A method and device for digital processing of paper documents

Abstract

The embodiment of the present invention discloses a method and device for digitizing paper documents. It includes: processing the content of paper documents to be classified, extracting key information of the paper documents to be classified; determining standard information matching the key information in the document type information library, and determining the document type to which the standard information belongs as the document type of the paper documents to be classified; generating structured information of the paper documents to be classified based on the structured format corresponding to the document type and the content of the paper documents to be classified; and associating and storing the paper documents to be classified, the document type of the paper documents to be classified, and the structured information. It can realize automatic classification and automatic document recording of documents, replace manual operations, have higher accuracy and processing efficiency, and make document data more specific and detailed.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of computer data processing, and in particular to a method and device for digitizing paper documents. Background Art

[0002] With the rapid development of information technology, document digitization has gradually become a development trend. However, there are still a large number of paper documents in government agencies, enterprises, schools and other units. Although paper documents can be stored in the database by scanning and entering, thus realizing the electronicization of paper documents, these paper documents still need to be manually classified. Due to the complex and changeable attributes of different documents, the large amount of data, and the different classification standards of each person, the efficiency and accuracy of document recording and classification are low, and the classification lacks objectivity. Summary of the invention

[0003] The content of the present invention is used to introduce the concepts in a brief form, which will be described in detail in the detailed implementation section below. The content of the present invention is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

[0004] The present invention proposes a method and device for digitizing paper documents to solve the technical problems mentioned in the above background technology section.

[0005] A first aspect of the present invention provides a method for digitizing a paper document, comprising:

[0006] Acquire a document image of a paper document to be classified, recognize the document image based on optical characters, and extract key information from the recognition result of the document image; determine standard information matching the key information in a document type information library, and determine the document type to which the standard information belongs as the document type of the paper document to be classified; generate structured information of the paper document to be classified based on a structured format corresponding to the document type and the content of the paper document to be classified; and store the paper document to be classified, the document type of the paper document to be classified, and the structured information in an associated manner.

[0007] Furthermore, the extracting of the key information from the recognition result of the document image includes: dividing the recognition result into multiple sub-recognition results according to the content type, each sub-recognition result corresponding to a different content type; extracting information from each sub-recognition result based on the information extraction rule corresponding to the content type to obtain the key information.

[0008] Furthermore, the method also includes: determining the characteristics of each document type based on existing classified documents; determining multiple standard information corresponding to each document type according to the characteristics of each document type; and storing the document type and the multiple standard information corresponding to the document type in the document type information library.

[0009] Further, the determining the standard information matching the key information in the document type information library includes: determining the standard information matching the key information from a plurality of standard information in the document type information library.

[0010] Furthermore, determining standard information matching the key information in the document type information library includes: inputting the key information and the standard information into a preset similarity model to obtain the similarity between the key information and the standard information; and determining standard information matching the key information based on the similarity.

[0011] Furthermore, the preset similarity model is a feature vector extraction model based on bidirectional coding representation, and the key information and the standard information are input into the preset similarity model to obtain the similarity between the key information and the standard information, including: inserting a first feature character at the front end of the key information, inserting a second feature character between the key information and the standard information to obtain an input sequence; inputting the input sequence into the similarity model, and outputting the similarity between the key information and the standard information.

[0012] Furthermore, the generating of structured information of the paper document to be classified based on the structured format corresponding to the document type and the content of the paper document to be classified includes: determining a plurality of description fields from the structured format corresponding to the document type; extracting the description information corresponding to each of the description fields from the content of the paper document to be classified; integrating the plurality of description fields and the description information corresponding to each of the description fields to generate the structured information of the paper document to be classified.

[0013] A second aspect of the present invention provides a device for digitizing paper documents, comprising:

[0014] A processing module is used to obtain a document image of a paper document to be classified, recognize the document image based on optical characters, and extract key information from the recognition result of the document image; a determination module is used to determine standard information matching the key information in a document type information library, and determine the document type to which the standard information belongs as the document type of the paper document to be classified; a generation module is used to generate structured information of the paper document to be classified based on a structured format corresponding to the document type and the content of the paper document to be classified; a storage module is used to associate and store the paper document to be classified, the document type of the paper document to be classified, and the structured information.

[0015] Furthermore, the processing module is further used to: divide the recognition result into multiple sub-recognition results according to the content type, each sub-recognition result corresponds to a different content type; extract information from each sub-recognition result based on the information extraction rule corresponding to the content type to obtain the key information.

[0016] Furthermore, the storage module is further used to determine the characteristics of each document type based on existing classified documents; determine multiple standard information corresponding to each document type according to the characteristics of each document type; and store the document type and the multiple standard information corresponding to the document type in the document type information library.

[0017] A third aspect of the present invention provides an electronic device, comprising: one or more processors; a storage device storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the above method.

[0018] According to a fourth aspect of the present invention, a computer-readable medium is provided, on which a computer program is stored, wherein the program implements any of the above methods when executed by a processor.

[0019] The present invention can realize automatic recognition, digitization, automatic classification and automatic recording of paper documents, and has higher recording accuracy and processing efficiency than manual work. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0021] Figure 1 is a flow chart of a method for digitizing a paper document provided by an embodiment of the present invention;

[0022] Figure 2 It is a flowchart of the process of establishing and updating a document type information library in a method for digitizing paper documents provided by an embodiment of the present invention;

[0023] Figure 3 is a flow chart of standard information determination of a method for digitizing paper documents provided by an embodiment of the present invention;

[0024] Figure 4 It is a structural schematic diagram of a device for digitizing paper documents provided by an embodiment of the present invention;

[0025] Figure 5 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.

[0027] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other.

[0028] It should be noted that the concepts such as "first" and "second" mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0029] It should be noted that the modifications of “one” and “plurality” mentioned in the present invention are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as “one or more”.

[0030] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only used for illustrative purposes, and are not used to limit the scope of these messages or information.

[0031] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0032] Figure 1The process of the method for digitizing paper documents according to the present invention is shown. The method for digitizing paper documents comprises the following steps:

[0033] Step S101 : acquiring a document image of a paper document to be classified, recognizing the document image based on optical characters, and extracting key information from the recognition result of the document image.

[0034] In the embodiment of the present invention, the paper document to be classified may be a resume, a contract, a certificate, etc. Before classifying the document, one or more keywords that can characterize the core content of the paper document to be classified need to be extracted.

[0035] As an optional implementation of the embodiment of the present invention, in order to facilitate the identification of key information of paper documents to be classified, it is necessary to digitize the paper documents to be classified. This step can be implemented by the following scheme: obtaining a document image of the paper document to be classified; recognizing the document image based on optical characters, and extracting the key information from the recognition result of the document image.

[0036] Specifically, first, the classified document can be scanned by a scanner to obtain a document image. After the document image is determined, the document image is recognized based on the optical character recognition technology (OCR) to obtain the character recognition result corresponding to the paper document to be classified, and then the character recognition result is semantically analyzed to extract the key information. The character recognition result of the document is composed of several segments of content. Here, the several segments of content can be split or clustered, and the key information is extracted based on the splitting result or clustering result.

[0037] As an optional implementation, key information is extracted based on the splitting results. Specifically, if the recognition result is pure text content, each paragraph of text content can be split into sentences, and the type of words in the text can be determined by the text structure of the sentence (syntax, part of speech, etc.), and the keywords of each sentence are determined accordingly. For example, the text content includes "I am mainly responsible for the traffic growth of the search business in Company A", and the subject, predicate and object analysis is performed to obtain the entity words: Company A, search business and traffic growth, and this entity word is used as a candidate keyword. After performing the above-mentioned speech analysis on each sentence of each paragraph of text content, multiple candidate keywords will be obtained, and the frequency of occurrence of each candidate keyword is counted, and the target keyword is determined as the key information according to the frequency of each candidate keyword. Generally speaking, the higher the frequency of occurrence of the candidate keyword, the more it can represent the core content of the paper document to be classified. In the process of determining the target keyword, different strategies can be adopted. For example, when the frequency of occurrence of each keyword is high, the selected threshold can be increased; when the frequency of occurrence of each keyword is low, the selected threshold can be lowered. For another example, when the frequencies of occurrence of various keywords vary greatly, the keywords with higher frequencies can be selected.

[0038] As another optional implementation, key information is extracted based on clustering results. Specifically, the recognition result is divided into multiple sub-recognition results according to the content type, and each sub-recognition result corresponds to a different content type; information is extracted from each sub-recognition result based on the information extraction rule corresponding to the content type to obtain the key information.

[0039] Generally speaking, there are many types of content in a document: text, image, table, etc. For text content, the more frequently a keyword appears in the text, the more it can represent the core content of the text; for image content, the name of the entity in the picture or the category of the entity can best represent the meaning of the picture (for example, a portrait photo in a resume, as long as the entity category is identified as a person, it represents the core content of the document); for table content, since the table itself has a strong structured attribute, the bibliographic information in the table can represent the key information represented by the table. Therefore, information extraction rules corresponding to different content types can be pre-set: image content extracts the name or category of the entity in the picture, table content extracts the bibliographic information in the table, and text content extracts the most frequently appearing keywords in the text. Cluster the above-mentioned content segments according to the content type to obtain the sub-recognition results corresponding to each content type, and then extract candidate keywords for each sub-recognition result according to the above-mentioned extraction rules, and perform similarity elimination on all candidate keywords to obtain keywords representing the core content of the document.

[0040] Step S102: determining standard information matching the key information in a document type information library, and determining the document type to which the standard information belongs as the document type of the paper document to be classified.

[0041] The preset document type information library stores multiple standard information corresponding to each document type. Before performing this step, Figure 2 As shown, the embodiment of the present invention also includes a process of establishing and updating a document type information library.

[0042] Step S201, determining the characteristics of each document type based on existing classified documents.

[0043] Step S202: determining a plurality of standard information corresponding to each document type according to the characteristics of each document type.

[0044] Step S203: storing the document type and a plurality of standard information corresponding to the document type in the document type information library.

[0045] First, by further analyzing the contents of the classified documents, the basic characteristics of various document types are determined. As shown in Table 1, a comparison table of several document types and characteristics is shown.

[0046] Table 1

[0047] Document Type feature contract Involving multiple types of entity objects, including keywords related to various rights and obligations CV There are keywords related to people (character images) and work experience Certificate Including keywords such as numbers (certificate number), time, issuing unit, etc. paper It also includes clue information such as "background, problem, plan, conclusion"

[0048] Secondly, analyze the characteristics of each document type, abstract the commonalities of the characteristics of each document type, and summarize them into multiple standard information. For example, for the characteristics of contract-type documents, summarize several synonyms and antonyms related to rights and obligations, as well as synonyms and antonyms related to Party A, Party B, location, and subject matter as standard information; for the characteristics of resume-type documents, summarize characters, boys, girls, job search intentions, work experience, work experience, project experience, etc. as standard information; for the characteristics of certificate-type documents, summarize random number strings, time, organization or unit names as standard information; for the characteristics of paper-type documents, summarize synonyms and antonyms related to background, problem, plan, and conclusion as standard information.

[0049] Finally, an association relationship is established between each document type and the standard relationship corresponding to the document type, and stored in the document type information library.

[0050] As some optional implementation modes of the embodiments of the present invention, the determining the standard information matching the key information in the document type information library includes: determining the standard information matching the key information from a plurality of standard information in the document type information library.

[0051] Specifically, since the key information may be composed of multiple keywords, when searching for standard information that matches the key information, the similarity between each keyword and multiple standard information under each document type can be determined one by one, and then the similarity of each keyword is weighted and summed to determine the average similarity between the key information and the document type. Finally, based on the average similarity of each document type, the standard information that matches the key information is determined.

[0052] In addition, one hot encoding can also be used to map the key information into a corresponding first vector matrix, and map the standard information of each document type into a second vector matrix for search. When the similarity between the first vector matrix and the second vector matrix of the standard information of a certain document type meets the requirements, the standard information is determined to be the standard information matching the key information. It should be noted that the method for vectorizing the key information and the standard information can adopt a frequency based embedding method.

[0053] As some other optional implementations of the embodiments of the present invention, Figure 3 As shown, determining standard information matching the key information in the document type information library includes:

[0054] Step S1021: input the key information and the standard information into a preset similarity model to obtain the similarity between the key information and the standard information.

[0055] Step S1022: Based on the similarity, determine standard information that matches the key information.

[0056] That is, the similarity between the key information and a plurality of standard information corresponding to each document type is calculated, and the standard information with the highest similarity is determined as the standard information matching the key information.

[0057] As an optional implementation of this step, the similarity model includes a feature vector extraction model. For any document type, first input multiple keywords in the key information and multiple standard keywords in the standard information of the document type into the feature vector extraction model respectively to obtain a first vector of the key information and a second vector of the standard information; then calculate the cosine value of the first vector and the second vector as the similarity between the key information and the standard information. Furthermore, the feature vector extraction model needs to be trained. When training the feature vector extraction model, first obtain a training sample, which includes multiple keywords, and mark the vector corresponding to each keyword.

[0058] As some other optional implementations of this step, the similarity between the key information and the standard information can be directly obtained based on the similarity model. Prior to this, the similarity model needs to be trained, and the training sample includes key information and multiple standard information, and the key information and multiple standard information are annotated by manual annotation. For example, there is a group of keywords Q in the key information, and multiple standard information include D1, D2, D3, D4, and D5, where the degree of association between each standard information and the key information can be determined based on the document type corresponding to the standard information and the document type from which the key information comes. If the document types corresponding to D1 and Q are both lease contracts, the correlation between the two is the highest; if the document type corresponding to D1 is a lease contract, and the document type corresponding to Q is a gift contract, the correlation between the two is slightly lower. In general, the correlation can be annotated by scoring, such as [Q, D1, 1], [Q, D2, 0.8]... [Q, D5, 0.2], etc. The training samples formed by multiple key and multiple standard information are used as the training sample set to train the similarity model. When the error between the similarity output by the model and the annotated similarity is within the preset range, it indicates that the similarity model training is completed.

[0059] As another optional implementation of this step, the preset similarity model can be a feature vector extraction model based on BERT representation (bidirectional coding representation). Specifically, a first feature character is inserted at the front end of the key information, and a second feature character is inserted between the key information and the standard information to obtain an input sequence; the input sequence is input into the similarity model, and the similarity between the key information and the standard information is output.

[0060] Furthermore, the first characteristic character tag is inserted at the beginning of the key information, and the second characteristic character tag is inserted between the key information and the standard information, and the whole is used as the input sequence after tagging. In order to predict the similarity between the key information and the standard information, the output of the tag can be transformed into a 2×1 vector through the classification layer in the similarity model, and the similarity probability between the two can be calculated through the softmax function.

[0061] The embodiment of the present disclosure calculates the similarity between key information and standard information through the BERT model, which facilitates the subsequent determination of document types and can effectively improve the accuracy of document classification.

[0062] Step S103: generating structured information of the paper document to be classified based on the structured format corresponding to the document type and the content of the paper document to be classified.

[0063] Since some documents have a lot of content, even if their corresponding types are determined, it is not possible to quickly understand the main content of each document in subsequent searches. Therefore, this solution can not only classify documents, but also perform structured extraction of the content based on the structured format corresponding to the document type to which each document belongs.

[0064] Specifically, this step can be implemented through the following scheme: determine multiple catalog fields from the structured format corresponding to the document type; extract the catalog information corresponding to each of the catalog fields from the content of the paper document to be classified; integrate the multiple catalog fields and the catalog information corresponding to each of the catalog fields to generate structured information of the paper document to be classified.

[0065] Generally speaking, the structured bibliographic information of the core content of each document type is the same. For example, the bibliographic fields of resume-type documents include headshot, industry field, years of work, work experience, graduation school, etc.; the bibliographic fields of contract-type documents include Party A, Party B, subject matter, by, performance date, etc. The display forms corresponding to different bibliographic fields are different. For example, the "headshot" bibliographic field needs to be displayed in the form of a picture, and the "work experience" bibliographic field needs to be displayed in the form of a table or line. Therefore, after determining the bibliographic fields of each document type, the structured layout corresponding to the document type can be determined in combination with the display form corresponding to the bibliographic field.

[0066] Furthermore, for a paper document to be classified whose document type has been determined, multiple description fields can be determined according to its structured layout, and the content corresponding to each description field can be extracted from the character recognition result as the description information, and the description information corresponding to each description field can be extracted from the content of the paper document to be classified; then, according to the structured layout, each description information corresponding to the description field is added to the corresponding position of the structured layout to generate the structured information of the paper document to be classified.

[0067] Step S104: The paper documents to be classified, the document types of the paper documents to be classified, and the structured information are stored in association with each other.

[0068] In this step, the attribute information of the paper document to be classified, such as the document ID and the user ID involved in the document, can be first determined, and the attribute information of the paper document to be classified, the document type, the paper document to be classified and the corresponding structured information are associated and stored. It should be noted that since the purpose of structured information is to quickly understand the core content of the document, the structured information can be stored in the format of an image, and the image can be used as the cover of the document file, which further improves the convenience and speed of document query preview.

[0069] As some optional implementations of the present invention, in some document types, a knowledge graph of the document is established based on the structured information of the document to facilitate the subsequent recommendation of similar documents. For example, for resume-type documents, a knowledge graph index can be established based on industry fields and years of work experience, so that relevant personnel can quickly recall relevant documents when screening candidates, or in response to relevant personnel reading a document for more than a preset time, similar resumes can be recommended to them based on the knowledge graph.

[0070] The above-mentioned embodiments of the present invention have the following beneficial effects: through the document classification method of some embodiments of the present invention, automatic classification and automatic document recording of documents can be realized, replacing manual operations, with higher accuracy and processing efficiency, making document data more specific and detailed.

[0071] Continue to refer Figure 4 As an implementation of the above methods in the above figures, the present invention provides some embodiments of a document classification device, and these device embodiments are Figure 1 Corresponding to the above method embodiments, the device can be specifically applied to various electronic devices.

[0072] like Figure 4 As shown, the document classification device of some embodiments includes: a processing module 401 , a determination module 402 , a generation module 403 and a storage module 404 .

[0073] Among them, the processing module 401 is used to obtain the document image of the paper document to be classified, recognize the document image based on optical characters, and extract key information from the recognition result of the document image; the determination module 402 is used to determine the standard information matching the key information in the document type information library, and determine the document type to which the standard information belongs as the document type of the paper document to be classified; the generation module 403 is used to generate the structured information of the paper document to be classified based on the structured layout corresponding to the document type and the content of the paper document to be classified; the storage module 404 is used to associate and store the paper document to be classified, the document type of the paper document to be classified and the structured information.

[0074] The processing module 401 is further used to divide the recognition result into multiple sub-recognition results according to the content type, and each sub-recognition result corresponds to a different content type; based on the information extraction rule corresponding to the content type, each sub-recognition result is respectively subjected to information extraction to obtain the key information.

[0075] The storage module 404 is also used to determine the characteristics of each document type based on existing classified documents; determine multiple standard information corresponding to each document type according to the characteristics of each document type; and store the document type and the multiple standard information corresponding to the document type in the document type information library.

[0076] The determination module 402 is further used to determine standard information matching the key information from a plurality of standard information in the document type information library.

[0077] The determination module 402 is further used to input the key information and the standard information into a preset similarity model to obtain the similarity between the key information and the standard information; and based on the similarity, determine the standard information matching the key information.

[0078] The determination module 402 is further used to determine that the preset similarity model is a feature vector extraction model based on bidirectional coding representation, insert a first feature character at the front end of the key information, insert a second feature character between the key information and the standard information, and obtain an input sequence; input the input sequence into the similarity model, and output the similarity between the key information and the standard information.

[0079] Optionally, the generation module 403 is further used to determine multiple catalog fields from the structured format corresponding to the document type; extract the catalog information corresponding to each of the catalog fields from the content of the paper document to be classified; integrate the multiple catalog fields and the catalog information corresponding to each of the catalog fields to generate structured information of the paper document to be classified.

[0080] It is understandable that the modules recorded in the device are similar to those in the reference Figure 1 Therefore, the operations, features and beneficial effects described above for the method are also applicable to the device and the modules contained therein, and will not be described in detail here.

[0081] Reference below Figure 5 , which shows an electronic device (eg, Figure 1 Schematic diagram of the structure of an electronic device). Figure 5 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0082] like Figure 5As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage device 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the electronic device are also stored. The processing device 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 909 is also connected to the bus 904.

[0083] Typically, the following devices may be connected to the I / O interface 909: input devices 906 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 907 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 908 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 909. The communication devices 909 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although Figure 5 An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead. Figure 5 Each block shown in the figure may represent one device, or may represent multiple devices as required.

[0084] In particular, according to some embodiments of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, some embodiments of the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In some such embodiments, the computer program can be downloaded and installed from the network through the communication device 909, or installed from the storage device 908, or installed from the ROM 902. When the computer program is executed by the processing device 901, the above-mentioned functions defined in the method of some embodiments of the present invention are executed.

[0085] It should be noted that the computer-readable medium in some embodiments of the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above.

[0086] The above descriptions are only some preferred embodiments of the present invention and the explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with the technical features with similar functions invented in the embodiments of the present invention (but not limited to) to form a technical solution.

Claims

1. A method for digitizing a paper document, comprising: Obtaining a document image of a paper document to be classified, recognizing the document image based on optical characters, clustering the recognition results of the document image to obtain a plurality of sub-recognition results of different content types, extracting the name or category of the entity in the image as a candidate keyword for the sub-recognition result of the image content type, extracting the bibliographic information in the table as a candidate keyword for the sub-recognition result of the table content type, extracting the keyword with the highest frequency of occurrence in the text as a candidate keyword for the sub-recognition result of the text content type, performing similarity elimination on all the candidate keywords, and obtaining key information representing the core content of the document; determining characteristics of each document type based on the classified documents; Determine multiple standard information corresponding to each document type according to the characteristics of each document type; store the document type and the multiple standard information corresponding to the document type in a document type information library; Determine the similarity between each keyword in the key information and multiple standard information corresponding to each document type in the document type information library one by one, then perform weighted summation on the similarity of each keyword, determine the average similarity between the key information and multiple standard information corresponding to the document type, and determine the standard information matching the key information according to the average similarity; Determining the document type to which the standard information belongs as the document type of the paper document to be classified; Determine a plurality of bibliographic fields from the structured format corresponding to the document type, extract bibliographic information corresponding to each bibliographic field from the content of the paper document to be classified, integrate the plurality of bibliographic fields and the bibliographic information corresponding to each bibliographic field, and generate structured information of the paper document to be classified; The paper documents to be classified, the document types of the paper documents to be classified, and the structured information are stored in association.

2. A paper document digitization processing device, comprising: A processing module is used to obtain a document image of a paper document to be classified, recognize the document image based on optical characters, cluster the recognition results of the document image to obtain a plurality of sub-recognition results of different content types, extract the name or category of the entity in the image as a candidate keyword for the sub-recognition result of the image content type, extract the bibliographic information in the table as a candidate keyword for the sub-recognition result of the table content type, extract the keyword with the highest frequency of occurrence in the text as a candidate keyword for the sub-recognition result of the text content type, perform similarity elimination on all the candidate keywords, and obtain key information representing the core content of the document; A standard information acquisition module for determining the characteristics of each document type based on the classified documents; Determine multiple standard information corresponding to each document type according to the characteristics of each document type; store the document type and the multiple standard information corresponding to the document type in a document type information library; A determination module, which determines the similarity between each keyword in the key information and a plurality of standard information corresponding to each document type in the document type information library one by one, and then performs a weighted summation on the similarity of each keyword to determine an average similarity between the key information and a plurality of standard information corresponding to the document type, and determines the standard information matching the key information according to the average similarity; Determining the document type to which the standard information belongs as the document type of the paper document to be classified; A generating module, used to determine a plurality of bibliographic fields from the structured format corresponding to the document type, extract bibliographic information corresponding to each bibliographic field from the content of the paper document to be classified, integrate the plurality of bibliographic fields and the bibliographic information corresponding to each bibliographic field, and generate structured information of the paper document to be classified; The storage module is used to store the paper documents to be classified, the document types of the paper documents to be classified and the structured information in an associated manner.

Images