution information, the device providing auxiliary text can recall one or more prior art texts from the prior art document library that are most relevant to the technical solution information. By extracting the features corresponding to the prior art texts and the technical solution information, and performing correlation analysis on paired features, retrieval results can be obtained to characterize the differences between the technical solution information and the prior art. In some possible implementations, a retrieval strategy configuration function is also provided, which is used to configure conditions for constraining one or more dimensions of the retrieval scope, such as application date, database, etc.

[0061] Furthermore, the device providing auxiliary text can generate auxiliary explanatory text based on technical solution information and search results corresponding to the technical solution information. Alternatively, it can send the technical solution information to a corresponding device (e.g., a distributed server, a central server, etc.) to trigger the corresponding device to generate auxiliary explanatory text based on the technical solution information and search results corresponding to the technical solution information. The device can also receive the auxiliary explanatory text returned by the corresponding device. Subsequently, the device providing auxiliary text can display the auxiliary explanatory text so that users can intuitively browse it. As a reference for users in forming instruction manuals, the auxiliary explanatory text can help users form instruction manuals more efficiently.

[0062] The auxiliary explanatory text is used to integrate technical solution information based on the search results, forming at least one implementation process for a technical solution with logical reasoning and semantic explanation. The goal of generating the auxiliary explanatory text is to provide diverse implementation methods for multiple technical solutions involved in the technical solution information, thereby supporting the feasibility of the technical solutions. Specifically, the technical solution information and search results can be used as input to a pre-trained large language model to generate the auxiliary explanatory text.

[0063] In some implementations, the "pre-trained large language model" used to generate supplementary explanatory text can be any combination of one or more known large models or model types. Non-limiting examples include: general-purpose pre-trained large models (e.g., models based on autoregressive or encoder-decoder structures), domain-specific models (e.g., models pre-trained or adapted for biological, chemical, or engineering domains), and retrieval-augmented generation (RAG) models. Specific examples include: "pre-trained large language models" such as BERT, RoBERTa, and the GPT series. BERT (Bidirectional Encoder Representations from Transformers) is a bidirectional Transformer encoder pre-trained model used to capture contextual semantic representations. RoBERTa is a robustly optimized BERT approach that improves pre-training performance through larger datasets and training strategies. The GPT (Generative Pre-trained Transformer) series is a pre-trained generative model based on an autoregressive decoder, adept at continuous text generation and dialogue. The above examples are merely illustrative and should not be construed as limiting the scope of this application. In practical applications, appropriate models or combinations thereof can be selected based on the language, scale, and requirements of the technical solution information. Appropriate prompting engineering, input splicing, or search result integration can be used as inputs to generate auxiliary explanatory text.

[0064] For example, supplementary explanatory text can be displayed in the supplementary explanatory text display area shown in Figure 4. This text can be displayed in a format that the instruction manual should conform to, depending on the requirements of the organization that accepts / reviews specialized documents containing the instruction manual. In some possible implementations, Figure 4 may also provide various functions, such as export controls for exporting the supplementary explanatory text to a specified format (e.g., .doc) for further user use, editing controls for online editing of the supplementary explanatory text, sharing controls for sharing to specified contacts / communities, and integration controls for integrating technical solution information and supplementary explanatory text into a complete specialized document. The specialized document conforms to the file format and form required by the relevant organization.

[0065] In some possible implementations, to improve the referenceability of the supplementary explanatory text, it includes the following parts: technical field, background art, invention summary, description of drawings, and detailed embodiments. Corresponding large models are trained for different parts. When supplementary explanatory text needs to be generated, the technical solution information and search results are used as inputs to the corresponding large model for each part. This large model is then used to generate content that meets the formal requirements of the corresponding part. This content can constitute the final supplementary explanatory text to be displayed to the user. Among these, supplementary explanatory text generated by combining search results and technical solution information is more reliable. Specifically, taking background art as an example, since the technical problems recorded in the technical solution information may be inaccurate, possibly due to limitations in the understanding of existing technology, compared to background art generated solely based on the recorded technical solution information, this application, by combining search results and technical solution information, can summarize more accurate problems existing in the prior art. Taking a specific implementation as an example, if the description of the technical solution information is similar to that of existing technical documents, this application can combine the differences between the technical solution information and existing technical documents characterized by the search results to generate auxiliary explanatory content to clarify the differences, such as the definition of the differences, the scope of the differences, the specific technical means adopted by the differences, and the specific implementation of the differences, so as to highlight the core innovations contained in the technical solution information and explain and support them.

[0066] In this embodiment, differentiating features refer to the technical attributes or key technical points used to characterize and distinguish the technical solution information relative to the retrieved prior art documents. Differentiating features can reflect at least one element among the following: "differences," "innovations," or "limitations / relaxations" of the technical solution relative to the prior art. Differentiating features can be high-level solution differences or at least one of the following: specific new functions, new structures, new parameters, and combination relationships. Differentiating features can be identified by semantically or feature-wise comparing the technical solution information with the prior art documents. Differentiating features can be categorized hierarchically into broad-scope (e.g., technical field / problem / overall architecture) differentiating features, medium-scope (e.g., solution module / method / call relationship) differentiating features, and narrow-scope (parameter / material / limiting details) differentiating features. By extracting and comparing these elements from the retrieved prior art, both innovative points can be identified, and these points can be used as the main thread for generating supplementary explanatory text, thereby improving the referenceability and differentiated expression of the supplementary explanatory text.

[0067] Furthermore, in some possible implementations, it is understood that the aforementioned technical field, background technology, invention summary, description of drawings, and specific embodiments can be further subdivided into multiple parts. For example, the background technology can be further subdivided into an introduction, a description of prior art, and a description of the technical problem. Therefore, more refined models can be trained for each subdivided part to generate the corresponding part.

[0068] In this embodiment, by acquiring user-inputted technical solution information and dynamically generating supplementary explanatory text based on the technical solution information and its corresponding search results, the supplementary text can automatically adapt to the multi-level protection scope requirements of the technical solution information. Compared to the traditional method relying on manual writing, this application combines real-time search results with the correlation analysis of technical solution information to generate supplementary explanatory text, reducing the workload of manually sorting out technical solution information. Relevant personnel can refer to the supplementary explanatory text to form the instruction manual, which helps to improve the efficiency of instruction manual formation. Furthermore, compared to the method of generating the instruction manual solely based on user input, using search results as one of the bases for generating supplementary explanatory text allows the supplementary explanatory text to be distinguished from related technologies as much as possible, thereby improving the referenceability of the supplementary explanatory text.

[0069] In some implementations, the technical solution information includes detailed technical description information and feature clauses extracted from the technical solution information. After obtaining the technical solution information input by the user, the device providing auxiliary text can further extract technical features from the feature clauses as first technical features; extract technical features from the detailed technical description information as second technical features; and generate modification suggestions for the feature clauses based on the relevance between the first technical features and the second technical features; wherein the modification suggestions are displayed through a visual interface to improve the relevance.

[0070] In this embodiment, to generate more accurate auxiliary explanatory text, the device providing the auxiliary text can perform feature comparison on the feature clauses and technical detail description information, thereby providing modification suggestions based on the relevance represented by the feature comparison results. This allows users to modify the feature clauses for reference, and based on the modified feature clauses and search results, a more accurate auxiliary explanatory text can be obtained. Specifically, a rule-based semantic parsing model can be invoked to extract technical features from the feature clauses as the first technical feature. For example, from the feature clause "A convolutional neural network structure for image recognition, characterized in that it includes: ***", the technical features "convolutional neural network" and "image recognition" can be identified as the first technical feature. For example, the semantic parsing model can be a CRF (Conditional Random Field) model; the CRF model is a discriminative probabilistic graphical model used to model label sequences under given observation sequences, capable of capturing dependencies between labels, and commonly used in sequence labeling tasks (such as word segmentation and named entity recognition). Furthermore, deep learning models (such as BERT-Technology) can be invoked to perform semantic segmentation on technical detail descriptions, identifying specific technical implementation details (i.e., technical features) as second technical features. For example, "residual modules" and "attention mechanisms" from the extracted technical detail descriptions can be used as second technical features. BERT-Technology refers to a pre-trained language model based on BERT, fine-tuned or extended for the technology / patent domain, used to extract and understand technical terms and descriptions. BERT (Bidirectional Encoder Representations from Transformers) is a bidirectional pre-trained language representation model used to learn deep semantic representations of words by simultaneously considering left and right contexts.

[0071] Furthermore, the relevance between the first technical feature and the second technical feature can be determined as follows: First and second technical features are vectorized, and the cosine similarity between the vectorized representations is calculated as the relevance. Alternatively, a knowledge graph is constructed with preset technical features as nodes and co-occurrence frequencies between technical features as edges, and the semantic distance between the first and second technical features is calculated using a shortest path algorithm as the relevance. Alternatively, a preset multi-level scoring mechanism is used to evaluate the functional consistency, parameter matching, and application scenario overlap between the first and second technical features, and the results are weighted and fused to obtain the relevance between the first and second technical features. A lower relevance indicates a weaker connection between the feature clauses and the technical detail description information; that is, there may be inaccurate descriptions of the technical solution by the feature clauses.

[0072] Furthermore, since feature clauses are typically compliance texts abstracted from technical specification information, the technical specification information can more accurately describe the technical solution as an intellectual achievement compared to feature clauses. Therefore, after clarifying the relevance between the first and second technical features, modification suggestions for the feature clauses can be provided based on this relevance. These modification suggestions can reach users in any form, such as SMS, web pages, pop-ups, etc. The modification suggestions indicate the feature clauses that should be modified and the corresponding modification methods. Feature clauses include, but are not limited to, the following dimensions: terminology optimization (prompting users to replace term A with term B1 / B2 / B3); feature supplementation (prompting users to supplement feature C to cover key points in the technical specification information); logical adjustment (prompting users to adjust the order or logical relationship of technical features in the feature clauses to better conform to the implementation logic in the technical specification information); and scope limitation (prompting users to narrow / expand the scope of protection of the feature clauses).

[0073] In some implementations, the technical solution information includes feature clauses extracted from the technical solution information. After obtaining the technical solution information input by the user, the device providing auxiliary text can further perform a formal check on the feature clauses according to a preset format specification corresponding to the feature clauses, to obtain a formal check result; wherein, the formal check result is used to display through a visual interface to indicate formal problems in the feature clauses that do not conform to the format specification.

[0074] In this embodiment, to improve the referenceability of the supplementary explanatory text, the device providing the supplementary text can provide a formal checking function to reduce formal problems existing in the technical solution information, thereby improving the referenceability of the supplementary explanatory text generated based on the technical solution information. Since technical solution information can contain a variety of information, it is understood that any piece of information contained within the technical solution information (e.g., claims) may correspond to a matching specification (e.g., the Implementing Regulations of the Patent Law), or may not correspond to any specification (e.g., a technical disclosure document may not correspond to any specification; however, if a user-specified specification exists, then the technical disclosure document may correspond to it).

[0075] For the claims / contractual clauses that serve as feature clauses, there are corresponding format specifications to constrain their writing style, ensuring they meet industry / legal / client requirements. These format specifications limit the various dimensions of conditions that feature clauses must satisfy (e.g., feature clauses should have numbers, and there should be no multiple references between feature clauses). In some possible implementations, when feature clauses are received, each clause can be compared against the format specifications using a predefined rule base to identify formal issues; natural language processing algorithms can be used to perform syntactic analysis on the feature clauses to identify grammatical errors or semantic ambiguities; models such as BERT-Technology can be used to analyze inconsistencies in technical terminology between feature clauses; and a reference relationship graph can be constructed, with nodes representing feature clauses and edges representing reference relationships, allowing graph traversal algorithms to verify the existence of multiple references or invalid references.

[0076] Furthermore, the identified formal issues can be provided to users as formal inspection results, allowing them to edit the feature clauses based on these results. The formal inspection results include, but are not limited to, information in the following fields: type of formal issue, location of the formal issue, description of the formal issue, and suggested modifications. After the feature clauses are edited / re-uploaded online, the supplementary explanatory text can be dynamically updated accordingly to ensure that the supplementary explanatory text and the feature clauses maintain correspondence and consistency.

[0077] In some implementations, the technical solution information includes detailed technical description information and feature clauses extracted from the technical solution information. The step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: calling the pre-trained large language model to determine at least one relevant information among the classification number, technology type, technical problem, technical means, and technical efficacy corresponding to the detailed technical description information; and generating auxiliary explanatory text corresponding to the technical solution information when at least one of the relevant information and the feature clauses is confirmed.

[0078] In this embodiment, if the received technical solution information is a feature clause and a detailed technical description, a confirmation function can be provided to offer corresponding supplementary explanatory text upon user confirmation, thereby improving the accuracy of the supplementary explanatory text. Examples of this include feature clauses being claims and detailed technical descriptions being technical disclosure documents.

[0079] Once the technical disclosure document is obtained, a classification number prediction model combining rule-based and deep learning (e.g., the BERT-IPC model) can be invoked to automatically match the corresponding classification number (e.g., IPC classification number, CPC classification number) based on the technical field and technical characteristics in the document. Furthermore, a pre-trained domain classification model (e.g., FastText, TextCNN) can be used to categorize the technical disclosure document into the corresponding technical field (e.g., mechanical, electrical, chemical, biological, etc.). Additionally, a pre-trained semantic recognition model can be used to determine the corresponding technology type (e.g., method, equipment). A semantic parsing model (e.g., the BERT-Problem model) can be invoked to extract the technical problem to be solved from the technical disclosure document. Furthermore, a semantic segmentation algorithm (e.g., SpanBERT) can be used to identify the necessary technical means for solving the technical problem from the technical disclosure document. Finally, an effect association model (e.g., causal reasoning model) can be used to analyze the technical effectiveness brought about by the necessary technical means. The BERT-IPC model is a BERT-based fine-tuned classification model used to automatically predict or match the corresponding IPC (International Patent Classification) classification number based on the technical text. FastText is an efficient text representation and classification method that uses word vector averaging and a linear classifier, making it suitable for large-scale, fast training. TextCNN is a convolutional neural network method for text classification that extracts local features and performs classification by applying multiple convolutional kernels and pooling operations to word vectors. The BERT-Problem model is a specialized model based on BERT fine-tuning, used to identify or extract "technical problems to be solved" and their descriptions from technical documents. SpanBERT is an improved pre-trained model of BERT that enhances performance on tasks such as cross-word segment extraction and coreference by masking and predicting continuous text segments and optimizing boundary representations.

[0080] Furthermore, the system can display the aforementioned classification number, technology type, technical problem, technical means, and technical effect for user confirmation, and supports online modification of any of these items. If the classification number, technology type, technical problem, technical means, or technical effect is confirmed, supplementary explanatory text can be generated directly based on this information.

[0081] Alternatively, if the classification number, technology type, technical problem, technical means, or technical effect is confirmed, a display interface for the claim terms can be provided. If the information in the online editing interface is confirmed, supplementary explanatory text can be generated based on the content in the online editing interface and the aforementioned classification number, technology type, technical problem, technical means, and technical effect. Furthermore, the online editing interface can also provide elements extracted from the claim terms; these elements are keywords used to solve the technical problem in the claim terms, or words present in multiple claim terms.

[0082] In some embodiments, the technical solution information includes feature clauses extracted from the technical solution information; the device for providing auxiliary text may also provide an editing function for the feature clauses; and the device for providing auxiliary text may also provide auxiliary explanatory text corresponding to the technical solution information based on the edited feature clauses and the technical solution information when the editing function is triggered (e.g., when an editing operation for the feature clauses is detected).

[0083] In this embodiment, the device providing auxiliary text can provide editing functions for feature clauses. These editing functions support adjusting the specific content of feature clauses, the relationship between feature clauses, deleting feature clauses, and adding new feature clauses, so that users can make personalized adjustments to the feature clauses that serve as the basis for generating auxiliary explanatory text. This helps to improve the referenceability of the auxiliary explanatory text.

[0084] In this embodiment, editing operations on the feature clauses refer to modifications made by the user to the feature clauses. Editing operations on the feature clauses can be categorized into three types based on their scope of impact: Large-scale editing operations, such as changing the overall structure or scope of protection of the claims or feature clauses (e.g., adding / deleting entire claims or groups of claims, merging / splitting claims, elevating / degrading dependent claims, or uploading complete replacement text), typically trigger a full re-search and regeneration; medium-scale editing operations, such as making content adjustments to single or multiple feature clauses, changing the technical meaning or implementation support (e.g., adding or deleting key features, replacing terms, adjusting parameter ranges or logical connections), typically triggering local searches and target paragraph updates; and small-scale editing operations, such as changes to the expression or formatting that do not change the technical substance (e.g., correcting typos, adjusting punctuation / numbering / layout, minor adjustments to synonyms), typically only triggering format consistency or numbering linkage checks.

[0085] In this embodiment, "detecting an editing operation on the feature clause" can refer to recognizing a change in the text content extracted as the feature clause. Such changes include at least one of the following forms: modification, addition, deletion, rearrangement, or replacement with a new document. These changes can be performed directly by the user in the editor or submitted via upload / interface. After the detection is triggered, the corresponding auxiliary explanatory text can be regenerated or updated using the edited feature clause along with the original technical solution information as input.

[0086] In some embodiments, the device providing auxiliary text can provide a continuous text format monitoring function; wherein the monitoring function is automatically triggered when the editing function is triggered, or automatically triggered after the auxiliary explanatory text is generated, or automatically triggered after the auxiliary explanatory text is updated, or passively triggered when a user-triggered operation is detected. The method further includes: triggering text format monitoring of the specified text upon detecting an editing operation on the specified text to obtain a monitoring result; or triggering text format monitoring of the specified text after the specified text is generated to obtain a monitoring result; or triggering text format monitoring of the specified text after the specified text is updated to obtain a monitoring result; or triggering text format monitoring of the specified text upon detecting a user-triggered operation to obtain a monitoring result; wherein the monitoring result is displayed through a visual interface to characterize the text format problems existing in the specified text, the specified text including the auxiliary explanatory text. The specified text may also include various texts such as claims and technical disclosure documents.

[0087] In this embodiment, the device providing auxiliary text can provide a monitoring function. This monitoring can be used to monitor the text format of various texts, such as claims, auxiliary explanatory text, and technical disclosure documents. Different types of text are adapted to different text formats. Specifically, the monitoring can call the corresponding text format based on the currently monitored text type and perform format checks on the corresponding text. Furthermore, the monitoring can be automatically triggered based on specified events (e.g., editing function triggered, auxiliary explanatory text generated, auxiliary explanatory text updated) or passively triggered based on user-triggered operations (e.g., clicking the monitoring function control, voice input operation to start the voice control monitoring function), thereby providing monitoring results characterizing text formatting problems for user reference and helping users save time on text formatting checks.

[0088] In some implementations, the technical solution information includes feature clauses and technical detail descriptions extracted from the technical solution information; the auxiliary explanatory text includes multiple implementation descriptions. The step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information may include: calling the pre-trained large language model to generate implementation description information based on the differentiating features between the technical solution information and existing technical documents indicated by the search results; or, calling the pre-trained large language model to generate implementation description information conforming to the technical solution layout based on the implementation schemes recorded in the technical detail description information; wherein the technical solution layout is formed based on the calling relationships between the feature clauses.

[0089] In some implementation methods, the implementation description information may refer to a set of textual or structured descriptions that specifically explain how to implement the technical solution. The implementation description information may include at least one of the following: implementation steps, device structure or module configuration, parameter range, specific embodiments, and their expected technical effects. The implementation description information can be used to improve the feasibility and understandability of the technical solution. This implementation description information can be generated based on the differentiating features from existing technologies indicated in the search results (to highlight and explain the specific implementation of the innovation points), or it can be directly generated based on the descriptions in the technical specifications and organized according to the technical solution layout. This layout can be determined by the calling relationships between the feature clauses.

[0090] In this embodiment, since the number of technical solutions defined by the feature clauses and technical detail description information can be multiple, and each technical solution can correspond to multiple implementation methods, in order to obtain auxiliary explanatory text supporting the feature clauses, the device providing the auxiliary text can call a semantic comparison model (e.g., the BERT-Diff model) to analyze the differences between the technical solution information and existing technical documents, identify the innovative features and distinguishing features present in the technical solution information, and generate progressive implementation description information accordingly. The progressive implementation description information includes a basic implementation method, an optimized implementation method, and a hardware-supported implementation method. The implementation methods included in the progressive implementation description information can be implementation methods already recorded in the technical solution information, or implementation methods extended based on the recorded technical solution information to support / supplement the technical solution. The BERT-Diff model is a difference detection model based on BERT fine-tuning, used to perform semantic comparison between the technical solution information and existing technical documents, automatically identifying and marking innovative features and distinguishing points.

[0091] The device providing auxiliary text can also determine the technical solution layout based on the relationship between the feature clauses and the identified technical problems, technical means, and technical effects. The technical solution layout is used to represent the parallel / progressive relationship between multiple technical solutions. Furthermore, it can generate implementation description information that conforms to the technical solution layout based on the implementation scheme recorded in the technical specification description information, so that each technical solution in the technical solution layout has corresponding implementation description information sufficient to support it.

[0092] For example, firstly, technical features that characterize the key technical points can be extracted from the feature clauses and technical details description information; based on these features, a search query is generated and relevant prior art documents are recalled. This step can employ publicly available rule parsing, semantic annotation, or vectorized representation. Then, a semantic comparison is performed between the recalled prior art and the information in this case (e.g., based on vector similarity, knowledge graph paths, or semantic comparison models exemplified in this specification) to identify a set of differentiated features and generate explanatory points (such as definitions, value ranges, and implementation methods) for each differentiated feature. Secondly, the differentiated features can be mapped to pre-set or dynamically generated implementation templates; for each key difference, at least one basic implementation is generated, and optimized or hardware / software supported implementations are generated as appropriate to ensure that the higher-level concepts and lower-level embodiments in the claims receive multi-level support. Furthermore, by analyzing the calling relationships between features, a technical solution layout (parallel or progressive structure) can be constructed, and the above-generated implementations can be organized according to the layout to ensure that each parallel branch and each progressive level has complete implementation support as much as possible. Finally, the generated implementation description can be checked for consistency, coverage, and format compliance (including linkage verification with the figure numbers); if insufficient coverage or logical inconsistencies are found, modification suggestions can be triggered, or the implementation description can be updated after user confirmation.

[0093] As can be seen, the implementation method of this application is driven by differentiated features, naturally organizing the description information of the implementation method around "where it differs, how it is implemented, and what effects it brings," thus forming auxiliary explanatory text that clarifies the differences in features. This makes the current technical solution clearer and distinguishable from the retrieved prior art documents, thereby enhancing the value of the auxiliary explanatory text as a reference. By generating at least one basic implementation method for each key differentiated feature and supplementing and optimizing / supporting implementation methods, it ensures that the superior concepts in the claims are supported by multiple embodiments, reducing the risk of being questioned during the examination stage due to "insufficient support / insufficient implementation teaching." The technical solution layout organizes the implementation methods based on feature call relationships, which can reduce logical breaks or contradictions between embodiments, thereby making the auxiliary explanatory text have a better logical closed loop between steps and embodiments. Integrating retrieval, comparison, implementation method generation and verification into a closed loop reduces the workload of manual repetitive comparison and sorting out differences, improves writing / drafting efficiency, and reduces low-level formatting or consistency errors. This allows the auxiliary explanatory text to adapt to the multi-level protection scope requirements of technical solution information.

[0094] In some embodiments, the technical solution information includes feature clauses and technical detail description information extracted from the technical solution information; the auxiliary explanatory text includes drawings; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: calling a pre-trained large language model to generate the drawings corresponding to the technical solution information; wherein, the drawings include at least one of the following: flowchart, structural diagram, formula diagram, chemical formula structural diagram, and circuit diagram; wherein, the flowchart is generated based on at least one of the feature relationships recorded in the feature clauses and the calling relationships between the feature clauses; the structural diagram is generated based on at least one of the structural information recorded in the feature clauses, the structural information recorded in the technical detail description information, and image information; the formula diagram, the chemical formula structural diagram, and the circuit diagram are generated based on the technical detail description information.

[0095] In this embodiment, to further reduce reliance on manual labor and save time spent manually drawing the accompanying drawings, thereby improving the efficiency of generating supplementary explanatory text, the device providing supplementary text provides accompanying drawings corresponding to the supplementary explanatory text. These accompanying drawings can be of various types, such as flowcharts, structural diagrams, formula diagrams, chemical formula structural diagrams, and circuit diagrams mentioned earlier. However, this does not mean that this application is limited to these types; in practical applications, this application can be used to generate various types of accompanying drawings.

[0096] Specifically, the device providing auxiliary text can invoke a semantic parsing model (e.g., SpanBERT) to extract technical features and their logical relationships from the feature clauses, and construct a call relationship diagram based on the extracted technical features and their logical relationships. Based on the call relationship diagram, a flowchart conforming to a flowchart style is generated for user reference or invocation. Furthermore, the device providing auxiliary text can extract structure-related descriptions from the feature clauses and technical specification information, extract accompanying drawings and image information (e.g., CAD drawings) from the technical specification information, and then invoke drawing tools to generate a structure diagram based on the extracted multi-dimensional information, ensuring that the logical relationships between modules are clear and correct. Additionally, the device providing auxiliary text can use a rendering engine to render the mathematical formulas recorded in the technical specification information as formula diagrams, use chemical structure drawing tools (e.g., RDKit, ChemDraw) to draw the molecular structures recorded in the technical specification information as chemical formula structure diagrams, and invoke circuit design tools (e.g., KiCad, Eagle) to generate circuit diagrams from the circuit connection relationships recorded in the technical specification information. Here, CAD drawings are engineering or product structure diagrams generated by computer-aided design software, used to represent components, dimensions, and assembly relationships. RDKit is an open-source cheminformatics toolkit for molecular representation, property calculation, and chemical reaction processing. ChemDraw is a commercial chemical drawing software used to draw chemical structural formulas and reaction formulas and generate publication-quality images. KiCad is an open-source electronic design automation (EDA) toolkit primarily used for schematic drawing and printed circuit board (PCB) layout design. Eagle is a commercial EDA software that provides schematic editing and PCB design functions and is commonly used in electronic product development. In some embodiments, the device providing auxiliary text can also determine the drawing numbers corresponding to each of the specification drawings; add corresponding labels to each part of the specification drawings according to the drawing numbers corresponding to the specification drawings; adjust the auxiliary explanatory text according to the specification drawings containing the labels; wherein the adjusted auxiliary explanatory text contains textual description information corresponding to each part of the specification drawings and the corresponding labels.

[0097] In this embodiment, to conform to the format specifications that the accompanying drawings should meet, the device providing auxiliary text can automatically add corresponding numbers to each part of the accompanying drawings, and avoid the situation where different parts are labeled with the same number. This saves time on manual numbering and improves the usability of the accompanying drawings. Specifically, each drawing can be assigned a unique drawing number according to the order of the accompanying drawings or user-specified rules, thereby identifying the key content of each part of the drawing and automatically generating corresponding numbers. Furthermore, a consistency check algorithm is used to verify whether the numbers match the content of the drawing, ensuring that there are no omissions or multiple labeling.

[0098] Based on this, the supplementary explanatory text can be updated according to the labeled illustrations in the instruction manual. For example, insert "Label 101 in Figure 1 represents the convolutional layer, whose function is feature extraction" into the supplementary explanatory text so that the supplementary explanatory text contains all the labels and contents marked in the illustrations in the instruction manual, thereby improving the correspondence between the supplementary explanatory text and the illustrations in the instruction manual.

[0099] In some implementations, after invoking a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information, the method of providing auxiliary text may further include: displaying the auxiliary explanatory text through a visual interface; and generating auxiliary explanatory text after the edit operation is detected. In other words, the device for providing auxiliary text may also provide an editing function for the auxiliary explanatory text; and generate auxiliary explanatory text after the edit operation is triggered (e.g., when an edit operation is detected).

[0100] In this embodiment, after providing supplementary explanatory text to the user, the user may have a need to modify the supplementary explanatory text. Therefore, the device providing the supplementary text also has an editing function. When the editing function is triggered, the current supplementary explanatory text content is automatically loaded into the editor, and the editable area is highlighted. The editor supports insertion, deletion, modification, and formatting operations. After the editing is detected to be complete, the original supplementary explanatory text is dynamically updated, and the updated supplementary explanatory text undergoes consistency verification and compliance checks to ensure that the updated supplementary explanatory text corresponds to the technical solution information and / or the accompanying drawings, and that the updated supplementary explanatory text complies with the requirements of relevant specifications.

[0101] In some embodiments, the process of generating the search results includes: invoking a pre-trained large language model to generate a search expression corresponding to the technical solution information; executing a search process based on the search expression to recall prior art documents related to the technical solution information; wherein the search process includes at least one of the following: a semantic search process, a classification number search process, and a keyword search process; and generating the search results based on the relevance between the prior art documents and the technical solution information. The device providing auxiliary text can also invoke a pre-trained large language model to generate a search expression corresponding to the technical solution information; execute a search process based on the search expression to recall prior art documents related to the technical solution information; wherein the search process includes at least one of the following: a semantic search process, a classification number search process, and a keyword search process; and generating the search results based on the relevance between the prior art documents and the technical solution information.

[0102] In this embodiment, to generate supplementary explanatory text that effectively distinguishes itself from existing technologies for user reference, the device providing supplementary text can generate a search query corresponding to one or more pieces of information in the technical solution information (e.g., claims and / or technical disclosure). Specifically, based on the technical solution information and a prompt instruction to instruct a large model to generate the search query, a pre-trained search query generation large model is invoked to trigger the large model to generate and display the search query. The search query is a logical statement that associates logical features through logical relationship identifiers (e.g., AND, OR, etc.), containing at least one of the following logical features: features in the technical solution information, and other features related to the technical solution information (e.g., synonymous features, near-synonymous features, etc.). In some embodiments, the type of search query generated based on the technical solution information can be one or more, such as semantic search queries, classification number search queries, keyword search queries, etc., and the number of search queries corresponding to each type can be one or more. With a clear search query, existing technologies related to technical solution information can be retrieved from one or more existing technology databases based on the search query. Based on the existing technologies, technical solution information, and prompts for instructing the large model to generate search results, the large model for generating search results is invoked to trigger the large model to perform feature comparison on the existing technologies and technical solution information and generate the corresponding search results.

[0103] It is understood that the generation and retrieval process of the above-mentioned search formulas are universal and not limited to a single search type or a single search expression. Search formulas can be semantic search formulas, classification number search formulas, keyword search formulas, or combinations of several of these. The retrieval process can include at least one of the following methods: semantic retrieval, classification number-based retrieval, and keyword-based retrieval. The above-mentioned general retrieval process is used to obtain search results that characterize the differentiated features between the technical solution information and existing technologies, and these search results are used as one of the bases for generating supplementary explanatory text, so as to highlight the distinguishing features from existing technologies in the generated explanatory text.

[0104] In some implementations, after generating the search results, the device providing auxiliary text can also generate modification suggestions for the technical solution information based on the feature comparison results between the technical solution information represented by the search results and existing technical documents; wherein, the modification suggestions are used to display them through a visual interface to broaden the differences between the technical solution information and existing technical documents.

[0105] In this embodiment, to generate supplementary explanatory text that effectively distinguishes it from existing technologies for user reference, the device providing the supplementary text can extract technical solution information so that the data volume of the extracted result is smaller than the data volume of the technical solution information. Feature recognition is performed on the extracted result to obtain the technical features to be compared, or feature recognition is directly performed on the technical solution information to obtain the technical features to be compared. These technical features are then compared with corresponding features in existing technical documents to obtain feature comparison results reflecting multiple sets of technical features (one set of technical features includes one feature from the extracted result and one feature from existing technical documents) and a comprehensive comparison result integrating all feature comparison results. The comprehensive comparison result can represent the overall relevance between the technical solution information and existing technical documents. Based on this, modification suggestions can be generated to broaden the differences between the technical solution information and existing technical documents, or, if the differences do not meet expectations, modification suggestions can be generated to broaden the differences between the technical solution information and existing technical documents. These modification suggestions instruct the user to make adaptive modifications to one or more parts of the technical solution information.

[0106] In some implementations, the step of invoking a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: determining a target text review standard suitable for the technical solution information from multiple text review standards based on the language corresponding to the technical solution information; and invoking the pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information based on a template corresponding to the target text review standard. In other words, the device providing auxiliary text can also determine a target text review standard suitable for the technical solution information from multiple text review standards based on the language corresponding to the technical solution information; and invoking the pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information based on a template corresponding to the target text review standard.

[0107] In this embodiment, different review bodies may have different text review standards for the specification. Therefore, as a reference for drafting the specification, if the supplementary explanatory text can conform to the corresponding text review standards, the efficiency of specification formation can be further improved. Therefore, the device providing supplementary text can output prompt information in the language corresponding to the technical solution information to remind the user to switch to the target text review standard, so that the supplementary explanatory text is generated based on the target text review standard. Among various text review standards, each text review standard can correspond to different languages ​​and different review bodies. The text review standards can be used to constrain the form and substantive content of the supplementary explanatory text.

[0108] In some embodiments, the method for providing auxiliary text may further include: exporting the auxiliary explanatory text to a specified file upon detecting a user-triggered export operation. In other words, the apparatus for providing auxiliary text may also provide an export function for the auxiliary explanatory text; wherein the export function is used to export the auxiliary explanatory text into text that conforms to the target text review specifications.

[0109] In this embodiment, in order to facilitate users to quickly generate instruction manuals using the content in the auxiliary instruction text, the device providing the auxiliary text can provide an export function. When the export function is triggered, auxiliary instruction text that conforms to the target text review specifications is output so that users can store, share, and edit it.

[0110] Please refer to Figure 5. An embodiment of this application also provides a method for providing auxiliary text, comprising: step S210: obtaining technical solution information input by a user to describe a technical solution; wherein the technical solution information includes detailed technical description information and feature clauses extracted from the technical solution information; step S220: generating a search query based on the feature clauses and the detailed technical description information; step S230: generating search results based on prior art documents recalled by the search query to characterize the differences between the technical solution information and prior art documents; step S240: generating auxiliary explanatory text corresponding to the technical solution information; wherein the auxiliary explanatory text is generated based on the technical solution information and the search results, the search results are used to characterize the differentiated features of the technical solution information from prior art documents, and the auxiliary explanatory text includes auxiliary explanatory content to clarify the differentiated features.

[0111] In some implementations, the technical solution information includes technical detail description information and feature clauses extracted from the technical solution information; the process of generating the search results includes: generating a search expression based on the feature clauses and the technical detail description information; and generating search results based on the prior art documents recalled by the search expression to characterize the differences between the technical solution information and the prior art documents.

[0112] For example, referring to Figure 5, the technical solution information received in step S210 is recorded as "a convolutional neural network structure for image recognition, the technical details of which include residual modules, attention mechanisms, and dilated convolutions as implementation details; and the features extracted from the feature terms extracted from the technical solution information include 'convolutional neural network', 'residual module', 'attention mechanism', etc." In step S220, a retrieval expression can be constructed exemplary, for example: ("convolutional neural network" AND ("residual module" OR "residual") AND ("attention mechanism" OR "self-attention")) AND IPC:G06N (or the corresponding semantic retrieval expression is generated by a pre-trained large language model in the form of semantic retrieval prompts). In step S230, based on the prior art documents retrieved by the retrieval query, the similarity and functional overlap of each technical feature are evaluated through vectorization comparison (such as vectorizing the first technical feature and the second technical feature and calculating the cosine similarity), knowledge graph path analysis, or multi-dimensional scoring mechanism, so as to obtain the differentiated features revealed in the retrieval results (e.g., prior art A discloses convolutional networks and residual structures but does not combine them with a specific type of attention mechanism, prior art B uses attention but does not use dilated convolution to expand the receptive field, etc.). Based on the above differential feature analysis, step S240 can generate specific modification suggestions and auxiliary explanatory content to clarify the differential features. For example, it is suggested to add the limitation "a residual-attention module is set after the third convolutional layer and the residual-attention module includes dilated convolution to expand the receptive field" to the claims to expand the difference from the prior art. At the same time, a pre-trained large language model is called to generate specific paragraphs for the specification, such as "Unlike prior art X, this embodiment introduces a residual-attention module after the third convolutional layer. This module combines dilated convolution to expand the receptive field, thereby improving the ability to recognize small-scale targets without increasing the saliency of parameters."

[0113] Considering that relevant large language models may have limitations in identifying subtle differences in technical features and distinguishing similar prior art when processing long texts (such as complete technical disclosures and multiple claims), this application, in some embodiments, strengthens the analysis of differentiated features by introducing search results into the generation path to improve the targeting of the generated text. Specifically, this may include the following steps: generating a search query (e.g., keyword search, semantic search, or classification number search) based on the extracted feature clauses and technical details; recalling prior art documents related to the technical solution based on the search query; performing feature extraction and comparison (using vectorized representation, knowledge graph paths, or multi-level scoring mechanisms) on the technical solution information and the recalled prior art documents to identify differentiated features and their scope that distinguish them from the prior art; and using the results of the differentiated feature analysis and the technical solution information together as input or generation prompts for the large language model (e.g., a pre-trained large language model) to guide the large language model to prioritize clarifying these differentiated features and providing explanations of corresponding embodiments, scopes, or technical means when generating auxiliary explanatory text. It can be seen that in the process of processing long text patents, the large language model can more accurately identify and highlight the key differences between the technical solution and the prior art based on the search results, thereby generating more targeted auxiliary explanatory text, which is conducive to more clearly defining the scope of protection and improving the referenceability of the specification.

[0114] In this embodiment, the specific functions and effects of the method for providing auxiliary text can be explained by referring to other embodiments of this application, and will not be repeated here.

[0115] Please refer to Figure 6. This application also provides an apparatus for providing auxiliary text. The apparatus for providing auxiliary text may include: an information acquisition module 101, used to acquire technical solution information input by a user; and a text generation module 104, used to invoke a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information; wherein the auxiliary explanatory text is generated based on the technical solution information and search results corresponding to the technical solution information, the search results being used to characterize the differentiated features of the technical solution information from existing technical documents, and the auxiliary explanatory text containing auxiliary explanatory content for clarifying the differentiated features.

[0116] In some embodiments, the device for providing auxiliary text may include one or more processors (e.g., general-purpose processors, microprocessors, digital signal processors, coprocessors, etc.), memory (including volatile memory and non-volatile memory), input / output interfaces (e.g., keyboard, mouse, touchscreen, camera), a display module, and a network interface module for network communication (e.g., Ethernet port, cellular communication module). In this embodiment, when the information acquisition module 101 is implemented by a general-purpose processor, the structure corresponding to the information acquisition module 101 may include: the processor and a memory for storing the computer program instructions corresponding to the information acquisition module 101. The function of the information acquisition module 101 can be implemented by loading and executing the program instructions stored in the memory on the processor. The program instructions receive and temporarily store technical solution information (e.g., text, images, videos, links, etc.) from the user through the input / output interface or network interface. Similarly, the structure corresponding to the text generation module 104 may include: a processor and a memory, wherein the program instructions are stored in the memory, and the program instructions are used to call local or remote models and algorithms to generate auxiliary explanatory text based on the technical solution information and its retrieval results. When implemented in hardware or for acceleration purposes, the information acquisition module and / or text generation module may be implemented wholly or partially by dedicated circuits (e.g., application-specific integrated circuits, field-programmable gate arrays) or hybrid hardware / software; the dedicated circuits or logic circuits are alternative structural embodiments. The processor, memory, network interface, dedicated circuits, etc., described above can be arbitrarily combined or distributed between the local device and the remote server for collaborative operation.

[0117] In some embodiments, the device for providing auxiliary text may include: at least one processor; a memory storing computer program instructions that can be loaded and executed by the processor; and a network interface module; wherein, when the computer program instructions are loaded and executed by the processor, the device enables the information acquisition module 101 and the text generation module 104 to perform their functions.

[0118] In this embodiment, the specific functions and effects of the device providing auxiliary text can be explained by referring to other embodiments of this application, and will not be repeated here.

[0119] Please refer to Figure 7. One embodiment of this application also provides an apparatus for providing auxiliary text, the apparatus comprising: an information acquisition module 101, configured to acquire technical solution information input by a user to describe a technical solution; wherein the technical solution information includes detailed technical description information and feature clauses extracted from the technical solution information; a search expression generation module 102, configured to generate a search expression based on the feature clauses and the detailed technical description information; a search module 103, configured to generate search results representing the differences between the technical solution information and existing technical documents based on prior art documents recalled by the search expression; and a text generation module 104, configured to invoke a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information; wherein the auxiliary explanatory text is generated based on the technical solution information and the search results, the search results representing the differentiated features of the technical solution information compared to existing technical documents, and the auxiliary explanatory text containing auxiliary explanatory content for clarifying the differentiated features.

[0120] In some embodiments, the structure corresponding to the search query generation module 102 may include a processor, a memory, and program instructions for generating keyword search queries or semantic search vectors on the processor. The structure corresponding to the search module 103 may include a processor, a network interface, and software (or a call to an external search service) for performing a search and returning recalled documents and relevance scores. A local / remote database can be used to store prior art documents related to technical solution information. The local / remote database can be an indexed storage of prior art documents (such as literature or patent data), and can be a remote storage stored locally or accessed via a network. The search query can employ keyword logical combinations, semantic vector representations, or a combination of both, constructed on the processor by corresponding program instructions and submitted through the search module 103. The search module 103 can receive the recall results and their relevance scores and then return the search results to the text generation module 104 for differential feature comparison and auxiliary explanatory text generation. The above modules can also be deployed on a distributed server cluster, using accelerators (such as GPUs or TPUs), or implemented through a hybrid hardware / software approach. GPU (Graphics Processing Unit) is a graphics processing unit; TPU (Tensor Processing Unit) refers to a tensor processing unit.

[0121] In some embodiments, the device for providing auxiliary text may include: at least one processor; a memory storing computer program instructions that can be loaded and executed by the processor; and a network interface module; wherein, when the computer program instructions are loaded and executed by the processor, the device enables the device to perform the functions of the information acquisition module 101, the search-based generation module 102, the search module 103, and the text generation module 104.

[0122] In this embodiment, the specific functions and effects of the device providing auxiliary text can be explained by referring to other embodiments of this application, and will not be repeated here.

[0123] Please refer to Figure 8. This application also provides a computer device, comprising: a memory and a processor, wherein the memory stores at least one computer program, which is loaded and executed by the processor to implement the method described above.

[0124] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, causes the processor to implement the method as described above.

[0125] This application also provides a computer program product containing instructions that, when executed by a processor, implement the method as described above.

[0126] The user information or user account information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, etc.) involved in various embodiments of this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws and regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0127] It is understood that the specific examples in this document are only intended to help those skilled in the art better understand the implementation methods of this application, and are not intended to limit the scope of this application.

[0128] It is understood that in the various embodiments of this application, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0129] It is understood that the various implementation methods described in this application can be implemented individually or in combination, and the implementation methods in this application are not limited in this respect.

[0130] Unless otherwise stated, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The term "and / or" as used in this application includes any and all combinations of one or more of the associated listed items. The singular forms "a," "the," and "the" as used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0131] It is understood that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiments can be completed by the integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly implemented by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0132] It is understood that the memory in the embodiments of this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Specifically, non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may be random access memory (RAM). It should be noted that the memory in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0133] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution information. Those 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 this application.

[0134] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the aforementioned method implementations, and will not be repeated here.

[0135] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0136] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0137] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0138] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution information of this application, essentially or in other words, the part that contributes to the prior art, or a portion of the technical solution information, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0139] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for providing auxiliary text, wherein, The method includes: Obtain technical solution information input by the user; A pre-trained large language model is invoked to generate auxiliary explanatory text corresponding to the technical solution information; wherein, the auxiliary explanatory text is generated based on the technical solution information and the search results corresponding to the technical solution information, the search results are used to characterize the differentiated features of the technical solution information from existing technical documents, and the auxiliary explanatory text contains auxiliary explanatory content to clarify the differentiated features.

2. The method according to claim 1, wherein, The technical solution information includes detailed technical description information and feature clauses extracted from the technical solution information; After obtaining the technical solution information input by the user, the method further includes: Extract the technical features from the aforementioned feature clauses as the first technical feature; Extract the technical features from the technical details description information as the second technical features; Based on the correlation between the first technical feature and the second technical feature, modification suggestions for the feature clause are generated; wherein, the modification suggestions are displayed through a visual interface to improve the correlation.

3. The method according to claim 1, wherein, The technical solution information includes feature clauses extracted from the technical solution information; After obtaining the technical solution information input by the user, the method further includes: The feature clauses are formally checked according to the preset format specifications corresponding to the feature clauses to obtain formal check results; wherein, the formal check results are displayed through a visual interface to indicate formal problems in the feature clauses that do not conform to the format specifications.

4. The method according to claim 1, wherein, The technical solution information includes feature clauses and detailed technical descriptions extracted from the technical solution information; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: Invoke a pre-trained large language model to determine at least one relevant information among the classification number, technology type, technology problem, technology means, and technology efficacy corresponding to the detailed description information of the technology; If at least one of the aforementioned relevant information and the aforementioned feature clauses is confirmed, auxiliary explanatory text corresponding to the technical solution information is generated.

5. The method according to claim 1, wherein, The technical solution information includes feature clauses extracted from the technical solution information; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: Upon detecting an editing operation on the feature clause, a pre-trained large language model is invoked to generate auxiliary explanatory text corresponding to the technical solution information, based on the feature clause after the editing operation and the technical solution information.

6. The method according to claim 5, wherein, The method further includes: Upon detecting an editing operation on a specified text, the text format of the specified text is monitored to obtain the monitoring results; or, After the specified text is generated, the text format of the specified text is monitored to obtain the monitoring results; or, After the specified text is updated, the text format of the specified text is monitored to obtain the monitoring results; or, Upon detecting a user-triggered operation, the text format of the specified text is monitored to obtain the monitoring results; The monitoring results are displayed through a visual interface to characterize the text formatting issues of the specified text, which includes the auxiliary explanatory text.

7. The method according to claim 1, wherein, The technical solution information includes feature clauses and detailed technical descriptions extracted from the technical solution information; the auxiliary explanatory text includes descriptions of multiple implementation methods; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: The pre-trained large language model is invoked to generate implementation description information based on the differences between the technical solution information indicated by the search results and existing technical documents; or, The pre-trained large language model is invoked, and implementation description information conforming to the technical solution layout is generated according to the implementation scheme recorded in the technical detail description information; wherein, the technical solution layout is formed based on the calling relationship between the feature clauses.

8. The method according to claim 1, wherein, The technical solution information includes feature clauses and detailed technical descriptions extracted from the technical solution information; the auxiliary explanatory text includes accompanying drawings; the step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: The pre-trained large language model is invoked to generate the accompanying drawings in the specification corresponding to the technical solution information; The accompanying drawings in the specification include at least one of the following: flowchart, structural diagram, formula diagram, chemical formula structural diagram, and circuit diagram; The flowchart is generated based on at least one of the feature relationships recorded in the feature clauses and the calling relationships between the feature clauses; the structure diagram is generated based on at least one of the structural information recorded in the feature clauses, the structural information recorded in the technical details description information, and image information; the formula diagram, the chemical formula structure diagram, and the circuit diagram are generated based on the technical details description information.

9. The method according to claim 8, wherein, The method further includes: Determine the drawing numbers corresponding to the drawings in each of the specifications; Add corresponding labels to each part of the accompanying drawings according to the drawing numbers in the specification; The auxiliary explanatory text is adjusted according to the accompanying drawings of the specification, which include reference numerals; wherein the adjusted auxiliary explanatory text contains textual descriptions of each part of the accompanying drawings and corresponding reference numerals.

10. The method according to claim 1, wherein, After invoking the pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information, the method further includes: The supplementary explanatory text is displayed through a visual interface; If an editing operation is detected on the auxiliary explanatory text, the auxiliary explanatory text after the editing operation is generated.

11. The method according to claim 1, wherein, The process of generating the search results includes: The pre-trained large language model is invoked to generate a retrieval expression corresponding to the information of the technical solution; The retrieval process is performed based on the retrieval query to recall existing technical documents related to the technical solution information; wherein the retrieval process includes at least one of the following: semantic retrieval process, classification number retrieval process, and keyword retrieval process; The search results are generated based on the relevance between the existing technical documents and the technical solution information.

12. The method according to claim 1, wherein, The technical solution information includes detailed technical description information and feature clauses extracted from the technical solution information; The process of generating the search results includes: A search query is generated based on the aforementioned feature clauses and the detailed technical description information; Based on the prior art documents retrieved using the search-based recall, search results are generated to characterize the differences between the technical solution information and the prior art documents.

13. The method according to claim 1, wherein, After generating the search results, the method further includes: Based on the feature comparison results between the technical solution information represented by the search results and existing technical documents, modification suggestions for the technical solution information are generated; The proposed modifications are presented through a visual interface to broaden the differences between the technical solution information and existing technical documents.

14. The method according to any one of claims 1 to 13, wherein, The step of calling a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information includes: Based on the language corresponding to the technical solution information, a target text review standard suitable for the technical solution information is determined from a variety of text review standards. The pre-trained large language model is invoked to generate auxiliary explanatory text corresponding to the technical solution information based on the template corresponding to the target text review specifications.

15. The method according to claim 14, wherein, The method further includes: If a user-triggered export operation is detected, the auxiliary explanatory text will be exported to a specified file.

16. An apparatus for providing auxiliary text, wherein, The device includes: The information acquisition module is used to acquire technical solution information input by the user; The text generation module is used to call a pre-trained large language model to generate auxiliary explanatory text corresponding to the technical solution information; wherein, the auxiliary explanatory text is generated based on the technical solution information and the search results corresponding to the technical solution information, the search results are used to characterize the differentiated features of the technical solution information from existing technical documents, and the auxiliary explanatory text contains auxiliary explanatory content to clarify the differentiated features.

17. The apparatus according to claim 16, wherein, The technical solution information includes detailed technical description information and feature clauses extracted from the technical solution information; the device further includes: A search query generation module is used to generate a search query based on the feature clauses and the technical detail description information; The retrieval module is used to generate retrieval results based on the prior art documents retrieved by the retrieval query, which characterize the differences between the technical solution information and the prior art documents.

18. A computer device, wherein, The computer device includes a memory and a processor, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to implement the method as described in any one of claims 1 to 15.

19. A computer-readable storage medium, wherein, The computer-readable storage medium stores at least one computer program, which, when executed by a processor, is capable of implementing the method as described in any one of claims 1 to 15.

20. A computer program product, wherein, The computer program product is used to implement the method as described in any one of claims 1 to 15.

Images