CN-115424272-B - Electronic bidding document detection method, electronic bidding document detection device, computer equipment and medium

Abstract

The application provides an electronic bidding document detection method, device, computer equipment and medium, wherein for each document image in electronic bidding document images, the document image is input into a document classification model to obtain a target document category of the document image, the document image is input into a target keyword extraction model to obtain at least one target keyword in the document image, whether the at least one target keyword contains all keywords in at least one preset standard keyword is judged, if the at least one target keyword contains all keywords in the at least one standard keyword, the document image is marked as a qualified image, whether each document image in the electronic bidding document image is a qualified image is judged, and if each document image in the electronic bidding document image is a qualified image, the electronic bidding document to be detected is marked as the qualified document. By adopting the method, the accuracy of detecting the electronic bidding document is improved.

Inventors

• Request for anonymity

Assignees

• 北京筑龙信息技术有限责任公司

Dates

Publication Date

20260512

Application Date

20220826

Claims (9)

1. 1. An electronic bidding document detection method, the method comprising: For each file picture in the electronic bidding file pictures, inputting the file picture into a trained file classification model to obtain a target file type to which the file picture belongs, wherein the electronic bidding file picture is obtained by converting at least one page contained in the electronic bidding file to be detected from an original format into a picture format; Inputting the file picture into a trained target keyword extraction model to obtain at least one target keyword contained in the file picture, wherein the target keyword extraction model is a keyword extraction model with the same category identification as the target file category to which the file picture belongs in all the keyword extraction models; Judging whether the at least one target keyword comprises all keywords in at least one preset standard keyword or not; if the at least one target keyword comprises all keywords in the at least one standard keyword, marking the file picture as a qualified picture; judging whether each file picture in the electronic bidding file pictures is a qualified picture or not; If each file picture in the electronic bidding file pictures is a qualified picture, marking the electronic bidding file to be detected as a qualified file; the target keyword extraction model comprises a target character recognition sub-model for character extraction and character positioning and a target structuring sub-model for structuring characters according to character categories; inputting the file picture into a trained target keyword extraction model to obtain at least one target keyword contained in the file picture, wherein the method comprises the following steps: Inputting the file picture into the target character recognition sub-model to obtain at least one target character contained in the file picture and coordinate information of each target character in the at least one target character in the file picture; And inputting the at least one target character and the coordinate information of each target character in the at least one target character in the file picture into the target structuring sub-model to obtain at least one target keyword contained in the file picture.
2. 2. The method of claim 1, wherein before determining whether all keywords in the preset at least one standard keyword are included in the at least one target keyword, the method further comprises: and extracting the at least one standard keyword from the standard electronic bidding document according to a preset regular expression.
3. 3. The method of claim 1, wherein after determining whether all keywords in the preset at least one standard keyword are included in the at least one target keyword, the method further comprises: And if all the keywords in the at least one standard keyword are not contained in the at least one target keyword, marking the file picture as a disqualified picture.
4. 4. The method of claim 3, wherein after determining whether each of the electronic bid document pictures is a qualified picture, the method further comprises: And if the electronic bidding document picture has an unqualified picture, marking the electronic bidding document to be detected as an unqualified document.
5. 5. The method of claim 4, wherein after marking the electronic bid document to be detected as a disqualifying document, the method comprises: Displaying unqualified pictures in the electronic bidding document pictures in a page; and/or displaying the keywords to be displayed in the disqualified pictures in the electronic bidding document pictures in the page, wherein the keywords to be displayed are keywords which exist in the at least one standard keyword but do not exist in the at least one target keyword.
6. 6. The method of claim 1, wherein after marking the electronic bid document to be detected as a qualified document, the method further comprises: displaying the qualified pictures in the electronic bidding document pictures in a page; and/or displaying at least one target keyword in the qualified picture in the electronic bidding document picture in the page.
7. 7. An electronic bidding document detection apparatus, the apparatus comprising: The file type determining module is used for inputting each file picture in the electronic bidding file pictures into the trained file classification model to obtain the target file type to which the file picture belongs, wherein the electronic bidding file pictures are obtained by converting at least one page contained in the electronic bidding file to be detected from an original format to a picture format; The target keyword determining module is used for inputting the file picture into the trained target keyword extracting model to obtain at least one target keyword contained in the file picture, wherein the target keyword extracting model is a keyword extracting model which has the same category identification as the target file category to which the file picture belongs in all the keyword extracting models; the first judging module is used for judging whether the at least one target keyword comprises all keywords in at least one preset standard keyword or not; the first picture marking module is used for marking the file picture as a qualified picture if all keywords in the at least one standard keyword are contained in the at least one target keyword; the second judging module is used for judging whether each file picture in the electronic bidding file pictures is a qualified picture or not; The first file marking module is used for marking the electronic bidding file to be detected as a qualified file if each file picture in the electronic bidding file pictures is a qualified picture; the target keyword extraction model comprises a target character recognition sub-model for character extraction and character positioning and a target structuring sub-model for structuring characters according to character categories; The target keyword determining module is specifically configured to, when being configured to input the file picture to a trained target keyword extraction model to obtain at least one target keyword included in the file picture: Inputting the file picture into the target character recognition sub-model to obtain at least one target character contained in the file picture and coordinate information of each target character in the at least one target character in the file picture; And inputting the at least one target character and the coordinate information of each target character in the at least one target character in the file picture into the target structuring sub-model to obtain at least one target keyword contained in the file picture.
8. 8. A computer device comprising a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory in communication via the bus when the computer device is in operation, the machine-readable instructions when executed by the processor performing the steps of the electronic bid file detection method of any of claims 1 to 6.
9. 9. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, performs the steps of the electronic bid file detection method according to any one of claims 1 to 6.

Description

Electronic bidding document detection method, electronic bidding document detection device, computer equipment and medium Technical Field The invention relates to the field of electronic bidding, in particular to a method, a device, computer equipment and a medium for detecting electronic bidding documents. Background In the prior art, when judging whether an electronic bidding document meets bidding requirements, the electronic bidding is usually checked by manpower, that is, the contents in the electronic bidding are read page by page line by a bidding management related person, and the contents of the bidding are checked after counting the conditions in the electronic bidding. The inventor found in the study that when examining the tender books, a great deal of information such as characters and pictures are required to be accurately counted and analyzed, and when examining through manpower, the examination result of errors is likely to be obtained when examining the tender documents due to the great variety of the contents of the tender books and the insufficient experience of the processing personnel, so that the accuracy of detecting the electronic tender documents is reduced. Disclosure of Invention Accordingly, the present invention is directed to a method, apparatus, computer device and medium for detecting electronic bidding documents, so as to improve the accuracy of detecting electronic bidding documents. In a first aspect, an embodiment of the present application provides a method for detecting an electronic bidding document, where the method includes: For each file picture in the electronic bidding file pictures, inputting the file picture into a trained file classification model to obtain a target file type to which the file picture belongs, wherein the electronic bidding file picture is obtained by converting at least one page contained in the electronic bidding file to be detected from an original format into a picture format; Inputting the file picture into a trained target keyword extraction model to obtain at least one target keyword contained in the file picture, wherein the target keyword extraction model is a keyword extraction model with the same category identification as the target file category to which the file picture belongs in all the keyword extraction models; Judging whether the at least one target keyword comprises all keywords in at least one preset standard keyword or not; if the at least one target keyword comprises all keywords in the at least one standard keyword, marking the file picture as a qualified picture; judging whether each file picture in the electronic bidding file pictures is a qualified picture or not; And if each file picture in the electronic bidding file pictures is a qualified picture, marking the electronic bidding file to be detected as a qualified file. Optionally, before determining whether the at least one target keyword includes all keywords in the preset at least one standard keyword, the method further includes: and extracting the at least one standard keyword from the standard electronic bidding document according to a preset regular expression. Optionally, the target keyword extraction model comprises a target character recognition sub-model for character extraction and character positioning, and a target structuring sub-model for structuring characters according to character categories; inputting the file picture into a trained target keyword extraction model to obtain at least one target keyword contained in the file picture, wherein the method comprises the following steps: Inputting the file picture into the target character recognition sub-model to obtain at least one target character contained in the file picture and coordinate information of each target character in the at least one target character in the file picture; And inputting the at least one target character and the coordinate information of each target character in the at least one target character in the file picture into the target structuring sub-model to obtain at least one target keyword contained in the file picture. Optionally, after determining whether the at least one target keyword includes all keywords in the preset at least one standard keyword, the method further includes: And if all the keywords in the at least one standard keyword are not contained in the at least one target keyword, marking the file picture as a disqualified picture. Optionally, after determining whether each of the electronic bidding document pictures is a qualified picture, the method further includes: And if the electronic bidding document picture has an unqualified picture, marking the electronic bidding document to be detected as an unqualified document. Optionally, after marking the electronic bid document to be detected as a disqualified document, the method includes: Displaying unqualified pictures in the electronic bidding document pictures in a page; and/or displaying the keywords to