An electronic dossier intelligent cataloging method and computer equipment

Abstract

The application discloses an electronic file intelligent cataloging method and computer equipment, the method comprises the following steps: obtaining an electronic file to be cataloged, and judging the file type of the electronic file to be cataloged; performing OCR identification on single-page images in stored images, and performing format analysis on text coordinate points and text styles to obtain a title analysis result; performing format analysis cataloging on the OCR identification result and the coordinate style title; performing customized cataloging for different business scenarios through pre-defined rule cataloging; performing named entity extraction through a natural language processing algorithm, defining the splicing of a cataloging name and an entity through a pre-set splicing mode according to the extraction result of the named entity extraction; continuously circulating the above process until cataloging of all single-page images in the stored images is completed; through the intelligent cataloging method, the accuracy of cataloging of diversified materials can be improved to more than 90%, meanwhile, a large amount of manual checking is not needed, maintenance and optimization costs can be reduced, and classification efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of electronic case file technology, and in particular to an intelligent cataloging method and computer equipment for electronic case files. Background Technology

[0002] Electronic case files record all kinds of documents generated throughout the entire event, which play a very important role in the review, retrieval and public disclosure of the event. For example, court case files, hospital medical records, social security payment and reimbursement records, law firm case materials, etc. With the advent of paperless technology, electronic case files have the outstanding advantages of fast retrieval and convenient storage.

[0003] Electronic case file cataloging refers to the process of analyzing, selecting, and describing the external and content characteristics of each entity (item) within an electronic case file according to certain standards and rules, recording these characteristics as entries, and then organizing these entries in a specific order to form a catalogue or bibliography. Electronic case file cataloging can quickly, accurately, and comprehensively annotate the electronic case file catalogue and material names, facilitating rapid location, retrieval, and visual utilization of electronic case file content, thus laying a solid foundation for the in-depth application of electronic case files.

[0004] Currently, the classification and cataloging of electronic case files is mainly used for archival purposes. The diverse types of materials and varying cataloging standards result in low cataloging accuracy, low classification efficiency, and are time-consuming and labor-intensive. Furthermore, most current automatic cataloging tools for electronic case files rely on regular expressions to match OCR recognition results for cataloging names. This method cannot accurately catalog diverse material types using existing matching methods, leading to low cataloging accuracy, requiring extensive manual verification, system maintenance, and optimization, making it unsustainable. Summary of the Invention

[0005] Based on this, and in response to the aforementioned technical problems, an intelligent cataloging method and computer device for electronic case files are provided, which solves the problem of low accuracy in electronic case file cataloging in existing technologies.

[0006] Firstly, an intelligent cataloging method for electronic case files, characterized by comprising:

[0007] Step S1: Obtain the electronic file to be cataloged and determine the file type of the electronic file to be cataloged; if the electronic file to be cataloged is an image file, store the electronic file to be cataloged directly; if the electronic file to be cataloged is a layout file, split the electronic file to be cataloged page by page into independent image files and store them.

[0008] Step S2: Perform OCR recognition on the single-page image in the stored image and obtain the OCR recognition result; perform layout parsing on the text coordinate points and text style to obtain the title analysis result, which includes the title information of the current single-page image;

[0009] Step S3: When performing layout analysis and cataloging on the current single-page image, obtain the title analysis results and the illustration analysis results based on whether illustration elements can be obtained. Then, catalog according to steps S31-S33:

[0010] Step S31: Determine whether the current single-page image contains an illustration;

[0011] Step S32: If it is determined that the current single-page image contains an illustration, determine whether the proportion of the illustration and the text reaches the preset proportion threshold; if it is determined that the proportion of the illustration and the text reaches the preset proportion threshold, then perform target detection on the illustration, and after the target detection is successful, use the catalog name corresponding to the detection result with a confidence level exceeding the preset confidence threshold as the definition catalog name of the current single-page image.

[0012] Step S33: If it is determined that the current single-page image does not contain illustrations, determine whether the title of the current single-page image meets the title rules; if the title meets the title rules, filter the title for special symbols and use the filtering result as the definition catalog name of the current single-page image; if the title does not meet the title rules, check whether there is a result for the catalog name of the previous single-page image; if there is a result, use the catalog name of the previous single-page image as the definition catalog name of the current single-page image; if there is no result, name the current single-page image "Other" as the definition catalog name of the current single-page image.

[0013] Step S4: When cataloging the current single-page image using predefined rule matching, cataloging is performed according to steps S41-S43:

[0014] Step S41: Obtain predefined rules and match the OCR recognition results and the title analysis results with the predefined rules;

[0015] Step S42: If the OCR recognition result and the title analysis result satisfy the matching condition of one of the predefined rules, then the corresponding catalog name result of the predefined rule that satisfies the matching condition shall be used as the definition catalog name of the current single-page image.

[0016] Step S43: If the OCR recognition result and the title analysis result do not meet the matching conditions of any predefined rule, check whether there is a result for the catalog name of the previous single-page image. If there is a result, use the catalog name of the previous single-page image as the defined catalog name of the current single-page image. If there is no result, name the current single-page image as something else as the defined catalog name of the current single-page image.

[0017] Step S5: Obtain the defined catalog name and entity extraction rules, and determine whether named entity extraction is required. If named entity extraction is required, use a natural language processing algorithm to extract named entities. After the named entity extraction is completed, obtain a pre-set splicing method, and use the pre-set splicing method to splice the extracted results of the named entity extraction with the defined catalog name and the entity. Use the spliced ​​result as the final catalog name of the current single-page image. If named entity extraction is not required, directly use the defined catalog name as the final catalog name of the current single-page image.

[0018] Step S6: Repeat steps S2-S5 until all single-page images in the stored images are cataloged.

[0019] Optionally, the image file is specifically a JPEG file or a PNG file, and the layout file is specifically a PDF file or an OFD file.

[0020] Optionally, the step of performing layout parsing on text coordinates and text styles to obtain the title analysis results includes:

[0021] Based on the OCR recognition results, determine whether there are coordinate points in the text style that have a larger font size or bolder font than the surrounding text;

[0022] When a coordinate point is determined to have a larger font size or bolder font than the surrounding text, the text at that coordinate point and the text extending to the left and right of that coordinate point are obtained, and it is determined whether all the obtained text is centered or centered at the top; if it is determined that all the obtained text is centered or centered at the top, then all the obtained text is parsed as a title.

[0023] If it is determined that there are no coordinate points with a larger font size or bolder font than the surrounding text, the text of the top center coordinate point and the text extending to the left and right of that coordinate point are obtained, and it is determined whether the left and right margins of the text below the top center coordinate point are filled; if it is determined that the left and right margins of the text below the top center coordinate point are filled, then all the obtained text is parsed as a title.

[0024] Optionally, an object detection neural network can be used for object detection.

[0025] Further optionally, the target detection neural network is specifically YOLO v3, RefineDet, M2Det, or Spiking-YOLO.

[0026] Optionally, the preset confidence threshold is 0.9.

[0027] Optionally, step S32 further includes:

[0028] If the ratio of illustration to text does not reach the preset ratio threshold, then it is determined whether the title of the current single-page image meets the title rules; if the title meets the title rules, then the title is filtered for special symbols, and the filtering result is used as the definition catalog name of the current single-page image; if the title does not meet the title rules, then it is checked whether there is a result for the catalog name of the previous single-page image; if there is a result, then the catalog name of the previous single-page image is used as the definition catalog name of the current single-page image; if there is no result, then the current single-page image is named "Other" as the definition catalog name of the current single-page image.

[0029] If object detection fails, it checks whether the title of the current page image meets the title rules. If the title meets the title rules, it filters the title for special characters and uses the filtered result as the definition catalog name of the current page image. If the title does not meet the title rules, it checks whether the catalog name of the previous page image exists. If it does, it uses the catalog name of the previous page image as the definition catalog name of the current page image. If it does not exist, it names the current page image as something else as the definition catalog name of the current page image.

[0030] Optionally, the natural language processing algorithm is a hidden Markov model algorithm, a Viterbi algorithm, or an Aho-Corasick algorithm.

[0031] Optionally, the entity includes the case number of the judgment, the name on the identification document, and the name of the institution.

[0032] In a second aspect, a computer device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the method described in any of the first aspects.

[0033] This application has at least the following beneficial effects:

[0034] This application provides a detailed method for intelligent cataloging of electronic case files, including extracting material titles through layout parsing, cataloging the layout using OCR recognition results and coordinate style titles, performing customized cataloging for different business scenarios using predefined rules, extracting important information from images using NLP named entity extraction, and continuously repeating the above process until all single-page images in the stored images are cataloged. The intelligent cataloging method provided by this application can improve the accuracy of cataloging diverse materials to over 90%, while eliminating the need for extensive manual verification, reducing maintenance and optimization costs, and improving classification efficiency.

[0035] In addition, using an object detection neural network for automatic object detection during layout parsing and cataloging can further reduce manual costs and improve classification efficiency. Attached Figure Description

[0036] Figure 1 A flowchart illustrating an intelligent cataloging method for electronic case files, provided as an embodiment of this application;

[0037] Figure 2 This is a schematic diagram of the layout analysis and cataloging process in one embodiment of this application;

[0038] Figure 3 This is a schematic diagram of the predefined rule matching and cataloging process in one embodiment of this application;

[0039] Figure 4 This is a schematic diagram of the named entity extraction process in one embodiment of this application;

[0040] Figure 5 A complete flowchart illustrating an intelligent cataloging method for electronic case files is provided as an embodiment of this application;

[0041] Figure 6 This is an internal structural diagram of a computer device provided in one embodiment of this application. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0043] In one embodiment, such as Figure 1 As shown, an intelligent cataloging method for electronic case files is provided, including the following steps:

[0044] Step S1: After uploading the electronic dossier to be cataloged, obtain the electronic dossier to be cataloged and determine the file type of the electronic dossier to be cataloged; if the electronic dossier to be cataloged is an image file, store the electronic dossier to be cataloged directly; if the electronic dossier to be cataloged is a layout file, split the electronic dossier to be cataloged page by page into independent image files and store them.

[0045] Specifically, image files are either JPEG (JPG) or PNG files, while layout files are either PDF or OFD files. In other words, PDF or OFD files are split into single pages and stored as multiple single-page images; image files are stored directly. The stored image list is then retrieved.

[0046] Step S2 involves performing OCR recognition on the single-page image in the stored images and obtaining the OCR recognition result; then performing layout analysis on the text coordinates and text styles to obtain the title analysis result, which includes the title information of the current single-page image. In other words, it involves extracting a single image, performing OCR recognition, obtaining the recognition result, and then performing layout analysis on the result.

[0047] The title analysis results obtained by parsing the text coordinates and text styles include:

[0048] (1) Based on the OCR recognition results, determine whether there are coordinate points in the text style that have a larger font size or bolder font than the surrounding text;

[0049] (2) When it is determined that there is a coordinate point with a larger font size than the surrounding text (i.e., larger font size) or bold font, the text at that coordinate point and the text extending to the left and right of that coordinate point are obtained, and it is determined whether the position of all obtained text is centered or centered at the top; where, extending to the left and right means extending the entire line where the obtained coordinate point is located; if it is determined that the position of all obtained text is centered or centered at the top, then all obtained text is parsed as a title;

[0050] (3) When it is determined that there is no coordinate point with a larger font size than the surrounding text or a bold font, obtain the text of the top center coordinate point and the text extending to the left and right of that coordinate point, and determine whether the left and right margins of the text below the top center coordinate point are filled; if it is determined that the left and right margins of the text below the top center coordinate point are filled, then parse all the obtained text into a title.

[0051] Subsequent cataloging is performed either through layout analysis or by matching predefined rules. In practice, the choice between layout analysis and predefined rules can be pre-defined as needed. For most materials, layout analysis can be used for coding. However, in some cases or application scenarios involving material aliases (e.g., a material titled "Ruling" but needing to be renamed "*** Court Ruling"), predefined rules can be used for matching.

[0052] Step S3, when performing layout analysis and cataloging on the current single-page image, such as Figure 2 As shown, obtain the title analysis results, and obtain the illustration analysis results based on whether illustration elements can be obtained. Then, catalog according to the following steps S31-S33:

[0053] Step S31: Determine whether the current single-page image contains an illustration;

[0054] Step S32: If it is determined that the current single-page image contains an illustration, determine whether the proportion of the illustration and the text reaches a preset proportion threshold; if it is determined that the proportion of the illustration and the text reaches the preset proportion threshold, then perform target detection on the illustration. After the target detection is successful, use the catalog name corresponding to the detection result with a confidence level exceeding the preset confidence threshold as the definition catalog name of the current single-page image; wherein, the preset proportion threshold can be determined according to the actual situation, and the preset confidence threshold can be set to 0.9;

[0055] Step S33: If it is determined that the current single-page image does not contain illustrations, determine whether the title of the current single-page image meets the title rules; if the title meets the title rules, filter the title for special symbols and use the filtering result as the definition catalog name of the current single-page image; if the title does not meet the title rules, check whether the catalog name of the previous single-page image has a result; if a result exists, use the catalog name of the previous single-page image as the definition catalog name of the current single-page image; if no result exists, name the current single-page image "Other" as the definition catalog name of the current single-page image.

[0056] Furthermore, step S32 also includes:

[0057] If the ratio of illustration to text does not reach the preset ratio threshold, then it is determined whether the title of the current single-page image meets the title rules; if the title meets the title rules, then the title is filtered for special symbols, and the filtering result is used as the definition catalog name of the current single-page image; if the title does not meet the title rules, then it is checked whether there is a result for the catalog name of the previous single-page image; if there is a result, then the catalog name of the previous single-page image is used as the definition catalog name of the current single-page image; if there is no result, then the current single-page image is named "Other" as the definition catalog name of the current single-page image.

[0058] If object detection fails, it checks whether the title of the current page image meets the title rules. If the title meets the title rules, it filters the title for special characters and uses the filtered result as the definition catalog name of the current page image. If the title does not meet the title rules, it checks whether the catalog name of the previous page image exists. If it does, it uses the catalog name of the previous page image as the definition catalog name of the current page image. If it does not exist, it names the current page image as something else as the definition catalog name of the current page image.

[0059] Furthermore, object detection is specifically achieved through an object detection neural network, which can be, but is not limited to, YOLO v3, RefineDet, M2Det, or Spiking-YOLO network models.

[0060] Step S4, when cataloging the current single-page image using predefined rule matching, such as... Figure 3 As shown, cataloging shall be carried out according to the following steps S41-S43:

[0061] Step S41: Obtain predefined rules and match the OCR recognition results and title analysis results with the predefined rules;

[0062] Among them, the predefined rule is defined as "If the content contains xx and / or the title contains xx, then the catalog name result is xx". The title in "the title contains xx" is also obtained through layout parsing. Multiple rules can be entered for the same catalog name result and their AND-OR relationship can be defined.

[0063] Step S42: If the OCR recognition result and the title analysis result satisfy the matching condition of one of the predefined rules, then the corresponding catalog name result of the predefined rule that satisfies the matching condition shall be used as the definition catalog name of the current single-page image.

[0064] Step S43: If the OCR recognition result and the title analysis result do not meet the matching conditions of any predefined rule, check if there is a result for the catalog name of the previous single-page image. If there is a result, use the catalog name of the previous single-page image as the defined catalog name of the current single-page image. If there is no result, name the current single-page image "Other" as the defined catalog name of the current single-page image. In other words, if no rule is matched, check if there is a result for the catalog name of the previous page document. If there is a result, follow the catalog name of the previous page document. If there is no result, name it "Other" as the defined catalog name.

[0065] Step S5, as follows Figure 4 As shown, the system obtains the defined catalog name and entity extraction rules, and determines whether named entity extraction is required. If named entity extraction is required, a natural language processing algorithm is used to extract the named entities. After the named entity extraction is completed, a pre-set concatenation method is obtained. The extracted results are then concatenated with the defined catalog name and the entity using the pre-set concatenation method. The concatenated result is used as the final catalog name of the current single-page image. If named entity extraction is not required, the defined catalog name is directly used as the final catalog name of the current single-page image.

[0066] In other words, for some single-page images, key entity information can be extracted from the content, such as the case number of the judgment, the name on the identification document, and the name of the organization. After S3 or S4 cataloging, a defined catalog name is obtained, and it is matched whether extraction is needed. If extraction is needed, named entity extraction is performed. After extraction, a predefined concatenation method is obtained, and the extracted results are concatenated with the catalog name and entities using the predefined concatenation method. The concatenated result forms a new catalog name, which is used as the final catalog name.

[0067] Named entity extraction can be implemented using natural language processing algorithms, specifically hidden Markov models, the Viterbi algorithm, or the Aho-Corasick algorithm.

[0068] (1) Hidden Markov Model is a statistical model that is widely used in the field of natural language processing. Generally, a hidden Markov model can be concisely represented by a triple λ = (A, B, π). The hidden Markov model is actually an extension of the standard Markov model, which adds the set of observable states and the probabilistic relationship between these states and the hidden states.

[0069] (2) The Viterbi algorithm is a dynamic programming algorithm with the goal of finding the optimal path. After calculating the normalized probability matrix and transition probability matrix of each token, the optimal probability output of the entire text sentence is finally obtained according to the Viterbi algorithm.

[0070] (3) The Aho-Corasick algorithm, or AC algorithm for short, preprocesses the pattern string into a deterministic finite state automaton and can finish after scanning the text once; its complexity is O(n), which is independent of the number and length of the pattern string.

[0071] The automaton accepts characters in the order of the text characters and performs state transitions. These states cache the jump and output situations in three cases: "successful transition by character (but not the end of the pattern string)," "successful transition by character (the end of the pattern string)," and "failed transition by character," thus reducing complexity.

[0072] Step S6: Continuously repeat steps S2-S5 until all single-page images in the stored images have been cataloged. If there are uncataloged single-page images, repeat steps S2-S5 until all single-page images have been cataloged.

[0073] A complete flowchart of this method is shown below. Figure 5 As shown.

[0074] This application provides a detailed method for intelligent cataloging of electronic case files, including extracting material titles through layout analysis, cataloging based on OCR recognition results and coordinate style titles through layout analysis, classifying and cataloging images through object detection image classification, performing customized cataloging for different business scenarios through predefined rules, and extracting important information from images through NLP named entity extraction. This process is repeated until all single-page images in the stored files are cataloged. This application proposes two cataloging approaches: layout analysis cataloging and predefined rule matching cataloging, applicable to diverse materials. The intelligent cataloging method provided by this application can improve the accuracy of cataloging diverse materials to over 90%, while eliminating the need for extensive manual verification, reducing maintenance and optimization costs, and improving classification efficiency.

[0075] In addition, using an object detection neural network for automatic object detection during layout parsing and cataloging can further reduce manual costs and improve classification efficiency.

[0076] It should be understood that, although Figure 1-5 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1-5 At least some of the steps in the process may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the steps or stages in other steps.

[0077] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 6As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements an intelligent cataloging method for electronic case files. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0078] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0079] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program relating to all or part of the processes in the methods of the above embodiments.

[0080] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon relating to all or part of the processes in the methods of the above embodiments.

[0081] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0082] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0083] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for intelligent cataloging of electronic case files, characterized in that, include: Step S1: Obtain the electronic file to be cataloged and determine the file type of the electronic file to be cataloged; If the electronic file to be cataloged is an image file, it is stored directly. If the electronic file to be cataloged is a layout file, it is split into independent image files page by page and stored. Step S2: Perform OCR recognition on the single-page image in the stored image and obtain the OCR recognition result; The title analysis results are obtained by parsing the text coordinates and text styles, and the title analysis results include the title information of the current single-page image; wherein, the title analysis results obtained by parsing the text coordinates and text styles include: Based on the OCR recognition results, determine whether there are coordinate points in the text style that have a larger font size or bolder font than the surrounding text; When a coordinate point is determined to have a larger font size or bolder font than the surrounding text, the text at that coordinate point and the text extending to the left and right of that coordinate point are obtained, and it is determined whether all the obtained text is centered or centered at the top; if it is determined that all the obtained text is centered or centered at the top, then all the obtained text is parsed as a title. When it is determined that there are no coordinate points with a larger font size or bolder font than the surrounding text, the text of the top center coordinate point and the text extending to the left and right of that coordinate point are obtained, and it is determined whether the left and right margins of the text below the top center coordinate point are filled; if it is determined that the left and right margins of the text below the top center coordinate point are filled, then all the obtained text is parsed as a title. Step S3: When performing layout analysis and cataloging on the current single-page image, obtain the title analysis results and the illustration analysis results based on whether illustration elements can be obtained. Then, catalog according to steps S31-S33: Step S31: Determine whether the current single-page image contains an illustration; Step S32: If it is determined that the current single-page image contains an illustration, determine whether the proportion of the illustration and the text reaches the preset proportion threshold; if it is determined that the proportion of the illustration and the text reaches the preset proportion threshold, then perform target detection on the illustration, and after the target detection is successful, use the catalog name corresponding to the detection result with a confidence level exceeding the preset confidence threshold as the definition catalog name of the current single-page image. Step S33: If it is determined that the current single-page image does not contain illustrations, determine whether the title of the current single-page image meets the title rules; if the title meets the title rules, filter the title for special symbols and use the filtering result as the definition catalog name of the current single-page image; if the title does not meet the title rules, check whether there is a result for the catalog name of the previous single-page image; if there is a result, use the catalog name of the previous single-page image as the definition catalog name of the current single-page image; if there is no result, name the current single-page image "Other" as the definition catalog name of the current single-page image. Step S4: When cataloging the current single-page image using predefined rule matching, cataloging is performed according to steps S41-S43: Step S41: Obtain predefined rules and match the OCR recognition results and the title analysis results with the predefined rules; Step S42: If the OCR recognition result and the title analysis result satisfy the matching condition of one of the predefined rules, then the corresponding catalog name result of the predefined rule that satisfies the matching condition shall be used as the definition catalog name of the current single-page image. Step S43: If the OCR recognition result and the title analysis result do not meet the matching conditions of any predefined rule, check whether there is a result for the catalog name of the previous single-page image. If there is a result, use the catalog name of the previous single-page image as the defined catalog name of the current single-page image. If there is no result, name the current single-page image as something else as the defined catalog name of the current single-page image. Step S5: Obtain the defined catalog name and entity extraction rules, and determine whether named entity extraction is required. If named entity extraction is required, use a natural language processing algorithm to extract named entities. After the named entity extraction is completed, obtain a pre-set splicing method, and use the pre-set splicing method to splice the extracted results of the named entity extraction with the defined catalog name and the entity. Use the spliced ​​result as the final catalog name of the current single-page image. If named entity extraction is not required, directly use the defined catalog name as the final catalog name of the current single-page image. Step S6: Repeat steps S2-S5 until all single-page images in the stored images are cataloged.

2. The intelligent cataloging method for electronic case files according to claim 1, characterized in that, The image file is specifically a JPEG file or a PNG file, and the layout file is specifically a PDF file or an OFD file.

3. The intelligent cataloging method for electronic case files according to claim 1, characterized in that, Object detection is performed using an object detection neural network.

4. The intelligent cataloging method for electronic case files according to claim 3, characterized in that, The target detection neural network is specifically YOLO v3, RefineDet, M2Det, or Spiking-YOLO.

5. The intelligent cataloging method for electronic case files according to claim 1, characterized in that, The preset confidence threshold is 0.

9.

6. The intelligent cataloging method for electronic case files according to claim 1, characterized in that, Step S32 further includes: If the ratio of illustration to text does not reach the preset ratio threshold, then it is determined whether the title of the current single-page image meets the title rules; if the title meets the title rules, then the title is filtered for special symbols, and the filtering result is used as the definition catalog name of the current single-page image; if the title does not meet the title rules, then it is checked whether there is a result for the catalog name of the previous single-page image; if there is a result, then the catalog name of the previous single-page image is used as the definition catalog name of the current single-page image; if there is no result, then the current single-page image is named "Other" as the definition catalog name of the current single-page image. If object detection fails, it checks whether the title of the current page image meets the title rules. If the title meets the title rules, it filters the title for special characters and uses the filtered result as the definition catalog name of the current page image. If the title does not meet the title rules, it checks whether the catalog name of the previous page image exists. If it does, it uses the catalog name of the previous page image as the definition catalog name of the current page image. If it does not exist, it names the current page image as something else as the definition catalog name of the current page image.

7. The intelligent cataloging method for electronic case files according to claim 1, characterized in that, The natural language processing algorithm is a hidden Markov model algorithm, a Viterbi algorithm, or an Aho-Corasick algorithm.

8. The intelligent cataloging method for electronic case files according to claim 1, characterized in that, The entities mentioned include the case number of the judgment, the name on the identification document, and the name of the organization.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images