Bidding document processing method and system
By applying bid error correction and dilution check models in the bid processing system, the bid documents are automatically processed, which solves the problem of inefficient screening of traditional bid documents, and achieves efficient and accurate error correction and dilution checking effects.
Patent Information
- Application Number
- CN202411938050.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-26
- Publication Date
- 2025-05-30
AI Technical Summary
Traditional bid documents are inefficient in the screening process, the content screening accuracy is not high, and relying on labor causes time-consuming and labor-consuming, and are easily affected by human factors.
Provide a bid handling method and system, by obtaining bid documents, performing bid error correction and plagiarism checking methods, and automatically processing bid documents using the trained bid error correction model and plagiarism checking model to improve the efficiency of error correction and plagiarism checking.
It realizes automatic error correction and plagiarism checking of bid documents, improves processing efficiency and accuracy, reduces manual intervention, and can screen out errors and duplicate content in bid documents more quickly.
Smart Images
Figure CN120068839A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of artificial intelligence technology and relates to a tender document processing method and system. Background Art
[0002] In the process of bidding, the tenderer publicly releases a tender notice to potential bidders, inviting eligible suppliers to participate in the competition. After receiving the tender documents, the bidder needs to submit a tender document according to its requirements. At the same time, in order to ensure the originality and accuracy of the tender document, it is necessary to screen the content of the tender document. However, traditional tender documents face problems such as low processing efficiency and low accuracy in content screening during the screening process. Summary of the Invention
[0003] The purpose of this application is to provide a tender document processing method and system for improving the processing efficiency of tender documents.
[0004] In a first aspect, this application provides a tender document processing method. The tender document processing method includes: obtaining a plurality of tender documents; performing a tender document error correction method on the tender documents to obtain corrected tender documents, and / or performing a tender document duplication check method on the tender documents to obtain a tender document duplication check result; the process of performing a tender document error correction method on the tender documents to obtain corrected tender documents includes: reading and splitting the tender documents by using a tender document error correction model to obtain tender document sentences; wherein, the training method of the tender document error correction model includes: extracting training tender documents from a training set to obtain correct tender document texts; generating abnormal tender document texts according to the correct tender document texts; training the tender document error correction model by using the correct tender document texts and the abnormal tender document texts; comparing the tender document sentences through the tender document error correction model to obtain the tender document error correction result; the process of performing a tender document duplication check method on the tender documents to obtain the tender document duplication check result includes: removing the text content similar to the tender documents from the tender documents to obtain processed tender documents; combining multiple processed tender documents in pairs to obtain multiple tender document combinations; comparing the text content of each tender document combination by using a tender document duplication check model to obtain a similarity duplication check result; according to the similarity duplication check result, comparing the meanings of similar tender document sentences in multiple processed second tender documents in pairs to obtain the tender document duplication check result.
[0005] In an implementation manner of the first aspect, the process of extracting the correct tender document text from the data set of the tender documents includes: extracting training tender documents from the training set; extracting the training tender documents to obtain multiple paragraphs of the tender documents; segmenting the paragraphs by using a sentence segmentation algorithm to obtain tender document sentences; analyzing and filtering the tender document sentences to obtain the correct tender document text.
[0006] In one implementation of the first aspect, the process of generating bid exception text based on the correct bid text includes: automatically generating the bid exception text by performing data augmentation operations and context-related error simulations on the correct bid text.
[0007] In one implementation of the first aspect, the process of removing the text content similar to the bidding document from the bid document and obtaining the processed bid document includes: using regular expressions and a word segmentation library to extract and filter the text content similar to the bidding document in the bid document, so as to obtain the processed bid document.
[0008] In one implementation of the first aspect, the process of using a bid duplication checking model to compare the text content of each bid document combination and obtain the similarity duplication checking result includes: using the bid duplication checking model to compare the text content of each bid document combination to obtain a text content similarity value; taking the text content with the text content similarity value greater than the similarity threshold as the similarity duplication checking result.
[0009] In one implementation of the first aspect, the bid processing method further includes: grouping the similarity duplication checking results through a clustering algorithm to obtain bid statements corresponding to different themes.
[0010] In one implementation of the first aspect, according to the similarity duplication checking result, the process of comparing the semantic meanings of similar bid statements in multiple processed second bid documents pairwise to obtain the bid duplication checking result includes: comparing the character differences of similar bid statements in multiple processed second bid documents to obtain a semantic comparison result; screening the similar bid statements according to the semantic comparison result to obtain the bid duplication checking result.
[0011] In the second aspect, the present application provides a bid processing system, which includes: an acquisition module for acquiring multiple bid documents; a bid error correction module and / or a bid duplication checking module, where the bid error correction module is used to execute the above-mentioned bid error correction method, and the bid duplication checking module is used to execute the above-mentioned bid duplication checking method.
[0012] In one implementation of the second aspect, the bid error correction module includes a bid error correction execution unit and a bid error correction model training unit, where the bid error correction execution unit is used to execute the above-mentioned bid error correction method, and the bid error correction model training unit is used to execute the above-mentioned training method of the bid error correction model.
[0013] In an implementation of the second aspect, the bid document duplicate checking module includes: a deletion unit, configured to delete the text content similar to the bidding document from the bid document to obtain a processed bid document; a combination unit, configured to pairwise combine multiple processed bid documents to obtain multiple combinations of bid documents; a text content comparison unit, configured to use a bid document duplicate checking model to compare the text content of each combination of bid documents to obtain a similarity duplicate checking result; and a bid document duplicate checking result obtaining unit, configured to compare the meanings of pairwise similar bid sentences in multiple processed second bid documents according to the similarity duplicate checking result to obtain the bid document duplicate checking result.
[0014] As described above, the bid document processing method, system, medium, and electronic device provided in this application have the following beneficial effects:
[0015] The bid document processing method provided in this application can process bid documents, correct incorrect bid sentences in bid documents based on a trained bid document error correction model, and can also perform similarity duplicate checking on the text content in bid documents to obtain duplicate checking results of multiple bid documents. Through the bid document processing method of this application, bid documents can be automatically processed, improving the efficiency of bid document error correction and duplicate checking. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It shows a schematic process diagram of the bid document processing method described in an embodiment of this application.
[0017] Figure 2 It shows a schematic process diagram of obtaining a corrected bid document described in an embodiment of this application.
[0018] Figure 3 It shows a schematic process diagram of the bid document error correction model training method described in an embodiment of this application.
[0019] Figure 4 It shows a schematic process diagram of obtaining a bid document duplicate checking result described in an embodiment of this application.
[0020] Figure 5 It shows a schematic process diagram of obtaining the correct text of a bid document described in an embodiment of this application.
[0021] Figure 6 It shows a schematic process diagram of obtaining a similarity duplicate checking result described in an embodiment of this application.
[0022] Figure 7 It shows a schematic process diagram of obtaining a bid document duplicate checking result described in an embodiment of this application.
[0023] Figures 8 - 9 It shows a schematic structural diagram of the T5 model training described in an embodiment of this application.
[0024] Figure 10 It shows a schematic structural diagram of the tender processing system described in the embodiments of the present application.
[0025] Explanation of component numbers
[0026] 1 Tender processing system
[0027] 11 Acquisition module
[0028] 12 Tender error correction module
[0029] 13 Tender duplicate checking module
[0030] Steps S11 - S12
[0031] Steps S21 - S22
[0032] Steps S31 - S33
[0033] Steps S41 - S44
[0034] Steps S51 - S54
[0035] Steps S61 - S62
[0036] Steps S71 - S72
[0037] Steps S81 - S82 Specific implementation manners
[0038] The following uses specific examples to illustrate the implementation manners of the present application. Those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in this specification. The present application can also be implemented or applied through other different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, without conflict, the following embodiments and the features in the embodiments can be combined with each other.
[0039] It should be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the present application in a schematic manner. Therefore, only the components related to the present application are shown in the diagrams, rather than being drawn according to the number, shape, and size of the components in actual implementation. The types, quantities, and proportions of the components in actual implementation can be arbitrarily changed, and the component layout type may also be more complex.
[0040] In the process of bidding, the tenderer publicly releases an invitation to tender to potential bidders through a tender notice, inviting eligible suppliers to participate in the competition. The tender notice details important information such as the project requirements, technical specifications, and contract terms. After receiving the tender documents, the bidder is required to submit a tender document according to the requirements. At the same time, in order to ensure the originality and accuracy of the tender document, content screening of the tender document is required. However, in the traditional tender document screening process, the screening of errors and duplicate content often relies on manual work. This method not only takes time and effort but is also easily affected by human factors, resulting in low screening efficiency. In addition, with the increase in tender projects and tender documents, the manual screening method has become difficult to meet the growing demand, facing problems such as low processing efficiency and accuracy of content screening in the screening process.
[0041] At least for the above problems, the following embodiments of the present application provide a method for processing tender documents.
[0042] The following will describe in detail the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application.
[0043] Figure 1 It shows a schematic process diagram of the method for processing tender documents in an embodiment of the present application. As Figure 1 shown, the method for processing tender documents includes:
[0044] S11, obtaining a plurality of tender documents and asynchronously storing them temporarily. The tender documents may include tender documents and bid invitation documents.
[0045] S12, performing a tender document error correction method to correct the tender documents to obtain corrected tender documents, and / or performing a tender document duplicate check method to check the tender documents for duplicates to obtain tender document duplicate check results. Among them, the corrected tender documents contain error markings of the original tender documents, and the corrected tender documents can be manually modified or automatically modified using a tender document error correction model.
[0046] It should be noted that the tender document processing method includes three processing methods: first correcting errors and then checking for duplicates, first checking for duplicates and then correcting errors, and executing separately. Among them, the processing method of correcting errors and then checking for duplicates is: first, the tender document error correction method is used to perform a comprehensive error check and correction on the original tender document document to obtain a tender document document that has been corrected. Subsequently, the tender document duplication checking method is executed to check for duplicates on the tender document document to obtain the tender document duplication checking result. This processing method can ensure that the tender document document undergoes strict quality control before submission. The processing method of checking for duplicates first and then correcting errors is: first, the tender document duplication checking method is executed to check for duplicates on the tender document document to obtain the tender document duplication checking result. After understanding the duplication checking situation of the tender document document, the tender document error correction method is executed to correct the tender document document to obtain the corrected tender document document. This processing method helps to identify and correct possible duplicate content. The processing method of executing separately is: executing the tender document error correction method separately to correct the tender document document, or executing the tender document duplication checking method separately to check for duplicates on the tender document document. This processing method can select the most appropriate combination of steps according to the specific situation.
[0047] See also Figure 2 The process of executing the tender document error correction method to correct the tender document file and obtaining the corrected tender document file includes:
[0048] S21, using a tender error correction model to read and split the tender file to obtain tender statements.
[0049] S22, comparing the bid statement with the bid error correction model to obtain the bid error correction result.
[0050] Among them, see Figure 3 , the training method of the bid error correction model includes:
[0051] S31, extracting the training bid documents in the training set to obtain the correct text of the bid documents.
[0052] S32, generating an abnormal tender text according to the correct tender text.
[0053] S33, using the correct tender text and the abnormal tender text to train the tender error correction model.
[0054] See also Figure 4 The process of executing the bid duplication checking method to check the bid duplication of the bid document and obtaining the bid duplication checking result includes:
[0055] S41, removing text contents in the bid document that are similar to the bidding document, and obtaining a processed bid document.
[0056] S42, combining the plurality of processed bid documents in pairs to obtain a plurality of bid document combinations.
[0057] S43. Use the tender document duplication checking model to compare the text content of each tender document combination, and obtain the similarity duplication checking result.
[0058] S44. According to the similarity duplication checking result, compare the meanings of pairwise similar tender document sentences in multiple processed second tender documents to obtain the tender document duplication checking result.
[0059] As can be seen from the above description, the tender document processing method provided by this application can process tender documents, correct the error sentences of tender documents based on the trained tender document error correction model, and can also perform similarity duplication checking on the text content in tender documents to obtain the duplication checking results of multiple tender documents. Through the tender document processing method of this application, tender documents can be automatically processed, improving the efficiency of tender document error correction and duplication checking.
[0060] Figure 5 It is shown as a schematic diagram of the process of obtaining the correct tender document text in an embodiment of this application. As Figure 5 shown, the process of extracting the correct tender document text from the dataset of the tender document includes:
[0061] S51. Extract the training tender documents from the training set.
[0062] S52. Extract the training tender documents to obtain multiple paragraphs of the tender documents.
[0063] S53. Use the sentence segmentation algorithm to segment the paragraphs to obtain tender document sentences.
[0064] S54. Analyze and filter the tender document sentences to obtain the correct tender document text.
[0065] Exemplarily, use the sentence segmentation algorithm to segment the paragraphs in the tender document, determine the end markers of the sentences, such as full stops, question marks, exclamation marks. Use regular expressions to match the end markers of the sentences. When the regular expressions match the end markers of the sentences, it will extract the end markers of the sentences and the text content before the end markers of the sentences as a complete sentence. Verify the extracted sentences to ensure they are correct sentences and not wrongly segmented.
[0066] In an embodiment of this application, the process of generating the tender document abnormal text according to the correct tender document text includes: automatically generating the tender document abnormal text by performing data augmentation operations and context-related error simulations on the correct tender document text. Generate tender document abnormal sentences containing various error types by processing the correct tender document sentences. Perform data augmentation operations on the sentences in the correct tender document text to simulate the error sentences that may occur in actual use.
[0067] Exemplarily, the data augmentation operations include transposing adjacent Chinese characters, replacing homophones, randomly adding or deleting characters. Context-related error simulation includes adjusting word order, misusing technical terms, semantic errors, logical errors, formatting errors, etc. Transposing adjacent Chinese characters simulates typing errors by randomly selecting two adjacent Chinese characters in a sentence and swapping their positions. Replacing homophones simulates dictation errors by replacing a certain Chinese character in a sentence with a Chinese character with a similar pronunciation. Randomly adding or deleting characters simulates text errors by randomly adding or deleting characters in a sentence. Adjusting word order simulates grammar errors by changing the order of words in a sentence. Misusing technical terms simulates errors in professional knowledge by replacing a technical term in a sentence with an inappropriate synonym. Semantic errors simulate comprehension errors by replacing words in a sentence to change its semantics. Logical errors simulate reasoning errors by changing the logical relationship in a sentence. Formatting errors simulate typesetting errors by changing the format in a sentence, such as the conversion between numbers and words.
[0068] Through data augmentation operations and context-related error simulation, a rich dataset of abnormal sentences can be generated to support the training and testing of the bid correction model. These abnormal sentences not only contain simple spelling and grammar errors, but also more complex semantic and logical errors, which helps the model better understand and correct errors in bid texts.
[0069] In a tender project, according to the tender documents issued by the tenderer, the bidder will submit a proposal for a certain service. The bidder may make various errors during the preparation of the bid, such as spelling mistakes, grammar errors, inappropriate word usage, formatting errors, etc. To ensure the quality and accuracy of the bid documents, the bidder can use the bid correction method of this application to process the bid documents.
[0070] First, the bidder uploads the prepared bid document through the model prediction interface provided by the system. Subsequently, the system performs intelligent identification and analysis on the bid document, automatically identifies the bid document using the trained bid correction model, accurately identifies various errors in the bid document, and displays the correction results in a visual way such as text highlighting and error type display. The bidder can view the correction results and modify the document according to the modification suggestions provided by the model. Through the bid correction method, the bidder can ensure the quality and accuracy of the bid document and improve the success rate of the bid. The automatic identification and correction process of the model for the document is carried out efficiently in the background, while the bidder receives the correction results in real time through the user interface. The correction results are presented in a visual form in an intuitive way such as text highlighting and error type display. For example, the incorrect words are highlighted, and the error types such as grammar errors and word usage errors are clearly marked, enabling the bidder to quickly locate and understand the specific location and nature of the errors.
[0071] When reviewing the error correction results, the tenderer can make targeted modifications to the document according to the modification suggestions provided by the model. These suggestions not only point out the errors but also provide possible correction solutions, greatly reducing the difficulty and error rate of the tenderer's self-modification. In this way, the tenderer can efficiently correct the errors in the tender document, improving the overall quality and professionalism of the document. Finally, through the optimization of the tender document error correction method, the tender document submitted by the tenderer to the tender inviting party is not only accurate in content but also standardized in format and clear in expression, greatly increasing the success rate of the tender. The application of the tender document error correction method not only enhances the tenderer's competitive advantage but also reflects its remarkable effect in improving work efficiency and document quality.
[0072] The tenderer can pre-train the tender document error correction model, upload the training tender document, and use regular expressions and the word segmentation library to extract sentences from the training tender document, removing format information such as line breaks and spaces. Perform data augmentation operations such as adjacent Chinese character transposition, homophone replacement, random character addition and deletion on the extracted correct sentences, and error simulation related to context such as word order adjustment and misuse of professional terms to generate abnormal sentences. Use the extracted correct sentences and the generated abnormal sentences to train the tender document error correction model.
[0073] Before preparing the tender document, the tenderer can pre-train the tender document error correction model so that the model can better adapt to the text processing requirements of a specific field. The tenderer first collects and organizes a series of training samples related to the tender document, uses regular expressions and the word segmentation library to extract sentences from the preprocessed document content to obtain correct tender document statements. Perform data augmentation operations such as adjacent Chinese character transposition, homophone replacement, random character addition and deletion on the extracted correct sentences to simulate common input errors, and introduce error simulation related to context such as word order adjustment and misuse of professional terms to simulate complex grammar and professional knowledge errors. Generate a series of abnormal sentences containing different types of errors through data augmentation operations and context-related error simulation. These abnormal sentences are still semantically reasonable but contain various errors, which helps to improve the generalization ability of the tender document error correction model. Use the extracted correct sentences and the generated abnormal sentences to train the tender document error correction model. By learning these sentences, the model can identify and correct errors in the tender document text, thus improving the accuracy and efficiency of error correction. By pre-training the tender document error correction model, the tenderer can ensure that the model can better adapt to the characteristics and error types of the tender document text. This method not only improves the performance of the model but also saves the time and effort of the tenderer in correcting the tender document during the actual tender process.
[0074] In an embodiment of the present application, the process of removing the text content similar to the bidding document from the tender document and obtaining the processed tender document includes: using regular expressions and a word segmentation library to extract and filter the text content similar to that in the bidding document in the tender document, so as to obtain the processed tender document.
[0075] Figure 6 It shows a schematic diagram of the process of obtaining the similarity check result in an embodiment of the present application. As Figure 6 shown, the process of using the tender document similarity check model to compare the text content of each tender document combination and obtain the similarity check result includes:
[0076] S61, using the tender document similarity check model to compare the text content of each tender document combination and obtain the text content similarity value. Based on word frequency, syntactic structure, etc., use the tender document similarity check model to calculate the text content similarity in each tender document combination, and obtain the similarity value between sentences in each tender document combination. The similarity value reflects the semantic similarity degree between two text contents.
[0077] S62, taking the text content with the text content similarity value greater than the similarity threshold as the similarity check result. For example, taking the text content with a similarity degree exceeding 80% as the similarity check result, but the present application is not limited thereto.
[0078] In an embodiment of the present application, the tender document processing method further includes: grouping the similarity check results through a clustering algorithm to obtain the tender document sentences corresponding to different themes. The clustering algorithm is, for example, DBSCAN or K-means. The clustering algorithm can automatically identify the repeated patterns or themes between similar sentences and group them together. Each group represents a repeated pattern or theme, and the sentences within the group have a high semantic similarity.
[0079] Figure 7 It shows a schematic diagram of the process of obtaining the tender document similarity check result in an embodiment of the present application. As Figure 7 shown, according to the similarity check result, the process of comparing the meanings of two similar tender document sentences in multiple processed second tender documents to obtain the tender document similarity check result includes:
[0080] S71, comparing the character differences of the similar tender document sentences in multiple processed second tender documents to obtain the semantic comparison result.
[0081] S72. Screen the similar tender document sentences according to the semantic comparison result to obtain the tender document duplication check result. The tender document sentences corresponding to different themes can be displayed in a visual way such as text highlighting and similarity score display, so as to quickly identify duplicate content. According to the tender document duplication check result, the user can delete and modify the duplicate content and similar content to ensure the originality and compliance of the tender document.
[0082] In a tender project, multiple companies submit bids for the tender documents of the same tenderer. To ensure the fairness and originality of the bids, the tenderer can use the tender document duplication check technology to check the duplication of multiple tender documents and require the tenderers to make rectifications according to the duplication check result. In the case of project publicity, the tenderer can also use the tender document duplication check technology to conduct duplication check and analysis on the tender documents of other companies to ensure the originality and independence of its own tender document. Next, the tender document duplication check method of this application will be described by taking the duplication check analysis of the tenderer as an example:
[0083] The tenderer issues a tender project and provides tender documents. Multiple companies submit their respective tender documents according to the requirements of the tender documents. First, the tenderer uses the tender document duplication check model to process the tender documents and filters out the content similar to the tender documents in the tender documents. Next, the tenderer combines the tender documents of multiple companies in pairs and uses the tender document duplication check model to conduct similarity comparison. The similarity score between sentences in each combination is calculated through the tender document duplication check model, and the sentences with similarity greater than the similarity threshold are marked as duplicate content. The text content of the same theme is divided into a group by using the clustering algorithm. Semantic comparison is carried out on the duplicate text content, the character difference between two sentences is compared, and it is determined whether the meanings of the two sentences are consistent. Finally, the text similarity and similar sentences of the tender document duplication check are output.
[0084] Both the tender document error correction model and the tender document duplication check model can be trained based on the T5 model. Figures 8 - 9 It is a structural schematic diagram for the training of the T5 model. The T5 model is a pre-trained language model that adopts the model architecture of encoder-decoder, uniformly uses a Transformer network to process all tasks, and can transform natural language processing into a text-to-text conversion problem. Transformer processes sequence data through the self-attention mechanism and positional encoding, does not require a recursive network, supports parallel computing, and greatly improves the training efficiency.
[0085] The T5 model uses the AdaFactor optimizer for model training and greedy decoding for decoding. When performing model pre-training, it is trained for 524k steps on the training set, with a maximum sequence length of 512 and a batch size of 128. In this way, the model processes approximately 34B tokens in total. At the same time, the reciprocal of the root mean square is used as the learning rate strategy, with a warmup step value of 10,000 and an initial learning rate of 0.01. The learning rate strategy ensures adaptive adjustment of the learning rate in the case of uncertain training steps, enabling the model to adjust the learning speed according to the training progress, thereby improving training efficiency. When the model is in the fine-tuning stage, each task is fine-tuned for 262k steps, using a constant learning rate of 0.001, saving the weights every 5k steps, and saving the model with the best performance on the validation set. This strategy helps to quickly find the optimization points for each task and ensure that the model performs optimally on various tasks.
[0086] In the process of correcting errors and checking for duplicates in the tender documents, models such as CSC (Chinese Speech and Language, Chinese speech processing model) and LoRA (Low-Rank Adaptation, fine-tuning) can also be used for auxiliary processing. CSC is used to optimize the training of the Chinese content in the tender documents. By considering the uniqueness of word segmentation methods and word structures in Chinese, the model's understanding of Chinese text and the accuracy of Chinese text generation are improved. LoRA is used to effectively fine-tune the tender document error correction model and the tender document duplicate checking model without changing the number of model parameters, reducing the model's computational cost and storage requirements.
[0087] In summary, the tender document processing method provided by this application can train tender documents, correct error statements in tender documents through the trained tender document error correction model, and can also check for similarity in the text content of tender documents to obtain duplicate checking results for multiple tender documents. Through the tender document processing method of this application, tender documents can be automatically processed, improving the efficiency of tender document error correction and duplicate checking.
[0088] The protection scope of the tender document processing method described in the embodiments of this application is not limited to the execution order of the steps listed in this embodiment. Any solution achieved by adding or subtracting steps of the prior art and replacing steps according to the principles of this application is included in the protection scope of this application.
[0089] The embodiments of this application also provide a tender document processing system. The tender document processing system can implement the tender document processing method described in this application. However, the implementation devices of the tender document processing method described in this application include but are not limited to the structure of the tender document processing system listed in this embodiment. Any structural deformation and replacement of the prior art made according to the principles of this application are included in the protection scope of this application.
[0090] Figure 10 Shown is a schematic structural diagram of a tender document processing system in an embodiment of the present application. As Figure 10 shown, the tender document processing system 1 includes: an acquisition module 11, a tender document error correction module 12, and / or a tender document duplicate check module 13. Among them, the acquisition module 11 is used to acquire multiple tender document files. The tender document error correction module 12 is used to execute the above-mentioned tender document error correction method. The tender document duplicate check module 13 is used to execute the above-mentioned tender document duplicate check method. The tender document error correction module includes a tender document error correction execution unit and a tender document error correction model training unit. The tender document error correction execution unit is used to execute the above-mentioned tender document error correction method, and the tender document error correction model training unit is used to execute the above-mentioned tender document error correction model training method. The tender document duplicate check module includes: an elimination unit, a combination unit, a text content comparison unit, and a tender document duplicate check result acquisition unit. Among them, the elimination unit is used to eliminate the text content similar to the tender invitation document in the tender document file to obtain a processed tender document file. The combination unit is used to combine multiple processed tender document files in pairs to obtain multiple tender document file combinations. The text content comparison unit is used to use a tender document duplicate check model to compare the text content of each tender document file combination to obtain a similarity duplicate check result. The tender document duplicate check result acquisition unit is used to, according to the similarity duplicate check result, compare the meanings of similar tender document statements in multiple processed second tender document files in pairs to obtain the tender document duplicate check result. It should be noted that Figure 10 each module in the shown tender document processing system 1 corresponds one by one to Figure 1 the steps in the above-mentioned tender document processing method, which will not be elaborated here.
[0091] In several embodiments provided by the present application, it should be understood that the disclosed system, device or method can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of modules / units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple modules or units can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the shown or discussed coupling or direct coupling or communication connection to each other can be through some interfaces, and the indirect coupling or communication connection of devices or modules or units can be in electrical, mechanical or other forms.
[0092] The modules / units described as separate components may or may not be physically separated, and the components shown as modules / units may or may not be physical modules, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules / units can be selected according to actual needs to achieve the objectives of the embodiments of the present application. For example, in the embodiments of the present application, the functional modules / units can be integrated in one processing module, or each module / unit can exist physically alone, or two or more modules / units can be integrated in one module / unit.
[0093] Those of ordinary skill in the art should further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.
[0094] The descriptions of the processes or structures corresponding to the above respective drawings each have their own emphases. For parts not detailed in a certain process or structure, reference can be made to the relevant descriptions of other processes or structures.
[0095] The above embodiments are only illustrative of the principles and effects of the present application and are not used to limit the present application. Any person familiar with this technology can modify or change the above embodiments without departing from the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by those of ordinary skill in the art within the spirit and technical ideas disclosed in the present application should still be covered by the claims of the present application.
Claims
1. A method for processing a tender document, characterized in that: The bid processing method includes: Obtain multiple tender documents; Execute the tender document error correction method to correct the tender document to obtain the corrected tender document, and / or Execute the bid duplication checking method to check the bid duplication of the bid document to obtain the bid duplication checking result; The process of executing the tender document error correction method to correct the tender document and obtain the corrected tender document includes: The tender document file is read and split using a tender document error correction model to obtain tender document statements; wherein the training method of the tender document error correction model includes: Extract the training bid documents in the training set to obtain the correct text of the bid documents; Generate abnormal tender text according to the correct tender text; Using the correct text of the tender document and the abnormal text of the tender document to train the tender document error correction model; Compare the bid statement with the bid error correction model to obtain the bid error correction result; The process of executing the bid duplication checking method to check the bid duplication of the bid document and obtaining the bid duplication checking result includes: Eliminate the text content in the bidding document that is similar to the bidding document to obtain a processed bidding document; Combining the processed bid documents in pairs to obtain multiple bid document combinations; Use the bid duplication checking model to compare the text content of each bid document combination and obtain the similarity duplication checking results; According to the similarity checking result, the meanings of similar bidding statements in the plurality of processed second bidding documents are compared in pairs to obtain the bidding checking result.
2. The bid processing method according to claim 1, characterized in that: The process of extracting the data set of the tender document and obtaining the correct text of the tender document includes: Extracting training bid documents from the training set; Extracting the training bid document to obtain multiple paragraphs of the bid document; Segment the paragraphs using a sentence segmentation algorithm to obtain bid statements; The bid statements are analyzed and filtered to obtain the correct text of the bid.
3. The bid processing method according to claim 1, characterized in that: The process of generating abnormal tender text according to the correct tender text includes: automatically generating the abnormal tender text by performing data enhancement operation and context-related error simulation on the correct tender text.
4. The bid processing method according to claim 1, characterized in that: The process of removing the text content in the bidding document that is similar to the bidding document and obtaining the processed bidding document includes: Regular expressions and word segmentation libraries are used to extract and filter text content in the bid document that is similar to the tender document, so as to obtain the processed bid document.
5. The bid processing method according to claim 1, characterized in that: The process of using the bid duplication checking model to compare the text content of each bid document combination and obtain the similarity duplication checking results includes: Using the bid duplication checking model to compare the text content of each bid document combination, and obtain a text content similarity value; The text content whose text content similarity value is greater than the similarity threshold is used as the similarity checking result.
6. The bid processing method according to claim 1, characterized in that: The bid processing method also includes: grouping the similarity check results by a clustering algorithm to obtain bid statements corresponding to different topics.
7. The bid processing method according to claim 1, characterized in that: According to the similarity checking result, the meaning of similar bidding documents in the plurality of processed second bidding documents is compared between two of them, and the process of obtaining the bidding document checking result includes: Comparing the character differences of similar bidding statements in the plurality of processed second bidding documents to obtain a semantic comparison result; The similar bid documents sentences are screened according to the semantic comparison result to obtain the bid duplication checking result.
8. A tender document processing system, characterized in that: The tender document processing system comprises: An acquisition module is used to acquire multiple bidding documents; A tender document error correction module and / or a tender document duplication checking module, wherein the tender document error correction module is used to execute the tender document error correction method described in any one of claims 1-7, and the tender document duplication checking module is used to execute the tender document duplication checking method described in any one of claims 1-7.
9. The tender document processing system according to claim 8, characterized in that: The bid error correction module includes a bid error correction execution unit and a bid error correction model training unit. The bid error correction execution unit is used to execute the bid error correction method described in any one of claims 1 to 7, and the bid error correction model training unit is used to execute the bid error correction model training method described in claim 1.
10. The tender document processing system according to claim 8, characterized in that: The bid duplication checking module comprises: A removal unit, used to remove text content in the bid document that is similar to the bidding document, and obtain a processed bid document; A combining unit, used for combining the plurality of processed tender documents in pairs to obtain a plurality of tender document combinations; A text content comparison unit is used to compare the text content of each bid document combination using a bid duplication checking model to obtain a similarity duplication checking result; The unit for obtaining the result of checking for duplicate tender documents is used to compare the meanings of similar tender documents sentences in pairs in the plurality of processed second tender documents files according to the result of checking for duplicate tender documents with similarity, so as to obtain the result of checking for duplicate tender documents.