Text automation-based bid invitation file intelligent generation method and system

Abstract

The invention discloses an intelligent bid invitation file generation method and system based on text automation, and relates to the technical field of artificial intelligence and data processing, and the method comprises the steps: carrying out the semantic analysis of a bid invitation demand text through a project type recognition model, and extracting a feature vector and a keyword set of a project; based on the feature vector, a cosine similarity algorithm is adopted to retrieve a matching template in a preset bid invitation file template library, a candidate template list is generated, and label classification and parameterization setting are carried out on templates of the candidate template list; based on the keyword set, carrying out content adaptation on templates of the candidate template list, and generating an initial bid invitation file; and performing segmented auditing on the initial bid invitation file, adjusting the keyword set in combination with auditing opinions, and synchronously updating the keyword set to the parameterized database through the mapping relationship to generate the bid invitation file. According to the invention, closed loop from demand to achievement is realized, labor cost of bid invitation file compiling is reduced, and bid invitation file quality and generation efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of artificial intelligence and data processing, and particularly to an intelligent generation method and system for tender documents based on text automation. Background Art

[0002] As an important basis for project implementation, government procurement, and enterprise tendering activities, the quality and integrity of tender documents are directly related to project compliance and implementation effects. However, there are a large number of manual operations in the existing tender document preparation process, and text screening and integration need to be carried out according to project categories, laws and regulations, and historical experience. The process is cumbersome and error-prone, and the generated tender documents often lack unified specifications and logical consistency, resulting in uneven document quality and affecting project evaluation and winning bid effects. Especially in a large-scale and multi-industry tendering environment, how to efficiently and accurately generate tender documents that meet the requirements to reduce manual intervention, improve document writing quality and generation efficiency is a technical problem that urgently needs to be solved at present.

[0003] CN117473971A discloses an automatic generation method and system for tender documents based on a procurement text library. Through natural language processing technology and a keyword extraction algorithm of machine learning, it extracts core text information including laws and regulations and requirement specifications from data sources to form a preliminary text library, and combines knowledge graphs, template matching, and automatic summarization algorithms to improve the speed and quality of text generation, and at the same time realizes the continuous learning and optimization of the system. However, this method mainly focuses on the construction of the text library and template matching, lacks in-depth modeling of the logical relationships between different modules in the tender documents, and cannot perform customized text generation for specific project categories, resulting in certain adaptability deficiencies when dealing with diversified and complex tendering requirements.

[0004] CN118036584A discloses an intelligent tender document generation method and system. By determining the project category of the file to be processed, obtaining historical tender documents and performing similarity comparison, and generating new tender documents based on similar historical documents and their templates, it realizes the automatic generation of documents and the improvement of writing efficiency. However, this method mainly relies on the similarity matching of historical tender documents, lacks in-depth modeling of the semantic consistency between laws and regulations, project requirements, and context, and it is difficult to ensure a high degree of adaptation of the generated tender documents in terms of legal compliance and project specificity. Moreover, for emerging project categories or scenarios lacking historical samples, there is still much room for improvement in the applicability and accuracy of the method. Summary of the Invention

[0005] In view of the problems of poor template adaptability, lack of logical consistency, and difficulty in dealing with emerging project categories in the existing tender document generation methods, the present invention is proposed.

[0006] Therefore, the problem to be solved by the present invention is how to implement an intelligent generation method for tender documents that can adapt to multiple project categories, ensure text logical consistency, and have efficient automation capabilities, so as to improve the efficiency and quality of tender document generation and effectively reduce the workload of manual intervention.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, an embodiment of the present invention provides an intelligent generation method for tender documents based on text automation, which includes: Performing semantic analysis on the tender requirement text through a project type recognition model to extract the feature vector and keyword set of the project; Based on the feature vector, using the cosine similarity algorithm to retrieve and match templates in a preset tender document template library, generating a list of candidate templates, and performing label classification and parameter setting on the templates in the candidate template list; Based on the keyword set, performing content adaptation on the templates in the candidate template list to generate an initial tender document; Performing segmented review on the initial tender document, adjusting the keyword set in combination with the review opinions, and synchronously updating it to the parameterized database through the mapping relationship to generate a tender document.

[0008] As a preferred solution of the intelligent generation method for tender documents based on text automation according to the present invention, wherein: the method for generating the tender document is as follows: Dividing the initial tender document into multiple text paragraphs according to the chapter structure to generate a segmented review list; Constructing a review rule library, performing rule matching on each text paragraph in the segmented review list, and marking the content points that do not meet the specifications; Summarizing the content points that do not meet the specifications to form a review opinion set, where the review opinion set includes problem descriptions, modification suggestions, and associated parameters; Based on the review opinion set, correcting the corresponding parameters in the keyword set to generate an optimized keyword set; Writing the optimized keyword set into the parameterized database and establishing a parameter version tracking record; Calling the latest parameter values of the parameterized database to adjust the initial tender document and output a tender document.

[0009] As a preferred solution of the intelligent generation method for tender documents based on text automation according to the present invention, wherein: the method for generating the initial tender document is as follows: Obtain the templates in the candidate template list, parse the template structure, and identify the placeholder tags, where the placeholder tags include a project scale parameter placeholder, a technical requirement parameter placeholder, and a budget range parameter placeholder; Based on the parsing result of the template structure, establish a template structure index table for identifying the text positions and attribute information of each placeholder in the template; Match the parameters in the keyword set with the placeholder tags in the template structure index table and perform replacement in sequence, where the keyword set includes a project scale parameter, a technical requirement parameter, and a budget range parameter; For the replaced template content, detect whether there are unmatched placeholder tags; Based on the detection result, perform text format specification processing on the detected template content to form an initial bidding document; Compare the text of the initial bidding document with the template structure of the original candidate template list, and update the tag status of the template structure index table.

[0010] As a preferred solution of the intelligent generation method of the bidding document based on text automation according to the present invention, wherein: the generation method of the candidate template list is, Use the cosine similarity algorithm to perform similarity matching between the feature vector of the project and the feature vectors in each template of the preset bidding document template library, and obtain a similarity matching value, where the preset bidding document template library includes a tender announcement template, a tender notice template, an evaluation method template, and a contract clause template; Based on the matching result, select the templates with the similarity matching value greater than the preset threshold to generate a candidate template list; Classify the tags of the templates in the candidate template list, where the tags include a project type tag, an applicable scope tag, and a file type tag; Based on the tag classification result, extract the changeable elements of the templates in the candidate template list and convert them into parameter placeholders; Establish a mapping relationship between the parameter placeholders and the keyword set to form a parameterized configuration table; Perform parameterized settings on the templates in the candidate template list according to the parameterized configuration table to generate a standardized template set.

[0011] As a preferred solution of the intelligent generation method of the bidding document based on text automation according to the present invention, wherein: performing similarity matching between the feature vector of the project and the feature vectors in each template of the preset bidding document template library includes: When the similarity matching value is less than the first preset threshold, the edit distance algorithm is used to recalculate the matching degree; When the similarity matching value is greater than or equal to the first preset threshold and less than the second preset threshold, secondary matching is performed in combination with the word frequency feature; When the similarity matching value is greater than or equal to the second preset threshold, it is directly included in the candidate template list.

[0012] As a preferred solution of the intelligent generation method of bidding documents based on text automation according to the present invention, wherein: the extraction method of the project feature vector and the keyword set is as follows. Receive the bidding requirement text input by the user, perform segmented parsing on the bidding requirement text to form a text paragraph set; Input the text paragraph set into a pre-trained project type recognition model for semantic understanding, calculate the weight distribution of each word in the text, and form a text semantic feature matrix; Based on the text semantic feature matrix, extract the feature vector representing the project type; Identify numerical and descriptive information from the bidding requirement text, and extract project scale parameters according to the preset parameter classification rules; Use a professional dictionary library to perform term matching on the bidding requirement text and extract technical requirement parameters; Match the amount information in the bidding requirement text through regular expression patterns and extract budget range parameters; Integrate the project scale parameters, the technical requirement parameters, and the budget range parameters into a keyword set.

[0013] As a preferred solution of the intelligent generation method of bidding documents based on text automation according to the present invention, wherein: the construction method of the project type recognition model is as follows. Adopt a three-layer encoder structure to establish a pre-trained language model based on BERT, wherein the pre-trained language model includes a number of attention heads; Based on the pre-trained language model, import a professional dictionary library in the bidding field and perform annotation according to three dimensions of engineering, goods, and services, wherein the professional dictionary library includes a number of commonly used words in the bidding field; Use the completed bidding documents as the training data set to perform domain adaptation adjustment on the pre-trained language model; For different types of bidding projects, set an attention weight matrix and pay attention to keywords and phrases related to the project type; Evaluate the performance of the pre-trained language model through a cross-validation method, and select the version of the pre-trained language model with the highest evaluation score as the project type recognition model; Deploy the project type recognition model as a semantic parsing service module to support batch text processing.

[0014] Second aspect, an embodiment of the present invention provides an intelligent tender document generation system based on text automation, which includes: An extraction module, configured to perform semantic parsing on the tender requirement text through a project type recognition model, and extract the feature vector and keyword set of the project, where the keyword set includes project scale parameters, technical requirement parameters, and budget range parameters; A template matching module, based on the feature vector, uses the cosine similarity algorithm to retrieve and match templates in a preset tender document template library, generates a candidate template list, and performs label classification and parameter setting on the templates in the candidate template list; A content adaptation module, based on the keyword set, adapts the content of the templates in the candidate template list to generate an initial tender document; A segmented review and generation module, configured to perform segmented review on the initial tender document, adjust the keyword set in combination with the review opinions, and synchronously update it to the parameterized database through a mapping relationship to generate a tender document.

[0015] Third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program instructions are executed by the processor, the steps of the intelligent tender document generation method based on text automation as described in the first aspect of the present invention are implemented.

[0016] Fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program instructions are executed by the processor, the steps of the intelligent tender document generation method based on text automation as described in the first aspect of the present invention are implemented.

[0017] Compared with the prior art, the beneficial effects of the present invention are: by introducing a project type recognition model, semantic parsing of the tender requirement text is realized, the project feature vector and keyword set are accurately extracted, effectively solving the problems of large manual understanding deviation and information omission in the preparation of traditional tender documents; by matching the preset tender document template library through the cosine similarity algorithm based on the feature vector, and combining label classification and parameter setting, the candidate template list can be quickly screened and generated, significantly improving the accuracy and applicability of template matching; content adaptation based on the keyword set realizes the automated generation of the initial tender document, reducing manual intervention and improving the professionalism and standardization of content generation; adopting segmented review and rule matching, combining review opinions to correct the keyword set and synchronously update the parameterized database, ensuring that the generated tender document is effectively guaranteed in terms of content integrity, standardization, and consistency; this method realizes an efficient closed-loop from requirements to results, significantly reducing the labor cost of tender document preparation and improving the quality and generation efficiency of tender documents. Description of the Drawings

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings. Among them: Figure 1 It is a flowchart of an intelligent generation method for tender documents based on text automation in Embodiment 1. Specific embodiments

[0019] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention with reference to the accompanying drawings of the specification.

[0020] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0021] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that mutually excludes other embodiments.

[0022] Embodiment 1 According to Embodiment 1 of the present invention, in combination with Figure 1 the flowchart shown, an intelligent generation method for tender documents based on text automation specifically includes the following steps: S1: Perform semantic parsing on the tender requirement text through a project type recognition model, and extract the feature vector and keyword set of the project.

[0023] Specifically, it includes: S1.1: Receive the tender requirement text input by the user, perform segmented parsing on the tender requirement text, and form a text paragraph set; S1.2: Input the text paragraph set into a pre-trained project type recognition model for semantic understanding, calculate the weight distribution of each word in the text, and form a text semantic feature matrix; In an alternative embodiment, the method for constructing the project type recognition model is to construct a pre-trained language model based on BERT using a three-layer encoder structure, where the pre-trained language model includes a number of attention heads; based on the pre-trained language model, import the professional dictionary library in the bidding field and annotate it according to three dimensions: engineering, goods, and services, where the professional dictionary library includes a number of commonly used words in the bidding field; use the completed bidding documents as the training data set to perform domain adaptation adjustment on the pre-trained language model; for different types of bidding projects, set the attention weight matrix and pay attention to the keywords and phrases related to the project type; evaluate the performance of the pre-trained language model through the cross-validation method, and select the version of the pre-trained language model with the highest evaluation score as the project type recognition model; deploy the project type recognition model as a semantic parsing service module to support batch text processing.

[0024] S1.3: Based on the text semantic feature matrix, extract the feature vectors representing the project type; S1.4: Identify numerical and descriptive information from the bidding requirement text, and extract the project scale parameters according to the preset parameter classification rules; S1.5: Use the professional dictionary library to perform term matching on the bidding requirement text and extract the technical requirement parameters; S1.6: Match the amount information in the bidding requirement text through regular expression patterns and extract the budget range parameters; S1.7: Integrate the project scale parameters, technical requirement parameters, and budget range parameters into a keyword set.

[0025] S2: Based on the feature vectors, use the cosine similarity algorithm to retrieve and match templates in the preset bidding document template library, generate a list of candidate templates, and perform label classification and parameter setting on the templates in the list of candidate templates.

[0026] Specifically include: S2.1: Use the cosine similarity algorithm to perform similarity matching between the feature vectors of the project and the feature vectors in each template of the preset bidding document template library to obtain the similarity matching value, where the preset bidding document template library includes a bidding announcement template, a tender notice template, an evaluation method template, and a contract clause template; Preferably, the specific formula for the similarity matching value is as follows: ; Where, is the similarity matching value, is the project feature vector, is the template feature vector, is the dimension of the feature vector, is the weight value of the i-th dimension of the feature, is the value of the i-th dimension of the project feature vector, is the value of the i-th dimension of the template feature vector, is the vector similarity weight coefficient, and its value range is [0, 0.7], is the category matching weight coefficient, and its value range is [0, 0.3], is the j-th category feature of the project, is the j-th category feature of the template, is the number of category features, is the category matching function.

[0027] S2.2: Based on the matching results, select the templates with similarity matching values greater than the preset threshold to generate a candidate template list; In an alternative embodiment, when the similarity matching value is less than the first preset threshold, the edit distance algorithm is used to recalculate the matching degree; when the similarity matching value is greater than or equal to the first preset threshold and less than the second preset threshold, secondary matching is performed in combination with word frequency features; when the similarity matching value is greater than or equal to the second preset threshold, it is directly included in the candidate template list.

[0028] It should be noted that the first preset threshold is determined based on the similarity statistical value of the historical bidding documents of this project category, and is obtained by analyzing the average similarity of a large number of similar projects; the second preset threshold is determined based on the comprehensive statistical value of the usage frequency and success rate of the template.

[0029] S2.3: Classify the tags of the templates in the candidate template list, where the tags include project type tags, applicable scope tags, and file type tags; In an alternative embodiment, when classifying the tags of the candidate template list, when the same template has multiple tags at the same time, the following rules are used to determine the priority: if there is a conflict between the project type tag and the applicable scope tag, the project type tag is preferred; if the applicable scope tag overlaps with the file type tag, the tag weight is dynamically adjusted according to the project budget scale; if there is a complementary relationship between the tags, the combined tag is retained.

[0030] In an alternative embodiment, when the number of templates in the candidate template list exceeds the upper limit threshold, the following screening mechanism is started: calculate the comprehensive score of the historical usage frequency and project similarity of each template; evaluate the complementarity between templates and retain the template combinations with strong complementarity; analyze the template update time and preferentially retain the templates updated recently.

[0031] S2.4: Based on the tag classification results, extract the changeable elements of the templates in the candidate template list and convert them into parameter placeholders; In an alternative embodiment, variable elements of the templates in the candidate template list are extracted, and special processing is performed when the following situations occur: If there is an associated dependency relationship between the variable elements, the relevant elements are combined into a parameter group; If the expression of the same element is inconsistent in different templates, a unified parameter mapping rule is established; If the value range of an element is restricted by multiple conditions, a parameter constraint relationship tree is constructed.

[0032] S2.5: Establish a mapping relationship between the parameter placeholder and the keyword set to form a parameterized configuration table; S2.6: Perform parameterized settings on the templates in the candidate template list according to the parameterized configuration table to generate a standardized template set; S3: Based on the keyword set, adapt the content of the templates in the candidate template list to generate an initial tender document.

[0033] Specifically, it includes: S3.1: Obtain the templates in the candidate template list, parse the template structure, and identify the placeholder tags, where the placeholder tags include the project scale parameter placeholder, the technical requirement parameter placeholder, and the budget range parameter placeholder; S3.2: Based on the parsing result of the template structure, establish a template structure index table for identifying the text positions and attribute information of each placeholder in the template; S3.3: Match the parameters in the keyword set with the placeholder tags in the template structure index table and perform replacement in sequence, where the keyword set includes the project scale parameter, the technical requirement parameter, and the budget range parameter; In an alternative embodiment, the project scale parameter in the keyword set is matched with the corresponding project scale parameter placeholder in the template structure index table, and the project scale parameter placeholder is sequentially replaced with the specific content of the project scale parameter; Similarly, the technical requirement parameter in the keyword set is matched with the corresponding technical requirement parameter placeholder in the template structure index table and replaced with the specific content of the technical requirement parameter; The budget range parameter in the keyword set is matched with the corresponding budget range parameter placeholder in the template structure index table and replaced with the specific content of the budget range parameter.

[0034] S3.4: For the replaced template content, detect whether there are unmatched placeholder tags; In an alternative embodiment, if an unmatched placeholder tag is detected, the unmatched placeholder tag is marked as abnormal information and output to the review prompt module for reference in the subsequent review stage; If no unmatched placeholder tag is detected, the replaced template content is used as the content body of the initial tender document.

[0035] S3.5: Based on the detection result, perform text format specification processing on the detected template content to form an initial tender document; S3.6: Compare the text of the initial tender document with the template structure of the original candidate template list, and update the label status of the template structure index table.

[0036] It should be noted that the marked replaced placeholders are in the filled state, and the un-replaced placeholders remain in the to-be-processed state.

[0037] S4: Conduct a segmented review of the initial tender document, adjust the keyword set based on the review opinions, and synchronously update it to the parametric database through the mapping relationship to generate the tender document.

[0038] Specifically include: S4.1: Divide the initial tender document into multiple text paragraphs according to the chapter structure to generate a segmented review list; S4.2: Construct a review rule library, perform rule matching on each text paragraph in the segmented review list, and mark the content points that do not meet the specifications; It should be noted that the review rule library includes format specifications, element integrity, and compliance checks.

[0039] In an alternative embodiment, when performing the format specification check, If the text paragraph is at the title level, perform the following rule matching: the font size of the first-level title must be bold No. 3; the font size of the second-level title must be bold No. 4; the title number format must conform to the hierarchical relationship of "1.", "1.1", "1.1.1"; if the title level exceeds four levels, it will be automatically marked and adjustment is recommended; If the text paragraph is the main content, perform the following rule matching: the font size of the main text must be Song typeface No. 4; the line spacing is 1.5 times; the first line of the paragraph is indented by 2 characters; if the length of the text paragraph exceeds 500 words, it will be processed in segments.

[0040] In an alternative embodiment, when performing the element integrity check, If the text paragraph belongs to the tender announcement chapter, verify the following elements: the completeness of the purchaser information must reach 85%; the number of words in the project overview description must be ≥200; the number of bidder qualification requirement clauses must be ≥5; If the text paragraph belongs to the business terms chapter, verify the following elements: verification of the completeness of the payment method clause; verification of the completeness of the performance bond clause; verification of the completeness of the acceptance criteria clause; If the element integrity <90%, a list of missing items will be automatically generated.

[0041] In an alternative embodiment, when performing the compliance check, If the text paragraph involves eligibility conditions, the following rules shall be executed for matching: the registered capital requirement must be ≤ 2 times the tender budget; the performance requirement period must be ≤ 3 years; the requirement for the number of professional personnel must match the project scale; If the text paragraph involves the scoring criteria, the following rules shall be executed for matching: the weight of a single scoring item must be ≤ 30%; the proportion of technical scores must be ≥ 50%; the proportion of objective scores must be ≥ 60%.

[0042] S4.3: Summarize the content points that do not meet the specifications to form an audit opinion set, where the audit opinion set includes problem descriptions, modification suggestions, and associated parameters; S4.4: Based on the audit opinion set, correct the corresponding parameters in the keyword set to generate an optimized keyword set; S4.5: Write the optimized keyword set into the parametric database and establish a parameter version tracking record; S4.6: Call the latest parameter values in the parametric database to adjust the initial tender document and output the tender document.

[0043] In summary, the present invention realizes the semantic parsing of the tender requirement text by introducing a project type recognition model, accurately extracts the project feature vector and keyword set, and effectively solves the problems of large manual understanding deviation and information omission in the preparation of traditional tender documents; by matching the preset tender document template library through the cosine similarity algorithm based on the feature vector, combined with label classification and parametric settings, it can quickly screen and generate a candidate template list, significantly improving the accuracy and applicability of template matching; based on the keyword set for content adaptation, it realizes the automatic generation of the initial tender document, reduces manual intervention, and improves the professionalism and standardization of content generation; adopts segmented audit and rule matching, combines the audit opinions to correct the keyword set and synchronously updates the parametric database, ensuring that the generated tender document is effectively guaranteed in terms of content integrity, standardization, and consistency; this method realizes an efficient closed-loop from requirements to results, significantly reducing the labor cost of tender document preparation and improving the quality and generation efficiency of tender documents.

[0044] Embodiment 2 This is the second embodiment of the present invention. Further, this embodiment also provides an intelligent tender document generation system based on text automation, including: An extraction module for performing semantic parsing on the tender requirement text through a project type recognition model to extract the feature vector and keyword set of the project, where the keyword set includes project scale parameters, technical requirement parameters, and budget range parameters; The template matching module, based on the feature vector, uses the cosine similarity algorithm to retrieve and match templates in a preset tender document template library, generates a list of candidate templates, and performs label classification and parameter setting on the templates in the list of candidate templates; The content adaptation module, based on the keyword set, adapts the content of the templates in the list of candidate templates to generate an initial tender document; The segmented review and generation module is used to perform segmented review on the initial tender document, adjust the keyword set in combination with the review opinions, and synchronously update it to the parameter database through the mapping relationship to generate a tender document.

[0045] This embodiment also provides an electronic device, which includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a multi-task edge computing resource scheduling method. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad set on the shell of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0046] This embodiment also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by the processor, it implements the method proposed in the above embodiment.

[0047] The storage medium proposed in this embodiment and the method proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be referred to in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0048] From the above description of the embodiments, those skilled in the art can clearly understand that the present invention can be implemented by means of software and necessary general-purpose hardware. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disc of a computer, etc., including several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute the method of the embodiments of the present invention.

[0049] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

[0050] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages.

[0051] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0052] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the function specified in the flowchart(s) Figure 1 a flowchart or flowcharts and / or block(s) Figure 1 a block or blocks.

[0053] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the function specified in the flowchart(s) Figure 1 a flowchart or flowcharts and / or block(s) Figure 1 a block or blocks.

[0054] Although the preferred embodiments of the present application have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic creative concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present application.

[0055] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. An intelligent generation method for tender documents based on text automation, characterized in that: including, performing semantic parsing on the tender requirement text through a project type recognition model to extract the feature vector and keyword set of the project; based on the feature vector, using the cosine similarity algorithm to retrieve and match templates in a preset tender document template library, generating a list of candidate templates, and performing label classification and parameter setting on the templates in the list of candidate templates; based on the keyword set, adapting the content of the templates in the list of candidate templates to generate an initial tender document; performing segmented review on the initial tender document, adjusting the keyword set in combination with the review opinions, and synchronously updating to the parameterized database through the mapping relationship to generate a tender document.

2. The intelligent generation method of bidding documents based on text automation according to claim 1, characterized in that: The method for generating the tender document is as follows: dividing the initial tender document into multiple text paragraphs according to the chapter structure to generate a segmented review list; constructing a review rule library, performing rule matching on each text paragraph in the segmented review list, and marking the content points that do not meet the specifications; summarizing the content points that do not meet the specifications to form a review opinion set, where the review opinion set includes problem descriptions, modification suggestions, and associated parameters; based on the review opinion set, correcting the corresponding parameters in the keyword set to generate an optimized keyword set; writing the optimized keyword set into the parameterized database and establishing a parameter version tracking record; invoking the latest parameter values of the parameterized database to adjust the initial tender document and output a tender document.

3. The intelligent generation method of tender documents based on text automation according to claim 2, characterized in that: The method for generating the initial tender document is as follows: obtaining the templates in the list of candidate templates, parsing the template structure, and identifying placeholder tags, where the placeholder tags include project scale parameter placeholders, technical requirement parameter placeholders, and budget range parameter placeholders; based on the parsing result of the template structure, establishing a template structure index table for identifying the text positions and attribute information of each placeholder in the template; matching the parameters of the keyword set with the placeholder tags of the template structure index table and performing replacement in sequence, where the keyword set includes project scale parameters, technical requirement parameters, and budget range parameters; detecting whether there are unmatched placeholder tags in the replaced template content; based on the detection result, performing text format specification processing on the detected template content to form an initial tender document; comparing the text of the initial tender document with the template structure of the original list of candidate templates to update the label status of the template structure index table.

4. The intelligent generation method of bidding documents based on text automation according to claim 3, wherein: The method for generating the list of candidate templates is as follows: using the cosine similarity algorithm to perform similarity matching on the feature vector of the project and the feature vectors in each template of the preset tender document template library to obtain a similarity matching value, where the preset tender document template library includes a tender announcement template, a tender notice template, an evaluation method template, and a contract clause template; based on the matching result, selecting the templates with the similarity matching value greater than the preset threshold to generate a list of candidate templates; performing label classification on the templates in the list of candidate templates, where the labels include project type labels, applicable scope labels, and document type labels; Based on the label classification results, extract the changeable elements of the templates in the candidate template list and convert them into parameter placeholders; Establish a mapping relationship between the parameter placeholders and the keyword set to form a parameterized configuration table; Perform parameterized settings on the templates in the candidate template list according to the parameterized configuration table to generate a standardized template set.

5. The intelligent generation method of tender documents based on text automation according to claim 4, characterized in that: Perform similarity matching between the feature vector of the project and the feature vectors of each template in the preset tender document template library, including: When the similarity matching value is less than the first preset threshold, recalculate the matching degree using the edit distance algorithm; When the similarity matching value is greater than or equal to the first preset threshold and less than the second preset threshold, perform secondary matching by combining word frequency features; When the similarity matching value is greater than or equal to the second preset threshold, directly include it in the candidate template list.

6. The intelligent generation method of bidding documents based on text automation according to claim 4, characterized in that: The extraction method of the project feature vector and the keyword set is as follows: Receive the tender requirement text input by the user, perform segmented parsing on the tender requirement text to form a text paragraph set; Input the text paragraph set into a pre-trained project type recognition model for semantic understanding, calculate the weight distribution of each word in the text to form a text semantic feature matrix; Based on the text semantic feature matrix, extract the feature vector representing the project type; Identify numerical and descriptive information from the tender requirement text, and extract project scale parameters according to the preset parameter classification rules; Use a professional dictionary library to perform term matching on the tender requirement text to extract technical requirement parameters; Extract the budget range parameters by pattern matching the amount information in the tender requirement text through regular expressions; Integrate the project scale parameters, the technical requirement parameters, and the budget range parameters into a keyword set.

7. The intelligent generation method of bidding documents based on text automation according to claim 6, wherein: The construction method of the project type recognition model is as follows: Adopt a three-layer encoder structure to establish a pre-trained language model based on BERT, where the pre-trained language model includes a number of attention heads; Based on the pre-trained language model, import a professional dictionary library in the tender field and perform annotation according to three dimensions: engineering, goods, and services, where the professional dictionary library includes a number of common words in the tender field; Use the completed tender documents as the training data set to perform domain adaptation adjustment on the pre-trained language model; For different types of tender projects, set an attention weight matrix and pay attention to keywords and phrases related to the project type; Evaluate the performance of the pre-trained language model through the cross-validation method, and select the version of the pre-trained language model with the highest evaluation score as the project type recognition model; Deploy the project type recognition model as a semantic parsing service module to support batch text processing.

8. An intelligent tender document generation system based on text automation, based on the intelligent tender document generation method based on text automation according to any one of claims 1 to 7, characterized in that: Including: An extraction module for performing semantic parsing on the tender requirement text through a project type recognition model to extract the feature vector and keyword set of the project, where the keyword set includes project scale parameters, technical requirement parameters, and budget range parameters; A template matching module that, based on the feature vector, uses the cosine similarity algorithm to retrieve and match templates in the preset tender document template library, generates a candidate template list, and performs label classification and parameterized settings on the templates in the candidate template list; The content adaptation module adapts the content of the templates in the candidate template list based on the keyword set to generate an initial bidding document; The segmented review and generation module is used to perform segmented review on the initial bidding document, adjust the keyword set in combination with the review opinions, and synchronously update it to the parameterized database through the mapping relationship to generate a bidding document.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent generation method of the bidding document based on text automation according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent generation method of the bidding document based on text automation according to any one of claims 1 to 7.

Images