A tender generation method, device, equipment, medium and program

Abstract

The application discloses a kind of tender generation method, device, equipment, medium and program.The method includes: first respectively to technical scoring rules and technical specification are structured analysis, obtain the scoring element set comprising multiple scoring clauses and specification structure tree, and the semantic mapping relationship between scoring clause and format file is established, form mapping matrix, then can with scoring element set and mapping matrix as generation constraint, fusion historical tender data and specification structure tree, generate the tender catalog of covering all scoring clauses in scoring element set, then according to the attribute of each directory node in tender catalog, corresponding content generation strategy is matched, generates the tender content corresponding to each directory node, obtains complete technical tender, finally to complete technical tender is audited, and the integration of auditing result and complete technical tender, obtain target technical tender report.The application can guarantee the accuracy and compliance of tender content structure.

Description

Technical Field

[0001] This invention relates to the field of automatic document generation technology, and in particular to a method, apparatus, device, medium and program for generating tender documents. Background Technology

[0002] In the fields of engineering, IT services, and equipment procurement, the preparation of technical tender documents is a highly structured, specialized task with extremely high compliance requirements.

[0003] Currently, the preparation of mainstream technical tender documents in the industry still relies heavily on manual experience or semi-automated replacement based on traditional templates. However, technical tender documents generated manually or using fixed models are prone to problems such as redundant table of contents, disordered hierarchical relationships, and omissions of core scoring items. Furthermore, they are difficult to cover all rules and regulations, making it difficult to guarantee the accuracy and compliance of the technical tender document's content structure.

[0004] Therefore, there is an urgent need for a technical tender document generation method that can guarantee the accuracy and compliance of the content structure. Summary of the Invention

[0005] This invention provides a method, apparatus, device, medium, and program for generating tender documents, which can ensure the accuracy and compliance of the content structure of tender documents.

[0006] According to a first aspect of the present invention, a method for generating tender documents is provided, comprising: The uploaded technical scoring rules, format files, and technical specifications are obtained. The technical scoring rules and technical specifications are structured and parsed to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. A semantic mapping relationship between the scoring clauses and the format files is established to form a mapping matrix. Read historical tender data, and using the set of scoring elements and the mapping matrix as generation constraints, merge the historical tender data and the specification structure tree to generate a tender catalog covering all scoring clauses in the set of scoring elements; Based on the attributes of each directory node in the tender document directory, the corresponding content generation strategy is matched, and based on the content generation strategy, the tender document content corresponding to each directory node is generated to obtain a complete technical tender document. Based on the set of scoring elements and the mapping matrix, the complete technical tender is audited, and the audit results are integrated with the complete technical tender to obtain the target technical tender report.

[0007] According to a second aspect of the present invention, a tender document generation apparatus is provided, comprising: The acquisition module is used to acquire the uploaded technical scoring rules, format files, and technical specifications. It performs structured parsing on the technical scoring rules and the technical specifications to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. It also establishes a semantic mapping relationship between the scoring clauses and the format files to form a mapping matrix. The catalog generation module is used to read historical tender data and, using the set of scoring elements and the mapping matrix as generation constraints, merge the historical tender data and the specification structure tree to generate a tender catalog covering all scoring clauses in the set of scoring elements. The content generation module is used to match the corresponding content generation strategy according to the attributes of each directory node in the tender document directory, and generate the tender document content corresponding to each directory node based on the content generation strategy to obtain a complete technical tender document. The audit module is used to audit the complete technical tender document based on the set of scoring elements and the mapping matrix, and to integrate the audit results with the complete technical tender document to obtain the target technical tender document report.

[0008] According to a third aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the tender generation method according to any embodiment of the present invention.

[0009] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the tender generation method according to any embodiment of the present invention.

[0010] According to a fifth aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the tender generation method described in any embodiment of the present invention.

[0011] The technical solution of this invention can first perform structured parsing of the technical scoring rules and technical specifications to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. A semantic mapping relationship between the scoring clauses and the format file is then established, forming a mapping matrix. Next, historical tender data can be read, and using the set of scoring elements and the mapping matrix as generation constraints, the historical tender data and the specification structure tree are merged to generate a tender catalog covering all scoring clauses in the set of scoring elements. Then, based on the attributes of each directory node in the tender catalog, a corresponding content generation strategy is matched, and based on the content generation strategy, the tender content corresponding to each directory node is generated, resulting in a complete technical tender. Finally, based on the set of scoring elements and the mapping matrix, the complete technical tender is audited, and the audit results are integrated with the complete technical tender to obtain a target technical tender report. This method can construct a tender catalog covering all scoring clauses in the set of scoring elements, thereby generating a corresponding target technical tender report, ensuring the accuracy and compliance of the technical tender content structure.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of a tender document generation method provided in Embodiment 1 of the present invention; Figure 2 This is a flowchart of a tender document generation method provided in Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of a tender document generation device according to Embodiment 3 of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device that implements the tender document generation method of this invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0016] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] Example 1 Figure 1 This is a flowchart of a tender document generation method provided in Embodiment 1 of the present invention. This embodiment is applicable to the automated generation of tender documents. The method can be executed by a tender document generation device, which can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown, the method includes: S101. Obtain the uploaded technical scoring details, format file, and technical specification. Perform structured parsing on the technical scoring details and the technical specification to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. Establish a semantic mapping relationship between the scoring clauses and the format file to form a mapping matrix.

[0018] The technical evaluation details are the evaluation criteria document issued by the tendering party for reviewing technical bids, containing multiple evaluation elements, evaluation weights, and evaluation conditions. The format document can be a bid preparation template document issued by the tendering party, specifying the chapter structure, table of contents, and document style of the bid. The technical specification document can be a technical requirements document issued by the tendering party, containing specifications such as the project's technical parameters, implementation requirements, and delivery standards.

[0019] A scoring element set can be a data set containing multiple scoring clauses. A scoring clause can be understood as scoring rule information consisting of a scoring clause identifier, detailed review items, scoring weights, and scoring conditions. A specification structure tree can be a hierarchical document structure representation obtained after parsing the directory structure of a technical specification. For example, this structure can include specification content, specification table of contents, and specification template.

[0020] A mapping matrix can be understood as a structured data representation obtained by associating each scoring clause with the content of the corresponding format file.

[0021] For example, users can upload the technical evaluation rules, format documents, and technical specifications issued by the bidding party. Then, the bid document generation device can perform document parsing, text extraction, and semantic unit annotation on the technical evaluation rules to extract the evaluation clause identifier (id), detailed review item (item), evaluation weight (weigh), and evaluation condition bundle for each evaluation clause, and organize them into a set of evaluation elements C={c1,c2,...c n}, where each rating clause c i For a multi-attribute tuple, c i =(id,item,weight,bundary). Simultaneously, the technical specification document can be subjected to directory level extraction, extracting the directory level, text styles, and text content to construct a "content-directory-style" structure tree for the specification document.

[0022] Next, a pre-defined large language model can be used to perform semantic association analysis on each scoring clause in the scoring element set and the chapter units in the format file, establish the correspondence between scoring clauses and chapters in the format file, and form a mapping matrix of "scoring clause - format file - format content".

[0023] The advantages of this setup are: it enables deep semantic understanding and precise alignment of unstructured and semi-structured files, significantly improves processing efficiency, and effectively eliminates response blind spots caused by fragmentation of multi-source information.

[0024] Furthermore, after establishing the mapping matrix, if there are unresponsive scoring clauses that fail to establish a semantic mapping relationship with the format file, keyword matching and historical forced list comparison are performed on the unresponsive scoring clauses to obtain a first classification result and a corresponding confidence level; when the confidence level is lower than a preset threshold, a preset large language model is called to perform semantic reasoning classification on the unresponsive scoring clauses to obtain a second classification result; based on the first classification result or the second classification result, the scoring clause is marked as a technical scoring base document or other technical document to be provided.

[0025] Among these, unresponsive scoring clauses can be scoring clauses for which no corresponding section or content can be found in the format file. The historical mandatory list can be a list of response clauses pre-maintained based on the company's past experience in preparing tender documents.

[0026] The technical evaluation base documents can refer to fundamental documents such as the technical cover page, bid deviation table, and technical clause response summary table in the technical tender. Other technical documents to be provided can be other technical response documents that need to be provided separately and are not part of the technical evaluation base documents.

[0027] Understandably, the first classification result can be the category label and corresponding confidence score obtained by classifying the unresponsive rating terms through a rule engine. The second classification result can be the category label obtained by semantic reasoning through a large language model when the confidence score of the rule engine classification is insufficient.

[0028] For example, after the mapping matrix is ​​constructed, a gap detection process can be initiated to identify unresponsive scoring clauses that fail to establish a valid semantic mapping relationship with the format file and generate a list of tags.

[0029] First, a rule engine can be used to sequentially perform keyword matching and historical mandatory list comparison on each unresponsive rating clause in the flagged list. Keyword matching can be based on a pre-defined keyword dictionary, calculating the semantic similarity between the unresponsive rating clause and each term in the dictionary. Historical mandatory list comparison can determine whether the unresponsive rating clause is a mandatory clause that the company has historically been required to respond to. The keyword matching results and historical mandatory list comparison results are then weighted to obtain the first classification result and the corresponding rule confidence score.

[0030] Specifically, the process of determining the rule confidence corresponding to the first classification result can be expressed as follows: ; in, This indicates the i-th unresponsive rating clause. This represents the rule confidence level corresponding to the first category result of the i-th unresponsive scoring clause. This represents the keyword matching result for the i-th unresponsive rating clause. This represents a keyword dictionary. This represents the historical mandatory list comparison result for the i-th unresponsive scoring clause. This refers to a historical mandatory list. and This represents the weighting coefficient.

[0031] When the rule confidence level is higher than or equal to a preset threshold, the first classification result can be directly adopted as the final classification of the unresponsive scoring clause. When the rule confidence level is lower than the preset threshold, it is determined to be a rule blind spot. At this time, the original text of the unresponsive scoring clause, the content of the format file, the company's historical tender examples, and the category definition can be injected into a preset large language model. The large language model outputs the probability distribution of the unresponsive scoring clause belonging to each category, and the category corresponding to the maximum probability is taken as the second classification result.

[0032] Ultimately, based on the results of the first or second classification, unresponsive scoring clauses can be marked as technical scoring base documents or other technical documents to be provided.

[0033] S102. Read historical tender data, and using the set of scoring elements and the mapping matrix as generation constraints, merge the historical tender data and the specification structure tree to generate a tender catalog covering all scoring clauses in the set of scoring elements.

[0034] Among them, historical tender data can be a collection of various technical tender documents that the bidding policy has previously prepared and stored for the tendering party, including historical directory structure, historical text content, etc.

[0035] The table of contents can be a hierarchical structure of technical bid sections, and this table of contents must fully cover all scoring clauses.

[0036] For example, historical tender data can be read from a local database or cloud storage. Then, using the set of scoring elements and the mapping matrix as generation constraints, the directory nodes in the historical tender data can be associated with the scoring clauses. Combined with the chapter information extracted from the specification structure tree, a tender catalog covering all scoring clauses in the set of scoring elements can be constructed through iterative generation and verification.

[0037] S103. Based on the attributes of each directory node in the tender document directory, match the corresponding content generation strategy, and generate the tender document content corresponding to each directory node based on the content generation strategy to obtain a complete technical tender document.

[0038] The table of contents nodes can be individual chapter nodes within the tender document's table of contents. The attributes of these nodes can be semantic feature classifications determined by the type of scoring clauses, content source type, or technical response requirements they represent, such as technical response category, general solution category, performance proof category, compliance commitment category, etc. The content generation strategy can be pre-configured, employing different content generation methods for different node attributes. A complete technical tender document can be a tender file containing all chapter content and conforming to the required format.

[0039] For example, after determining the tender document catalog that covers all scoring clauses, semantic attribute identification can be performed on each catalog node in the tender document catalog. For example, it can be determined whether the catalog node specifically belongs to the type of technical specification response, general engineering technical solution, project performance proof, compliance commitment, etc. Then, according to the type of the target node, the content generation strategy that matches the node type can be automatically selected from multiple preset content generation strategies. Based on the content generation strategy, combined with the scoring element set, mapping matrix and technical specification content, the tender document content corresponding to each catalog node can be generated. Finally, all node contents are merged to obtain a complete technical tender document.

[0040] The advantage of this setup is that by dynamically matching the content generation strategy based on the directory node attributes, the generated content can better meet the scoring requirements and technical specifications, improving the relevance and compliance of the tender content, while reducing the cost of manual intervention.

[0041] Furthermore, the content generation strategy may include a voice conversion and compliance enhancement strategy, a content enhancement generation strategy, a performance grading generation strategy, a multi-source retrieval enhancement strategy, and other technical document processing strategies to be provided.

[0042] Among them, the language conversion and compliance enhancement strategy is used to locate the original paragraphs corresponding to the directory nodes in the specification structure tree, convert declarative language into the bidder's commitment language, and supplement compliance statements; the content enhancement generation strategy is used to retrieve semantically similar content to the directory nodes in the company's internal general specification library, and rewrite and enhance it in combination with project parameters; the performance grading generation strategy is used to grade and organize performance information according to three dimensions: corporate performance, team performance, and individual performance, and select effective performance to generate descriptive content based on scoring weights; the multi-source retrieval enhancement strategy is used to retrieve effective fragments and perform context splicing and logical reconstruction using technical scoring details, technical specifications, historical bid data, and internal general specifications as knowledge sources; and the other unprovided technical document processing strategy is used to generate commitment letters and point-to-point response records for the scoring clauses corresponding to unformatted documents.

[0043] For example, for table of contents nodes whose content can be traced back to the technical specification, a language conversion and compliance enhancement strategy can be adopted to locate the original paragraphs from the specification's structure tree. Through a large language model, declarative expressions such as "shall..." and "must meet..." are converted into commitment expressions such as "we promise..." and "we will strictly implement..." and bidding-specific compliance terms such as "fully responsive, without deviation" are automatically added to adapt to the bidding document writing specifications and finally generate a compliant commitment text.

[0044] For nodes with high semantic similarity to the company's internal general specification library, a content enhancement generation strategy can be adopted. This involves using a BGE vector model to calculate the similarity between directory nodes and the company's general specification library and historical high-quality bidding data. A preset number of optimally matching general specification contents are then selected to form a matching library. Based on the mapping matrix and the matching library, and combined with actual project parameters (obtained from the technical specifications), targeted rewriting can be performed, adding project-specific parameters and technical measures. Simultaneously, general content in the template that is incompatible with this project is automatically deleted, and technically relevant content that conforms to the project's actual situation is output.

[0045] For nodes related to project performance, a performance tiered generation strategy can be adopted. The performance classification agent can be used to classify performance into three dimensions: enterprise, team, and individual. Valid performance can be screened according to the scoring weight and threshold requirements, and a differentiated performance template can be called to generate a structured performance description.

[0046] For nodes not covered by the first three strategies, a multi-source retrieval enhancement strategy can be adopted. Using scoring rules, technical specifications, historical tender documents, and general corporate standards as knowledge sources, a vector index can be constructed and effective fragments can be retrieved. Contextual splicing, logical reconstruction, and original enhancement can be performed to generate exclusive content that fits the bidding scenario.

[0047] For scoring clauses without a formatted document, they are categorized under other technical documents to be provided, and the processing strategy for other technical documents to be provided is adopted to automatically generate a letter of commitment and a structured point-to-point response record arranged in the order of the specification chapters.

[0048] Furthermore, when a technical scoring base file exists, after generating the tender content corresponding to each directory node, the area to be filled in the technical scoring base file can be determined, and the area to be filled can be filled based on a preset large language model; the filled technical scoring base file is then merged with the tender content corresponding to each directory node to obtain a complete technical tender.

[0049] The technical scoring base documents can be basic, standardized documents such as technical covers, bid deviation tables, technical clause response summary tables, project basic overview descriptions, and basic information documents of bidding entities. These documents heavily rely on external fixed information such as basic business registration information, qualification information, and project filing information, and usually have no corresponding relationship with the scoring clauses. The areas to be filled can be blank areas or placeholder areas in the technical scoring base documents where specific information needs to be filled in; the preset large language model can be a pre-trained or configured neural network model for text generation and content filling.

[0050] For example, the technical scoring base file can be broken down into fill units to identify the areas to be filled. Then, based on the company's basic materials and the current project characteristics, a preset large language model can be invoked to generate and fill the content in the areas to be filled. For example, matching content can be automatically generated for text-based areas such as company introduction, project overview, and performance capability description. During the filling process, the format of the filled content should be consistent with the format of the content before and after the filling position. For areas that cannot be automatically filled or lack materials, they can be highlighted to prompt manual processing. Finally, the filled technical scoring base file can be merged with the tender document content corresponding to each generated directory node to form a complete technical tender document.

[0051] Optionally, after generating the tender document content corresponding to each directory node, structured annotations can be added to the content generated by different content generation strategies, and highly distinguishable colors can be used to mark the content generated by different content generation strategies. The structured annotations can include the generation strategy, content source, and source code. This ensures that the content source is traceable, facilitating targeted manual review and modification.

[0052] S104. Based on the set of scoring elements and the mapping matrix, audit the complete technical tender document, and integrate the audit results with the complete technical tender document to obtain the target technical tender document report.

[0053] Auditing can be understood as the process of checking the coverage and compliance of the scoring clauses in a complete technical tender document. Audit results can be used to indicate the coverage of the scoring clauses, identify risks, and provide optimization suggestions. The target technical tender report can refer to the final tender document formed by integrating the complete technical tender document with the audit information.

[0054] For example, the scoring element set and mapping matrix can be used as the review benchmark to verify the response content of each scoring clause in the complete technical tender document, calculate the coverage and content compliance, obtain the audit results, and then integrate the audit results with the complete technical tender document in a formatted manner to generate a target technical tender document report containing the main text, review information and optimization suggestions.

[0055] The technical solution of this embodiment can first perform structured parsing of the technical scoring rules and technical specifications to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. A semantic mapping relationship between the scoring clauses and the format file is then established, forming a mapping matrix. Next, historical tender data can be read, and using the set of scoring elements and the mapping matrix as generation constraints, the historical tender data and the specification structure tree are merged to generate a tender catalog covering all scoring clauses in the set of scoring elements. Then, based on the attributes of each directory node in the tender catalog, the corresponding content generation strategy is matched, and based on the content generation strategy, the tender content corresponding to each directory node is generated, resulting in a complete technical tender. Finally, based on the set of scoring elements and the mapping matrix, the complete technical tender is audited, and the audit results are integrated with the complete technical tender to obtain a target technical tender report. This method can construct a tender catalog covering all scoring clauses in the set of scoring elements, thereby generating the corresponding target technical tender report, ensuring the accuracy and compliance of the technical tender content structure.

[0056] Example 2 Figure 2 This is a flowchart of a tender document generation method provided in Embodiment 2 of the present invention. This embodiment can be further improved and described based on the above embodiments. Figure 2 As shown, the method includes: S201. Obtain the uploaded technical scoring details, format file, and technical specification. Perform structured parsing on the technical scoring details and the technical specification to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. Establish a semantic mapping relationship between the scoring clauses and the format file to form a mapping matrix.

[0057] S202. Read historical tender data, and using the set of scoring elements and the mapping matrix as generation constraints, associate the scoring terms in the set of scoring elements with the directory nodes in the historical tender data to generate an initial directory framework.

[0058] The initial directory framework can be a directory tree structure that is initially generated based on the set of scoring elements and historical tender data, but has not yet integrated the technical specification entries.

[0059] For example, the directory nodes and corresponding content summaries can be extracted from historical tender data first. Then, using the set of scoring elements and the mapping matrix as generation constraints, a dynamic prompt word template is constructed. The detailed review items, scoring weights, scoring conditions of each scoring clause, and highly similar directory nodes from historical tender data are used as prompt word inputs and injected into a pre-defined large language model. The large language model decomposes each scoring clause based on semantic understanding and, combined with the structural features of the directory nodes in historical tenders, generates an initial directory framework. During the generation process, the large language model can be forced to output a directory tree structure that conforms to a pre-defined JSON Schema. If the output does not conform to the schema validation, a secondary generation is triggered until a valid initial directory framework is obtained.

[0060] The advantage of this setup is that it uses the set of scoring elements and the mapping matrix as prior constraints for catalog generation, while learning from the validated catalog structure in historical tenders. This ensures that the initial catalog framework is aligned with the scoring details in terms of macro structure, avoiding large-scale structural deviations.

[0061] S203. Using each directory node in the initial directory framework as a query, perform semantic retrieval in the specification structure tree to extract the specification chapter titles and entries that are semantically related to each of the scoring clauses, forming a specification-side directory candidate set.

[0062] The candidate set of the specification side directory can be a collection of chapter titles and directory entries that are semantically related to each directory node in the initial directory framework, retrieved from the specification book structure tree.

[0063] For example, each leaf node in the initial directory framework can be used as the query object, and its text description can be used as the query vector to perform semantic retrieval in the specification tree structure. Specifically, a preset vector retrieval model (such as the BGE model) can be used to calculate the relevance score between the query vector and each text block in the specification tree structure, filter out text blocks with relevance scores lower than a preset threshold (such as 0.75), and return the top K (such as K=8) text blocks with the highest relevance.

[0064] Subsequently, the retrieved highly relevant text blocks can be injected into the prompts of the large language model to extract the chapter titles, technical requirements, and deliverables list implicit in the specification, forming a candidate set of specification side catalogs.

[0065] S204. Semantically merge the initial directory framework and the candidate set of the standard side directory to obtain a fused directory tree.

[0066] The fused directory tree can be a standardized tree-like directory structure obtained by removing redundancy, determining parent-child relationships, and reconstructing hierarchy from the initial directory framework and the candidate directory set of the standard side.

[0067] For example, the initial directory framework, the candidate set of the specification side directory, and the mapping matrix can be used as inputs, and a preset large language model can be called as a relation discriminator to perform semantic judgment on the subordinate relationships between directory nodes.

[0068] Furthermore, for directory entries with hierarchical relationships (such as "Project Schedule Requirements" belonging to "Project Schedule Control Measures"), the large language model can output parent-child relationships and establish tree edges. For entries with semantic repetition or redundancy, the large language model can output removal instructions. After redundancy merging, the merged tree structure can be topologically sorted and hierarchically compressed, and child nodes under the same parent node can be sorted according to the technical logic flow of "Basic Requirements → Implementation Content → Delivery / Others" to obtain a merged directory tree.

[0069] The advantage of this setup is that it solves the problems of overlapping entries, hierarchical confusion, and disordered subordinate relationships that may exist between the initial directory framework and the candidate set on the specification side, making the merged directory tree more rigorous and standardized in structure.

[0070] S205. Based on the set of scoring elements, the fused directory tree is detected to determine the list of directories to be supplemented.

[0071] The list of directories to be supplemented can be a collection of scoring terms that have been confirmed after coverage testing and have failed to be responded to by any directory node in the merged directory tree.

[0072] For example, the set of scoring elements can be used as the verification benchmark to perform full coverage testing on the fused directory tree.

[0073] Specifically, the large language model can be used to verify whether each scoring clause has a corresponding entry in the fusion directory tree to respond to it, and at the same time verify whether the directory entry covers the content requirements of the corresponding format file in the mapping matrix.

[0074] For rating clauses that are not covered, the verification agent can be used to check the necessity and accuracy of the uncovered catalog, and generate a list of catalogs to be supplemented and supplementary evidence.

[0075] S206. For the list of catalogs to be supplemented, iteratively perform the semantic retrieval and semantic merging operations until the resulting fused catalog tree covers all the scoring clauses or reaches the preset maximum number of iterations, and obtain the tender catalog.

[0076] The preset maximum number of iterations can be a pre-set upper limit for the number of catalog supplementation iterations, used to prevent infinite loops.

[0077] For example, for each item to be supplemented in the list of items to be supplemented, the semantic retrieval in step S203 and the semantic merging in step S204 can be automatically repeated to retrieve the corresponding content from the specification tree structure tree and generate a local supplementary subtree. The local supplementary subtree is then mounted to the fusion directory tree according to the topological relationship.

[0078] After each iteration, the coverage check in step S205 can be re-executed to determine whether all scoring clauses have been covered. If the coverage reaches 100%, the iteration can be terminated. Alternatively, if the number of iterations reaches the preset maximum number of iterations (e.g., 3 times) and the remaining uncovered clauses are determined to be non-mandatory by the enterprise rule base, the iteration can be stopped, and the final tender document catalog can be output.

[0079] The advantage of this setup is that it ensures the tender document catalog covers all scoring clauses to the greatest extent possible, while also enabling adaptive optimization of the catalog structure.

[0080] S207. Based on the attributes of each directory node in the tender document directory, match the corresponding content generation strategy, and generate the tender document content corresponding to each directory node based on the content generation strategy to obtain a complete technical tender document.

[0081] S208. Based on the mapping matrix and the complete technical tender, construct a four-dimensional mapping matrix; the four-dimensional mapping matrix includes: scoring clauses, format files, format content blocks, and actual generated content blocks.

[0082] The four-dimensional mapping matrix can be an extended mapping relationship formed by adding actual generated content blocks to the original mapping matrix (scoring clauses - format files - format content blocks), used to accurately align the scoring clauses with the actual generated content positions in the tender document.

[0083] For example, based on the mapping matrix, the generated complete technical tender can be retrieved. Then, by matching each "scoring clause - format file - format content block" association record in the mapping matrix with the corresponding content block actually generated in the complete technical tender, a four-dimensional mapping matrix is ​​formed. ;in, This represents the i-th rating clause. This represents the j-th format file. This represents the k-th format content block of the j-th format file. This represents the corresponding content block actually generated in the complete technical tender document.

[0084] S209. The scoring clauses are broken down to obtain multiple scoring detail blocks.

[0085] Among them, the scoring detail block can be an independent scoring point unit that cannot be further divided after the individual scoring clause is atomically broken down according to the review requirements, such as independent technical parameters, solution modules, service terms, qualification requirements or technical response points.

[0086] For example, a large language model can be invoked, using a single scoring clause as the smallest processing unit, to atomically and finely decompose all scoring content under each scoring clause. The decomposition granularity is precise down to individual technical parameters, solution modules, service terms, qualification requirements, or technical response points, forming several independent scoring detail blocks. During the decomposition process, the large language model is constrained to ensure that no scoring points are omitted and no points are merged or overlapped.

[0087] S210. Detect the coverage of each of the scoring detail blocks in the complete technical tender to obtain the coverage of a single clause.

[0088] In this context, the coverage of a single clause can be the proportion of the number of scoring detail blocks under a certain scoring clause that are actually responded to by the complete technical tender to the total number of all scoring detail blocks under that clause.

[0089] For example, keyword matching can be used to detect whether there is semantically equivalent, parameter-name consistent, and terminology-matching professional content in the actual generated content block corresponding to the scoring detail block. If a match is found, a hard hit indicator is output.

[0090] For scoring detail blocks that are not matched by keywords, a large language model is invoked to perform dynamic semantic determination to determine whether the corresponding content block in the complete technical tender fully covers the scoring requirements of the scoring detail block.

[0091] By combining the hit results from the two methods, and calculating the number of hit rating details for a single rating clause and the total number of rating details, the coverage of a single clause can be obtained using the formula "Single Clause Coverage = Number of Hit Details / Total Number of Details × 100%".

[0092] S211. Based on the coverage of the single clause and the level description in the scoring rules, construct a segmented scoring function to predict the score of each scoring clause.

[0093] The grade description can be a textual explanation of the score range or grade corresponding to different levels of response to the same scoring clause in the scoring rules. The segmented scoring function can be a mathematical function established based on the grade description and the coverage of the detailed blocks, used to map the coverage of the scoring detailed blocks to specific scores.

[0094] For example, the tender document generation device can call a large language model to parse the level description in each scoring clause, identify the set of scoring detail blocks and score ranges corresponding to different review levels, and construct a normalized segmented scoring function by combining the total number of scoring detail blocks under that clause. Where x∈[0,1] represents the detail block coverage.

[0095] Then, the calculated single-clause coverage can be used as input and substituted into the corresponding segmented scoring function to predict the score of that scoring clause.

[0096] S212. Generate risk markers and optimization suggestions for the scoring details that are not covered or deviate from the content, and use the scores of each scoring clause, the risk markers and optimization suggestions as the audit results.

[0097] Risk markers can be warning messages used to identify rating details that are not covered, have missing content, or deviate from semantics, along with their corresponding location information. Optimization suggestions can be actionable improvement guidelines generated to address the issues indicated by the risk markers.

[0098] For example, risk markers can be applied to uncovered or deviated scoring details, recording the corresponding content block location and risk type. Simultaneously, using the technical specification content and the actual generated content blocks corresponding to the scoring details as input, a large language model is invoked for in-depth analysis to determine whether the content blocks comply with the technical specification constraints, generating risk warnings for scoring details with insufficient expression or missing content.

[0099] Subsequently, the predicted scores, risk markers, and optimization suggestions for each scoring item can be aggregated as the audit results. The logic for generating optimization suggestions follows a "problem identification - solution suggestion" model, outputting specific and actionable improvement suggestions for each deduction risk item.

[0100] S213. Integrate the audit results and the complete technical tender to obtain the target technical tender report.

[0101] The target technical tender report can be a structured output document containing the complete technical tender text, audit review report, and optimization suggestion list.

[0102] For example, the complete technical tender can be formatted into chapters according to the tender table of contents, retaining all annotations and color-coded information. Simultaneously, the audit results can be structured according to review dimensions to generate an audit review report, including predicted scores for each scoring clause, risk assessments for deductions, and improvement suggestions. A visual representation of the overall predicted score can be provided, and all optimization suggestions can be sorted by risk level to generate a list of optimization suggestions.

[0103] Finally, the formatted complete technical tender document, audit review report, and optimization suggestion list can be integrated into a target technical tender document report, supporting output in mainstream document formats such as docx and pdf, and providing online interactive preview and editing functions.

[0104] The technical solution of this embodiment can further generate an initial directory framework and a candidate set of specification-side directories, then perform semantic merging to obtain a fused directory tree. The fused directory tree is then inspected to determine a list of directories to be supplemented. Next, for the list of directories to be supplemented, the semantic retrieval and semantic merging operations are iteratively performed. This ensures strict alignment between the tender document directory and the set of scoring elements, further improving the accuracy and compliance of the technical tender document's content structure. Simultaneously, by splitting each scoring clause into multiple scoring detail blocks, the coverage of each scoring detail block in the complete technical tender document is detected to obtain the coverage of a single clause. Based on the coverage of the single clause and the grade description in the scoring details, a segmented scoring function is constructed to predict the score of each scoring clause. Risk markers and optimization suggestions are generated for uncovered or deviating scoring details. The scores, risk markers, and optimization suggestions of each scoring clause are used as audit results. This allows for the unified integration of tender document content and audit details, achieving integrated report output and facilitating content revision work by the bidding party.

[0105] Example 3 Figure 3 This is a schematic diagram of a tender document generation device provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: The acquisition module 301 can be used to acquire the uploaded technical scoring rules, format files and technical specifications, perform structured parsing on the technical scoring rules and technical specifications respectively, obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree, and establish a semantic mapping relationship between the scoring clauses and the format files to form a mapping matrix; The catalog generation module 302 can be used to read historical tender data and, using the set of scoring elements and the mapping matrix as generation constraints, fuse the historical tender data and the specification structure tree to generate a tender catalog covering all scoring clauses in the set of scoring elements. The content generation module 303 can be used to match the corresponding content generation strategy according to the attributes of each directory node in the tender document directory, and generate the tender document content corresponding to each directory node based on the content generation strategy to obtain a complete technical tender document. Audit module 304 can be used to audit the complete technical tender based on the set of scoring elements and the mapping matrix, and integrate the audit results with the complete technical tender to obtain a target technical tender report.

[0106] The technical solution of this embodiment can first perform structured parsing of the technical scoring rules and technical specifications to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. A semantic mapping relationship between the scoring clauses and the format file is then established, forming a mapping matrix. Next, historical tender data can be read, and using the set of scoring elements and the mapping matrix as generation constraints, the historical tender data and the specification structure tree are merged to generate a tender catalog covering all scoring clauses in the set of scoring elements. Then, based on the attributes of each directory node in the tender catalog, the corresponding content generation strategy is matched, and based on the content generation strategy, the tender content corresponding to each directory node is generated, resulting in a complete technical tender. Finally, based on the set of scoring elements and the mapping matrix, the complete technical tender is audited, and the audit results are integrated with the complete technical tender to obtain a target technical tender report. This method can construct a tender catalog covering all scoring clauses in the set of scoring elements, thereby generating the corresponding target technical tender report, ensuring the accuracy and compliance of the technical tender content structure.

[0107] Optionally, the acquisition module can also be used for: If there are unresponsive scoring clauses that fail to establish a semantic mapping relationship with the format file, keyword matching and historical mandatory list comparison are performed on the unresponsive scoring clauses to obtain the first classification result and the corresponding confidence level. When the confidence level is lower than a preset threshold, a preset large language model is invoked to perform semantic reasoning classification on the unresponsive scoring terms to obtain a second classification result. Based on the first classification result or the second classification result, the unresponsive scoring clause is marked as a technical scoring base document or other technical document to be provided.

[0108] Optionally, the directory generation module 302 can be used for: Using the set of scoring elements and the mapping matrix as generation constraints, the scoring terms in the set of scoring elements are associated with the directory nodes in the historical tender data to generate an initial directory framework; Using each directory node in the initial directory framework as a query, semantic retrieval is performed in the specification tree structure to extract the specification chapter titles and entries that are semantically related to each of the scoring clauses, forming a specification-side directory candidate set. The initial directory framework and the candidate directory set of the specification side are semantically merged to obtain a fused directory tree; Based on the set of scoring elements, the fused directory tree is inspected to determine the list of directories to be supplemented; For the list of catalogs to be supplemented, the semantic retrieval and semantic merging operations are performed iteratively until the resulting fused catalog tree covers all the scoring clauses or reaches the preset maximum number of iterations, thus obtaining the tender catalog.

[0109] Optionally, the content generation strategy includes a voice conversion and compliance enhancement strategy, a content enhancement generation strategy, a performance-based generation strategy, a multi-source retrieval enhancement strategy, and other technical document processing strategies to be provided.

[0110] Optionally, the content generation module 303 can be used for: Determine the areas to be filled in the technical scoring base file, and fill the areas to be filled based on a preset large language model; The filled-in technical evaluation base file is merged with the tender content corresponding to each directory node to obtain the complete technical tender.

[0111] Optional, audit module 304, specifically can be used for: Based on the mapping matrix and the complete technical tender, a four-dimensional mapping matrix is ​​constructed; the four-dimensional mapping matrix includes: scoring clauses, format files, format content blocks, and actual generated content blocks. Each of the aforementioned scoring clauses is broken down to obtain multiple scoring detail blocks; The coverage of each of the scoring detail blocks in the complete technical tender is detected to obtain the coverage of a single clause; Based on the coverage of the single clause and the level description in the scoring rules, a segmented scoring function is constructed to predict the score of each scoring clause; Risk markers and optimization suggestions are generated for rating details that are not covered or deviate from the scope of the content, and the scores of each rating clause, the risk markers and optimization suggestions are used as audit results.

[0112] The tender document generation device provided in the embodiments of the present invention can execute the tender document generation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0113] Example 4 Figure 4A schematic diagram of an electronic device 400 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0114] like Figure 4 As shown, the electronic device 400 includes at least one processor 401 and a memory, such as a read-only memory (ROM) 402 or a random access memory (RAM) 403, communicatively connected to the at least one processor 401. The memory stores computer programs executable by the at least one processor. The processor 401 can perform various appropriate actions and processes based on the computer program stored in the ROM 402 or loaded into the RAM 403 from storage unit 408. The RAM 403 may also store various programs and data required for the operation of the electronic device 400. The processor 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0115] Multiple components in electronic device 400 are connected to I / O interface 405, including: input unit 406, such as keyboard, mouse, etc.; output unit 407, such as various types of displays, speakers, etc.; storage unit 408, such as disk, optical disk, etc.; and communication unit 409, such as network card, modem, wireless transceiver, etc. Communication unit 409 allows electronic device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0116] Processor 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 401 performs the various methods and processes described above, such as tender document generation methods.

[0117] In some embodiments, the tender generation method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by processor 401, one or more steps of the tender generation method described above may be performed. Alternatively, in other embodiments, processor 401 may be configured to perform the tender generation method by any other suitable means (e.g., by means of firmware).

[0118] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0119] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0120] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0121] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0122] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0123] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0124] In one embodiment, the present invention further includes a computer program product, which includes a computer program that, when executed by a processor, implements the tender generation method of any embodiment of the present invention.

[0125] In implementing the computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0126] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0127] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for generating tender documents, characterized in that, include: The uploaded technical scoring rules, format files, and technical specifications are obtained. The technical scoring rules and technical specifications are structured and parsed to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. A semantic mapping relationship between the scoring clauses and the format files is established to form a mapping matrix. Read historical tender data, and using the set of scoring elements and the mapping matrix as generation constraints, merge the historical tender data and the specification structure tree to generate a tender catalog covering all scoring clauses in the set of scoring elements; Based on the attributes of each directory node in the tender document directory, the corresponding content generation strategy is matched, and based on the content generation strategy, the tender document content corresponding to each directory node is generated to obtain a complete technical tender document. Based on the set of scoring elements and the mapping matrix, the complete technical tender is audited, and the audit results are integrated with the complete technical tender to obtain the target technical tender report.

2. The method according to claim 1, characterized in that, After establishing the semantic mapping relationship between the scoring terms and the format file, forming a mapping matrix, the method further includes: If there are unresponsive scoring clauses that fail to establish a semantic mapping relationship with the format file, keyword matching and historical mandatory list comparison are performed on the unresponsive scoring clauses to obtain the first classification result and the corresponding confidence level. When the confidence level is lower than a preset threshold, a preset large language model is invoked to perform semantic reasoning classification on the unresponsive scoring terms to obtain a second classification result. Based on the first classification result or the second classification result, the unresponsive scoring clause is marked as a technical scoring base document or other technical document to be provided.

3. The method according to claim 1, characterized in that, The step of generating a bid catalog covering all scoring clauses in the scoring element set by fusing the historical bid data and the specification structure tree with the set of scoring elements and the mapping matrix as generation constraints includes: Using the set of scoring elements and the mapping matrix as generation constraints, the scoring terms in the set of scoring elements are associated with the directory nodes in the historical tender data to generate an initial directory framework; Using each directory node in the initial directory framework as a query, semantic retrieval is performed in the specification tree structure to extract the specification chapter titles and entries that are semantically related to each of the scoring clauses, forming a specification-side directory candidate set. The initial directory framework and the candidate directory set of the specification side are semantically merged to obtain a fused directory tree; Based on the set of scoring elements, the fused directory tree is inspected to determine the list of directories to be supplemented; For the list of catalogs to be supplemented, the semantic retrieval and semantic merging operations are performed iteratively until the resulting fused catalog tree covers all the scoring clauses or reaches the preset maximum number of iterations, thus obtaining the tender catalog.

4. The method according to claim 1, characterized in that, The content generation strategy includes a voice conversion and compliance enhancement strategy, a content enhancement generation strategy, a performance-based generation strategy, a multi-source retrieval enhancement strategy, and other technical document processing strategies to be provided.

5. The method according to claim 1, characterized in that, The step of generating the tender document content corresponding to each directory node based on the content generation strategy to obtain a complete technical tender document also includes: Determine the areas to be filled in the technical scoring base file, and fill the areas to be filled based on a preset large language model; The filled-in technical evaluation base file is merged with the tender content corresponding to each directory node to obtain the complete technical tender.

6. The method according to claim 1, characterized in that, The audit of the complete technical tender document based on the set of scoring elements and the mapping matrix includes: Based on the mapping matrix and the complete technical tender, a four-dimensional mapping matrix is ​​constructed; the four-dimensional mapping matrix includes: scoring clauses, format files, format content blocks, and actual generated content blocks. Each of the aforementioned scoring clauses is broken down to obtain multiple scoring detail blocks; The coverage of each of the scoring detail blocks in the complete technical tender is detected to obtain the coverage of a single clause; Based on the coverage of the single clause and the level description in the scoring rules, a segmented scoring function is constructed to predict the score of each scoring clause; Risk markers and optimization suggestions are generated for rating details that are not covered or deviate from the scope of the content, and the scores of each rating clause, the risk markers and optimization suggestions are used as audit results.

7. A tender document generation device, characterized in that, include: The acquisition module is used to acquire the uploaded technical scoring rules, format files, and technical specifications. It performs structured parsing on the technical scoring rules and the technical specifications to obtain a set of scoring elements containing multiple scoring clauses and a specification structure tree. It also establishes a semantic mapping relationship between the scoring clauses and the format files to form a mapping matrix. The catalog generation module is used to read historical tender data and, using the set of scoring elements and the mapping matrix as generation constraints, merge the historical tender data and the specification structure tree to generate a tender catalog covering all scoring clauses in the set of scoring elements. The content generation module is used to match the corresponding content generation strategy according to the attributes of each directory node in the tender document directory, and generate the tender document content corresponding to each directory node based on the content generation strategy to obtain a complete technical tender document. The audit module is used to audit the complete technical tender document based on the set of scoring elements and the mapping matrix, and to integrate the audit results with the complete technical tender document to obtain the target technical tender document report.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the tender generation method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the tender generation method according to any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the tender generation method according to any one of claims 1-6.

Images