An electronic bidding document detection method and device, computer equipment and medium
By using document classification and keyword extraction models to automatically detect electronic tender documents, the problem of insufficient accuracy of manual detection has been solved, achieving efficient and accurate tender document detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 北京筑龙信息技术有限责任公司
- Filing Date
- 2022-08-26
- Publication Date
- 2026-05-12
AI Technical Summary
In existing technologies, the inspection of electronic tender documents relies on manual checks, which leads to insufficient accuracy and a high risk of errors.
This paper employs a document classification model and a target keyword extraction model to classify and extract keywords from images of electronic bidding documents, determine whether they contain preset standard keywords, identify document categories through the document classification model, and use the target keyword extraction model to identify and structure text content, thereby achieving automated detection.
It improves the accuracy of electronic bid document detection, ensures the accuracy and consistency of detection results, and reduces human error.
Smart Images
Figure CN115424272B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of electronic bidding, and more specifically, to a method, apparatus, computer equipment, and medium for detecting electronic bid documents. Background Technology
[0002] Electronic bidding is based on traditional bidding, but it involves digitizing the bid documents and transmitting and publishing them over the network. In the current technology, when judging whether the electronic bid documents meet the bidding requirements, they are usually checked manually. That is, relevant personnel in charge of bid document management read the contents of the electronic bid document line by line and page by page, and check the contents of the bid document after summarizing the conditions in the electronic bid document.
[0003] The inventors discovered in their research that reviewing tender documents typically requires precise statistical analysis of a large amount of text and images. However, when reviewing by humans, the sheer volume of tender documents and the lack of experience among the personnel can lead to erroneous results, thus reducing the accuracy of electronic tender document review. Summary of the Invention
[0004] In view of this, the purpose of the present invention is to provide a method, apparatus, computer equipment and medium for detecting electronic bid documents, so as to improve the accuracy of detecting electronic bid documents.
[0005] In a first aspect, embodiments of this application provide a method for detecting electronic bid documents, the method comprising:
[0006] For each file image in the electronic bid document image, the file image is input into the trained file classification model to obtain the target file category to which the file image belongs. The electronic bid document image is an image obtained by converting at least one page contained in the electronic bid document to be detected from the original format to the image format.
[0007] The image of the file is input into a trained target keyword extraction model to obtain at least one target keyword contained in the image of the file. The target keyword extraction model is a key extraction model among all keyword extraction models that has the same category identifier as the target file category to which the image of the file belongs.
[0008] Determine whether the at least one target keyword contains all the keywords in the at least one preset standard keyword;
[0009] If the at least one target keyword contains all the keywords of the at least one standard keyword, then the image in the file is marked as a qualified image;
[0010] Determine whether each file image in the electronic bid document images is a qualified image;
[0011] If each file image in the electronic bid document is a qualified image, then the electronic bid document to be tested will be marked as a qualified document.
[0012] Optionally, before determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword, the method further includes:
[0013] The at least one standard keyword is extracted from the standard electronic tender document according to a preset regular expression.
[0014] Optionally, the target keyword extraction model includes a target text recognition sub-model for text extraction and text localization, and a target structured sub-model for structuring text according to text category;
[0015] The step of inputting the image of the file into a trained target keyword extraction model to obtain at least one target keyword contained in the image of the file includes:
[0016] The file image is input into the target text recognition sub-model to obtain at least one target text contained in the file image and the coordinate information of each target text in the file image;
[0017] The coordinate information of the at least one target text and each of the at least one target text in the file image is input into the target structured sub-model to obtain at least one target keyword contained in the file image.
[0018] Optionally, after determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword, the method further includes:
[0019] If the at least one target keyword does not contain all the keywords of the at least one standard keyword, then the image file will be marked as an unqualified image.
[0020] Optionally, after determining whether each file image in the electronic bid document images is a qualified image, the method further includes:
[0021] If any of the images in the electronic bid document are unqualified, the electronic bid document to be tested will be marked as unqualified.
[0022] Optionally, after marking the electronic tender document to be inspected as an unqualified document, the method includes:
[0023] The page will display any non-compliant images from the electronic bid document.
[0024] And / or, display the keywords to be displayed from the unqualified images in the electronic tender document images on the page, wherein the keywords to be displayed are those that exist in the at least one standard keyword but do not exist in the at least one target keyword.
[0025] Optionally, after marking the electronic tender document to be tested as a qualified document, the method further includes:
[0026] Display the qualified images from the electronic bid document images on the page;
[0027] And / or, display at least one target keyword from the qualified images in the electronic tender document images on the page.
[0028] Secondly, embodiments of this application provide an electronic bid document detection device, the device comprising:
[0029] The document category determination module is used to input each document image in the electronic bid document image into a trained document classification model to obtain the target document category to which the document image belongs. The electronic bid document image is an image obtained by converting at least one page contained in the electronic bid document to be detected from the original format to an image format.
[0030] The target keyword determination module is used to input the file image into a trained target keyword extraction model to obtain at least one target keyword contained in the file image. The target keyword extraction model is a key extraction model among all keyword extraction models that has the same category identifier as the target file category to which the file image belongs.
[0031] The first judgment module is used to determine whether the at least one target keyword contains all the keywords in the at least one preset standard keyword;
[0032] The first image tagging module is used to tag the image of the file as a qualified image if the at least one target keyword contains all the keywords of the at least one standard keyword;
[0033] The second judgment module is used to determine whether each file image in the electronic bid document image is a qualified image;
[0034] The first document marking module is used to mark the electronic bid document to be tested as a qualified document if each document image in the electronic bid document image is a qualified image.
[0035] Optionally, the device further includes:
[0036] The standard keyword determination module is used to extract the at least one standard keyword from the standard electronic tender document according to a preset regular expression before determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword.
[0037] Optionally, the target keyword extraction model includes a target text recognition sub-model for text extraction and text localization, and a target structured sub-model for structuring text according to text category;
[0038] The target keyword determination module, when inputting the image of the file into the trained target keyword extraction model to obtain at least one target keyword contained in the image of the file, is specifically used for:
[0039] The file image is input into the target text recognition sub-model to obtain at least one target text contained in the file image and the coordinate information of each target text in the file image;
[0040] The coordinate information of the at least one target text and each of the at least one target text in the file image is input into the target structured sub-model to obtain at least one target keyword contained in the file image.
[0041] Optionally, the device further includes:
[0042] The second image marking module is used to mark the image of the file as an unqualified image if, after determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword, the at least one target keyword does not contain all the keywords of the at least one standard keyword.
[0043] Optionally, the device further includes:
[0044] The second document marking module is used to mark the electronic bid document to be tested as an unqualified document if there is an unqualified image among the electronic bid document images after determining whether each document image in the electronic bid document image is a qualified image.
[0045] Optionally, the device further includes:
[0046] The first display module is used to display the unqualified image of the electronic bid document on the page after the electronic bid document to be inspected is marked as an unqualified document;
[0047] And / or, display the keywords to be displayed from the unqualified images in the electronic tender document images on the page, wherein the keywords to be displayed are those that exist in the at least one standard keyword but do not exist in the at least one target keyword.
[0048] Optionally, the device further includes:
[0049] The second display module is used to display the qualified image of the electronic bid document on the page after the electronic bid document to be tested is marked as a qualified document;
[0050] And / or, display at least one target keyword from the qualified images in the electronic tender document images on the page.
[0051] Thirdly, embodiments of this application provide a computer device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the electronic tender document detection method described in any of the optional embodiments of the first aspect are performed.
[0052] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the electronic tender document detection method described in any of the optional embodiments of the first aspect.
[0053] The technical solution provided in this application includes, but is not limited to, the following beneficial effects:
[0054] For each file image in the electronic bid document image, the file image is input into a trained file classification model to obtain the target file category to which the file image belongs. The electronic bid document image is an image obtained by converting at least one page contained in the electronic bid document to be detected from its original format to an image format. Through the above steps, the electronic bid document can be converted into an image and then the image can be classified so that text can be extracted from the image based on the classification results.
[0055] The image of the file is input into a trained target keyword extraction model to obtain at least one target keyword contained in the image. The target keyword extraction model is a key extraction model that has the same category identifier as the target file category to which the image belongs. Through the above steps, the text content in the tender document image can be obtained quickly and accurately.
[0056] Determine whether the at least one target keyword contains all the keywords of the at least one preset standard keyword; if the at least one target keyword contains all the keywords of the at least one standard keyword, then mark the file image as a qualified image; through the above steps, it is possible to determine whether each page of the electronic tender document meets the bidding requirements.
[0057] Determine whether each file image in the electronic bid document is a qualified image; if each file image in the electronic bid document is a qualified image, then mark the electronic bid document to be tested as a qualified document; through the above steps, it is possible to determine whether the electronic bid document to be tested meets the bidding requirements.
[0058] Using the above method, after classifying the file images of the electronic bid documents to be detected using a file classification model, the text information in the images is extracted using the target keyword extraction model corresponding to each classification result and compared with the standard text information. Then, the detection result of the electronic bid document to be detected is determined based on the comparison result, so as to improve the accuracy of detecting electronic bid documents.
[0059] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description
[0060] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0061] Figure 1 The flowchart of an electronic bid document detection method provided in Embodiment 1 of the present invention is shown;
[0062] Figure 2 The flowchart of a target keyword determination method provided in Embodiment 1 of the present invention is shown;
[0063] Figure 3 This diagram illustrates the structure of an electronic bid document detection device provided in Embodiment 2 of the present invention.
[0064] Figure 4 This diagram illustrates the structure of another electronic bid document detection device provided in Embodiment 2 of the present invention;
[0065] Figure 5This diagram illustrates the structure of another electronic bid document detection device provided in Embodiment 2 of the present invention;
[0066] Figure 6 This diagram illustrates the structure of another electronic bid document detection device provided in Embodiment 2 of the present invention;
[0067] Figure 7 This diagram illustrates the structure of another electronic bid document detection device provided in Embodiment 2 of the present invention;
[0068] Figure 8 This diagram illustrates the structure of another electronic bid document detection device provided in Embodiment 2 of the present invention;
[0069] Figure 9 A schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention is shown. Detailed Implementation
[0070] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0071] Example 1
[0072] To facilitate understanding of this application, the following is combined with... Figure 1 The flowchart of the electronic tender document detection method provided in Embodiment 1 of the present invention will be described in detail for Embodiment 1 of this application.
[0073] See Figure 1 The above, Figure 1 A flowchart of an electronic bid document detection method according to Embodiment 1 of the present invention is shown, wherein the method includes steps S101 to S106:
[0074] S101: For each file image in the electronic bid document image, input the file image into the trained file classification model to obtain the target file category to which the file image belongs. The electronic bid document image is an image obtained by converting at least one page contained in the electronic bid document to be detected from the original format to the image format.
[0075] Specifically, electronic tender documents are usually stored on computers or in the cloud in a portable document format. Before testing the electronic tender documents, it is necessary to call a format conversion component to convert the format of the electronic tender documents from the original format to an image format so that the text content contained in the electronic tender documents can be extracted.
[0076] All pages in at least one page of the electronic bid document to be tested are converted into electronic bid document images. The number of pages in the electronic bid document to be tested is the same as the number of file images in the electronic bid document images. That is, each page in the electronic bid document to be tested is converted into a file image in the electronic bid document images.
[0077] For each file image in the electronic tender document, the file image is input into the trained file classification model to obtain the target file category to which the file image belongs. Based on the role of each component in the tender document, the target file category includes qualification certificates, financial statements, business licenses, permits, registration certificates, etc. The specific category can be set according to the bidding requirements.
[0078] Training a document classification model involves the following two steps:
[0079] Step 1: Use a small number of bid document image samples to generate a training set containing a large number of bid document image samples through a GAN (Generative Adversarial Network).
[0080] Step 2: Train the Swin transformer algorithm framework (an algorithm framework) using a training set containing a large number of samples to obtain a trained file classification model.
[0081] S102: Input the file image into the trained target keyword extraction model to obtain at least one target keyword contained in the file image, wherein the target keyword extraction model is a key extraction model among all keyword extraction models that has the same category identifier as the target file category to which the file image belongs.
[0082] Specifically, since different types of images require different models for keyword extraction, after determining the target file category of the image, the image is input into a trained target keyword extraction model (which is a key extraction model among all keyword extraction models that has the same category identifier as the target file category to which the image belongs) to extract the target keywords from the image.
[0083] S103: Determine whether the at least one target keyword contains all the keywords of the at least one preset standard keyword.
[0084] Specifically, the qualification of a bid document is reflected in whether the conditions in the bid document meet the requirements of the tender document. Since a bid document consists of multiple components (which can be regarded as multiple document pages), the bid document is first split into multiple pages, and the content of each page is judged to determine whether the content of each page meets the standard content to determine whether each page is a qualified page. Then, it is judged whether all pages are qualified pages to determine whether the bid document is a qualified document.
[0085] The tender documents are pre-designed, so the content (standard content or standard keywords) in the tender documents is also known. When judging whether the image in the document is qualified, the known standard keywords are compared with the keywords extracted from the image in the document, and the comparison results determine whether the image in the document is a qualified keyword.
[0086] S104: If the at least one target keyword contains all the keywords of the at least one standard keyword, then the image in the file is marked as a qualified image.
[0087] Specifically, target keywords are keywords in the tender document, while standard keywords are keywords in the bidding document. The tender document must meet all the requirements in the bidding document to be considered as meeting the requirements. Therefore, if the at least one target keyword contains all the keywords in the at least one standard keyword, the image of the document will be marked as a qualified image.
[0088] S105: Determine whether each file image in the electronic bid document images is a qualified image.
[0089] Specifically, all content in the tender documents must meet the requirements of the tender documents, so it is necessary to determine whether each file image in the electronic tender document images is a qualified image.
[0090] S106: If each file image in the electronic bid document image is a qualified image, then the electronic bid document to be tested is marked as a qualified document.
[0091] Specifically, if all contents in the tender document meet the requirements of the tender document, it means that the electronic tender document image belongs to the electronic tender document to be tested and is a qualified tender document.
[0092] In one feasible implementation, before determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword, the method further includes:
[0093] The at least one standard keyword is extracted from the standard electronic tender document according to a preset regular expression.
[0094] Specifically, regular expressions describe a pattern for matching strings. They can be used to check whether a string contains a certain substring, replace the matched substring, or extract a substring from a string that meets a certain condition. In other words, you can use a predefined regular expression to extract the target text from the text information that corresponds to the regular expression and save it in JSON (JavaScript Object Notation) format.
[0095] In one feasible implementation, the target keyword extraction model includes a target text recognition sub-model for text extraction and text localization, and a target structured sub-model for structuring text according to text category.
[0096] Specifically, the target structuring sub-model is used to perform structuring processing on the text and text positions extracted by the target text recognition sub-model to obtain keywords; the target text recognition sub-model is used to extract the text and text positions (coordinates) in the image to be detected. It is a recognition network consisting of four parts: shared convolution, text detection branch, ROIRotate algorithm (a rotation algorithm), and text recognition branch. The training of the text recognition sub-model includes the following steps:
[0097] Step 1: Use a small number of bid document image samples to generate a training set containing a large number of bid document image samples through a GAN (Generative Adversarial Network).
[0098] Step 2: Use shared convolution to extract feature maps from the training set. By sharing convolutional features, the network can detect and recognize text simultaneously with very little computational overhead, thus achieving real-time speed.
[0099] Step 3: Based on the extracted feature maps, construct a text detection branch based on a fully convolutional neural network to predict text detection boxes.
[0100] Step 4: Next, the RoIRotate algorithm is used to extract text candidate features corresponding to the detection results from the feature map. This operation unifies text detection and recognition into an end-to-end training process.
[0101] Step 5: Finally, the text candidate features are input into the recurrent neural network encoder and the connected temporal classification decoder for text recognition.
[0102] See Figure 2 The above, Figure 2 The flowchart of a target keyword determination method provided in Embodiment 1 of the present invention is shown, wherein the step of inputting the file image into a trained target keyword extraction model to obtain at least one target keyword contained in the file image includes steps S201 to S202:
[0103] S201: Input the file image into the target character recognition sub-model to obtain at least one target character contained in the file image and the coordinate information of each target character in the file image.
[0104] Specifically, the coordinate information can be in the form of two-dimensional coordinates, with the origin of the two-dimensional coordinates being the top left vertex of the image to be detected. At least one target text and the coordinate information of each target text are saved in JSON format.
[0105] S202: Input the coordinate information of the at least one target text and each target text in the file image into the target structured sub-model to obtain at least one target keyword contained in the file image.
[0106] Specifically, save at least one target keyword in JSON format.
[0107] In one feasible implementation, after determining whether the at least one target keyword contains all the keywords of at least one preset standard keyword, the method further includes:
[0108] If the at least one target keyword does not contain all the keywords of the at least one standard keyword, then the image file will be marked as an unqualified image.
[0109] Specifically, if the at least one target keyword does not contain all the keywords in the at least one standard keyword, it means that the page in the tender document to be tested corresponding to the file image is missing the conditions in the tender document, so the file image is marked as an unqualified image.
[0110] In one feasible implementation, after determining whether each file image in the electronic bid document images is a qualified image, the method further includes:
[0111] If any of the images in the electronic bid document are unqualified, the electronic bid document to be tested will be marked as unqualified.
[0112] Specifically, if there are unqualified images in the electronic bid document, it indicates that there are documents in the bid document that do not meet the bidding requirements, and the electronic bid document to be tested will be marked as an unqualified document.
[0113] In one feasible implementation, after marking the electronic tender document to be inspected as a non-compliant document, the method includes:
[0114] Displaying unqualified images from the electronic bid document images on the page; and / or displaying keywords from the unqualified images from the electronic bid document images on the page, wherein the keywords to be displayed are keywords that exist in the at least one standard keyword but do not exist in the at least one target keyword.
[0115] Specifically, the non-compliant images in the electronic bid document will be displayed on the page so that users can see the non-compliant content in the electronic bid document to be inspected.
[0116] For example, if at least one standard keyword is "length" and "width", and at least one target keyword is "quality" and "length", then the keyword to be displayed is "width".
[0117] In one feasible implementation, after marking the electronic tender document to be inspected as a qualified document, the method further includes:
[0118] Display qualified images from the electronic bid document images on the page; and / or display at least one target keyword from the qualified images from the electronic bid document images on the page.
[0119] Specifically, qualified images from the electronic bid documents will be displayed on the page so that users can see the content in the electronic bid documents that meets the requirements.
[0120] Example 2
[0121] See Figure 3 As shown, Figure 3 This diagram illustrates the structure of an electronic bid document detection device according to Embodiment 2 of the present invention, wherein, as shown... Figure 3 As shown, the electronic bid document detection device provided in Embodiment 2 of the present invention includes:
[0122] The document category determination module 301 is used to input each document image in the electronic bid document image into a trained document classification model to obtain the target document category to which the document image belongs. The electronic bid document image is an image obtained by converting at least one page contained in the electronic bid document to be detected from the original format to an image format.
[0123] The target keyword determination module 302 is used to input the file image into the trained target keyword extraction model to obtain at least one target keyword contained in the file image, wherein the target keyword extraction model is a key extraction model among all keyword extraction models that has the same category identifier as the target file category to which the file image belongs;
[0124] The first judgment module 303 is used to determine whether the at least one target keyword contains all the keywords of the at least one preset standard keyword;
[0125] The first image marking module 304 is used to mark the image of the file as a qualified image if the at least one target keyword contains all the keywords of the at least one standard keyword;
[0126] The second judgment module 305 is used to determine whether each file image in the electronic bid document image is a qualified image;
[0127] The first document marking module 306 is used to mark the electronic bid document to be tested as a qualified document if each document image in the electronic bid document image is a qualified image.
[0128] In one feasible implementation plan, see Figure 4 The above, Figure 4 A schematic diagram of another electronic bid document detection device provided in Embodiment 2 of the present invention is shown, wherein the device further includes:
[0129] The standard keyword determination module 401 is used to extract the at least one standard keyword from the standard electronic tender document according to a preset regular expression before determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword.
[0130] In one feasible implementation, the target keyword extraction model includes a target text recognition sub-model for text extraction and text localization, and a target structured sub-model for structuring text according to text category.
[0131] The target keyword determination module, when inputting the image of the file into the trained target keyword extraction model to obtain at least one target keyword contained in the image of the file, is specifically used for:
[0132] The file image is input into the target text recognition sub-model to obtain at least one target text contained in the file image and the coordinate information of each target text in the file image;
[0133] The coordinate information of the at least one target text and each of the at least one target text in the file image is input into the target structured sub-model to obtain at least one target keyword contained in the file image.
[0134] In one feasible implementation plan, see Figure 5 The above, Figure 5 A schematic diagram of another electronic bid document detection device provided in Embodiment 2 of the present invention is shown, wherein the device further includes:
[0135] The second image marking module 501 is used to mark the image file as an unqualified image if, after determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword, the at least one target keyword does not contain all the keywords of the at least one standard keyword.
[0136] In one feasible implementation plan, see Figure 6 The above, Figure 6 A schematic diagram of another electronic bid document detection device provided in Embodiment 2 of the present invention is shown, wherein the device further includes:
[0137] The second document marking module 601 is used to mark the electronic bid document to be tested as an unqualified document if there is an unqualified image among the electronic bid document images after determining whether each document image in the electronic bid document image is a qualified image.
[0138] In one feasible implementation plan, see Figure 7 The above, Figure 7 A schematic diagram of another electronic bid document detection device provided in Embodiment 2 of the present invention is shown, wherein the device further includes:
[0139] The first display module 701 is used to display the unqualified image of the electronic bid document on the page after the electronic bid document to be inspected is marked as an unqualified document;
[0140] And / or, display the keywords to be displayed from the unqualified images in the electronic tender document images on the page, wherein the keywords to be displayed are those that exist in the at least one standard keyword but do not exist in the at least one target keyword.
[0141] In one feasible implementation plan, see Figure 8 The above, Figure 8 A schematic diagram of another electronic bid document detection device provided in Embodiment 2 of the present invention is shown, wherein the device further includes:
[0142] The second display module 801 is used to display the qualified image of the electronic bid document on the page after the electronic bid document to be tested is marked as a qualified document;
[0143] And / or, display at least one target keyword from the qualified images in the electronic tender document images on the page.
[0144] Example 3
[0145] Based on the same application concept, see [link / reference] Figure 9 As shown, Figure 9 A schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention is shown, wherein, as Figure 9 As shown, the computer device 900 provided in Embodiment 3 of this application includes:
[0146] The computer device 900 includes a processor 901, a memory 902, and a bus 903. The memory 902 stores machine-readable instructions that can be executed by the processor 901. When the computer device 900 is running, the processor 901 communicates with the memory 902 through the bus 903. When the machine-readable instructions are executed by the processor 901, they perform the steps of the electronic tender document detection method shown in Embodiment 1 above.
[0147] Example 4
[0148] Based on the same concept, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of an electronic tender document detection method as described in any of the above embodiments.
[0149] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0150] The computer program product for detecting electronic tender documents provided in this embodiment of the invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.
[0151] The electronic bid document detection device provided in this embodiment of the invention can be specific hardware on a device or software or firmware installed on the device. The implementation principle and technical effects of the device provided in this embodiment of the invention are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the device embodiments can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.
[0152] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.
[0153] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0154] In addition, the functional units in the embodiments provided by the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0155] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0156] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0157] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. All should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for detecting electronic bid documents, characterized in that, The method includes: For each file image in the electronic bid document image, the file image is input into the trained file classification model to obtain the target file category to which the file image belongs. The electronic bid document image is an image obtained by converting at least one page contained in the electronic bid document to be detected from the original format to the image format. The image of the file is input into a trained target keyword extraction model to obtain at least one target keyword contained in the image of the file. The target keyword extraction model is a key extraction model among all keyword extraction models that has the same category identifier as the target file category to which the image of the file belongs. Determine whether the at least one target keyword contains all the keywords in the at least one preset standard keyword; If the at least one target keyword contains all the keywords of the at least one standard keyword, then the image in the file is marked as a qualified image; Determine whether each file image in the electronic bid document images is a qualified image; If each file image in the electronic bid document is a qualified image, then the electronic bid document to be tested is marked as a qualified document; The target keyword extraction model includes a target text recognition sub-model for text extraction and text localization, and a target structured sub-model for structuring text according to text category. The step of inputting the image of the file into a trained target keyword extraction model to obtain at least one target keyword contained in the image of the file includes: The file image is input into the target text recognition sub-model to obtain at least one target text contained in the file image and the coordinate information of each target text in the file image; The coordinate information of the at least one target text and each of the at least one target text in the file image is input into the target structured sub-model to obtain at least one target keyword contained in the file image.
2. The method according to claim 1, characterized in that, Before determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword, the method further includes: The at least one standard keyword is extracted from the standard electronic tender document according to a preset regular expression.
3. The method according to claim 1, characterized in that, After determining whether the at least one target keyword contains all the keywords of the at least one preset standard keyword, the method further includes: If the at least one target keyword does not contain all the keywords of the at least one standard keyword, then the image file will be marked as an unqualified image.
4. The method according to claim 3, characterized in that, After determining whether each file image in the electronic bid document is a qualified image, the method further includes: If any of the images in the electronic bid document are unqualified, the electronic bid document to be tested will be marked as unqualified.
5. The method according to claim 4, characterized in that, After marking the electronic tender document to be inspected as an unqualified document, the method includes: The page will display any non-compliant images from the electronic bid document. And / or, display the keywords to be displayed from the unqualified images in the electronic tender document images on the page, wherein the keywords to be displayed are those that exist in the at least one standard keyword but do not exist in the at least one target keyword.
6. The method according to claim 1, characterized in that, After marking the electronic tender document to be inspected as a qualified document, the method further includes: Display the qualified images from the electronic bid document images on the page; And / or, display at least one target keyword from the qualified images in the electronic tender document images on the page.
7. An electronic bid document inspection device, characterized in that, The device includes: The document category determination module is used to input each document image in the electronic bid document image into a trained document classification model to obtain the target document category to which the document image belongs. The electronic bid document image is an image obtained by converting at least one page contained in the electronic bid document to be detected from the original format to an image format. The target keyword determination module is used to input the file image into a trained target keyword extraction model to obtain at least one target keyword contained in the file image. The target keyword extraction model is a key extraction model among all keyword extraction models that has the same category identifier as the target file category to which the file image belongs. The first judgment module is used to determine whether the at least one target keyword contains all the keywords in the at least one preset standard keyword; The first image tagging module is used to tag the image of the file as a qualified image if the at least one target keyword contains all the keywords of the at least one standard keyword; The second judgment module is used to determine whether each file image in the electronic bid document image is a qualified image; The first document marking module is used to mark the electronic bid document to be tested as a qualified document if each document image in the electronic bid document image is a qualified image; The target keyword extraction model includes a target text recognition sub-model for text extraction and text localization, and a target structured sub-model for structuring text according to text category. The target keyword determination module, when inputting the image of the file into the trained target keyword extraction model to obtain at least one target keyword contained in the image of the file, is specifically used for: The file image is input into the target text recognition sub-model to obtain at least one target text contained in the file image and the coordinate information of each target text in the file image; The coordinate information of the at least one target text and each of the at least one target text in the file image is input into the target structured sub-model to obtain at least one target keyword contained in the file image.
8. A computer device, characterized in that, include: The computer device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the electronic tender document detection method as described in any one of claims 1 to 6.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the electronic tender document detection method as described in any one of claims 1 to 6.