Document processing, document display method and device, computer equipment and storage medium

Abstract

The application relates to a bill processing method and device, a bill display method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring a bill image corresponding to a target bill and a user identifier corresponding to the bill image; identifying a target bill type corresponding to the bill image, performing keyword identification on the bill image based on the target bill type, and obtaining a target keyword set corresponding to the target bill; acquiring a historical keyword set corresponding to a historical bill matched with the user identifier; a bill set composed of the target bill and the historical bill comprises bills corresponding to different bill providers; calculating a similarity corresponding to the target bill and the historical bill based on the target keyword set and the historical keyword set; and establishing a target association relationship between the target bill and the historical bill based on the similarity, wherein the target association relationship is used for bill display. The method can improve the information acquisition efficiency of the bill.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a document processing and document display method, apparatus, computer equipment, and storage medium. Background Art

[0002] In daily life, different institutions often store and maintain their own document data. For example, Hospital A and Hospital B each maintain their own medical records for users. Because these institutions independently store and maintain data, "data silos" gradually form.

[0003] In traditional technology, if a user wants to view a document, he or she needs to visit the institution corresponding to the document separately. To view documents provided by different institutions, he or she needs to visit different institutions. Users can only view similar documents provided by the same institution, and cannot view similar documents provided by different institutions, resulting in low information acquisition efficiency. Summary of the Invention

[0004] Based on this, it is necessary to provide a document processing, document display method, device, computer equipment and storage medium that can improve the efficiency of information acquisition in response to the above technical problems.

[0005] A document processing method, comprising:

[0006] Obtaining a document image corresponding to a target document and a user identifier corresponding to the document image;

[0007] Identifying a target document type corresponding to the document image, performing keyword recognition on the document image based on the target document type, and obtaining a target keyword set corresponding to the target document;

[0008] Obtaining a historical keyword set corresponding to a historical document matching the user identifier, wherein the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on historical images corresponding to the historical documents;

[0009] Calculating the similarity between the target document and the historical document based on the target keyword set and the historical keyword set;

[0010] A target association relationship is established between the target document and the historical documents based on the similarity, and the target association relationship is used for document display.

[0011] In one embodiment, identifying the target document type corresponding to the document image includes:

[0012] Performing text recognition on a preset image area in the document image to obtain a target title corresponding to the target document;

[0013] The target document type is determined based on the target title.

[0014] In one embodiment, the target keyword set and the historical keyword set both include keywords of multiple dimensions;

[0015] The calculating the similarity between the target document and the historical document based on the target keyword set and the historical keyword set includes:

[0016] In the target keyword set and the historical keyword set, keyword differences are calculated based on the target keywords and historical keywords corresponding to the same dimension to obtain keyword differences corresponding to each dimension;

[0017] Obtain target attention information corresponding to each dimension;

[0018] The similarity between the target document and the historical document is calculated based on the keyword differences and attention information corresponding to the various dimensions.

[0019] In one embodiment, obtaining target attention information corresponding to each dimension includes:

[0020] Obtain the training keyword differences corresponding to the training document pairs in each dimension, and obtain the training similarity corresponding to the training document pairs;

[0021] Get the initial attention information corresponding to each dimension;

[0022] Calculating the initial similarity corresponding to the training document pair based on the training keyword differences corresponding to the respective dimensions and the initial attention information;

[0023] A similarity loss value is calculated based on the initial similarity and training similarity corresponding to the training document, and the initial attention information corresponding to each dimension is adjusted based on the similarity loss value until the convergence condition is met, thereby obtaining the target attention information corresponding to each dimension.

[0024] In one embodiment, establishing a target association relationship between the target document and the historical document based on the similarity includes:

[0025] Acquire association relationship mapping information, wherein the association relationship mapping information includes reference similarities corresponding to at least two candidate association relationships; different candidate association relationships correspond to different document display modes;

[0026] The similarity is matched with the reference similarity in the association relationship mapping information, and the target association relationship is determined from various candidate association relationships based on the matching result.

[0027] In one embodiment, the document provider refers to the document attribution agency corresponding to the document, and the document attribution agency is used to collect data and analyze data through electronic equipment to obtain data analysis results, and send the data analysis results to the terminal corresponding to the user identifier. The data analysis results include user vital sign data corresponding to the user identifier.

[0028] A document processing device, comprising:

[0029] An image acquisition module, configured to acquire a document image corresponding to a target document and a user identifier corresponding to the document image;

[0030] a keyword recognition module, configured to identify a target document type corresponding to the document image, perform keyword recognition on the document image based on the target document type, and obtain a target keyword set corresponding to the target document;

[0031] a historical information acquisition module, configured to acquire a historical keyword set corresponding to historical documents matching the user identifier, wherein the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on historical images corresponding to the historical documents;

[0032] A similarity calculation module, configured to calculate the similarity between the target document and the historical document based on the target keyword set and the historical keyword set;

[0033] The association relationship establishing module is used to establish a target association relationship between the target document and the historical document based on the similarity, and the target association relationship is used for document display.

[0034] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0035] Obtaining a document image corresponding to a target document and a user identifier corresponding to the document image;

[0036] Identifying a target document type corresponding to the document image, performing keyword recognition on the document image based on the target document type, and obtaining a target keyword set corresponding to the target document;

[0037] Obtaining a historical keyword set corresponding to a historical document matching the user identifier, wherein the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on historical images corresponding to the historical documents;

[0038] Calculating the similarity between the target document and the historical document based on the target keyword set and the historical keyword set;

[0039] A target association relationship is established between the target document and the historical documents based on the similarity, and the target association relationship is used for document display.

[0040] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0041] Obtaining a document image corresponding to a target document and a user identifier corresponding to the document image;

[0042] Identifying a target document type corresponding to the document image, performing keyword recognition on the document image based on the target document type, and obtaining a target keyword set corresponding to the target document;

[0043] Obtaining a historical keyword set corresponding to a historical document matching the user identifier, wherein the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on historical images corresponding to the historical documents;

[0044] Calculating the similarity between the target document and the historical document based on the target keyword set and the historical keyword set;

[0045] A target association relationship is established between the target document and the historical documents based on the similarity, and the target association relationship is used for document display.

[0046] A document processing method, comprising:

[0047] Obtaining a document display request sent by a terminal; the document display request carries a target document identifier and a user identifier;

[0048] Obtaining a target association relationship between a target document corresponding to the target document identifier and a historical document corresponding to the user identifier; the document set consisting of the target document and the historical documents includes documents corresponding to different document providers; the target association relationship is established based on a similarity between the target document and the historical documents; the similarity is calculated based on a target keyword set corresponding to the target document and a historical keyword set corresponding to the historical documents; the target keyword set or the historical keyword set is obtained by performing keyword recognition on a current document image based on a current document type corresponding to the current document image; the current document image is the document image corresponding to the target document or the document image corresponding to the historical document;

[0049] Generating document display data corresponding to the document display request based on the target association relationship, and returning the document display data to the terminal, so that the terminal performs document display according to the document display data.

[0050] A document processing device, comprising:

[0051] A request acquisition module, configured to acquire a document display request sent by a terminal; the document display request carries a target document identifier and a user identifier;

[0052] an association relationship acquisition module, configured to acquire a target association relationship between a target document corresponding to the target document identifier and a historical document corresponding to the user identifier; the document set consisting of the target document and the historical documents includes documents corresponding to different document providers; the target association relationship is established based on a similarity between the target document and the historical documents, the similarity being calculated based on a target keyword set corresponding to the target document and a historical keyword set corresponding to the historical documents; the target keyword set or the historical keyword set being obtained by performing keyword recognition on a current document image based on a current document type corresponding to the current document image; the current document image being the document image corresponding to the target document or the document image corresponding to the historical document;

[0053] A display data determination module is used to generate document display data corresponding to the document display request based on the target association relationship, and return the document display data to the terminal so that the terminal performs document display according to the document display data.

[0054] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0055] Obtaining a document display request sent by a terminal; the document display request carries a target document identifier and a user identifier;

[0056] Obtaining a target association relationship between a target document corresponding to the target document identifier and a historical document corresponding to the user identifier; the document set consisting of the target document and the historical documents includes documents corresponding to different document providers; the target association relationship is established based on a similarity between the target document and the historical documents; the similarity is calculated based on a target keyword set corresponding to the target document and a historical keyword set corresponding to the historical documents; the target keyword set or the historical keyword set is obtained by performing keyword recognition on a current document image based on a current document type corresponding to the current document image; the current document image is the document image corresponding to the target document or the document image corresponding to the historical document;

[0057] Generating document display data corresponding to the document display request based on the target association relationship, and returning the document display data to the terminal, so that the terminal performs document display according to the document display data.

[0058] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0059] Obtaining a document display request sent by a terminal; the document display request carries a target document identifier and a user identifier;

[0060] Obtaining a target association relationship between a target document corresponding to the target document identifier and a historical document corresponding to the user identifier; the document set consisting of the target document and the historical documents includes documents corresponding to different document providers; the target association relationship is established based on a similarity between the target document and the historical documents; the similarity is calculated based on a target keyword set corresponding to the target document and a historical keyword set corresponding to the historical documents; the target keyword set or the historical keyword set is obtained by performing keyword recognition on a current document image based on a current document type corresponding to the current document image; the current document image is the document image corresponding to the target document or the document image corresponding to the historical document;

[0061] Generating document display data corresponding to the document display request based on the target association relationship, and returning the document display data to the terminal, so that the terminal performs document display according to the document display data.

[0062] A document display method, comprising:

[0063] Acquire a document collection instruction generated by a trigger operation on a document processing interface, perform image collection on a target document according to the document collection instruction, and obtain a document image corresponding to the target document;

[0064] The document image and the user identifier corresponding to the document image are sent to a server so that the server identifies a target document type corresponding to the document image; keyword recognition is performed on the document image based on the target document type to obtain a target keyword set corresponding to the target document; a historical keyword set corresponding to historical documents matching the user identifier is obtained; similarity between the target document and the historical documents is calculated based on the target keyword set and the historical keyword set; a target association relationship is established between the target document and the historical documents based on the similarity; and document display data is generated based on the target association relationship. The document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on the historical images corresponding to the historical documents.

[0065] The document display data returned by the server is obtained, and the document is displayed on a document display interface based on the document display data.

[0066] A document display device, comprising:

[0067] An image acquisition module is used to obtain a document acquisition instruction generated by a trigger operation on the document processing interface, and to acquire an image of a target document according to the document acquisition instruction to obtain a document image corresponding to the target document;

[0068] a request sending module, configured to send the document image and a user identifier corresponding to the document image to a server, so that the server identifies a target document type corresponding to the document image, performs keyword recognition on the document image based on the target document type to obtain a target keyword set corresponding to the target document, obtains a historical keyword set corresponding to historical documents matching the user identifier, calculates a similarity between the target document and the historical documents based on the target keyword set and the historical keyword set, establishes a target association relationship between the target document and the historical documents based on the similarity, and generates document display data based on the target association relationship, wherein the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on the historical images corresponding to the historical documents;

[0069] The document display module is used to obtain the document display data returned by the server and display the document on the document display interface based on the document display data.

[0070] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0071] Acquire a document collection instruction generated by a trigger operation on a document processing interface, perform image collection on a target document according to the document collection instruction, and obtain a document image corresponding to the target document;

[0072] The document image and the user identifier corresponding to the document image are sent to a server so that the server identifies a target document type corresponding to the document image; keyword recognition is performed on the document image based on the target document type to obtain a target keyword set corresponding to the target document; a historical keyword set corresponding to historical documents matching the user identifier is obtained; similarity between the target document and the historical documents is calculated based on the target keyword set and the historical keyword set; a target association relationship is established between the target document and the historical documents based on the similarity; and document display data is generated based on the target association relationship. The document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on the historical images corresponding to the historical documents.

[0073] The document display data returned by the server is obtained, and the document is displayed on a document display interface based on the document display data.

[0074] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0075] Acquire a document collection instruction generated by a trigger operation on a document processing interface, perform image collection on a target document according to the document collection instruction, and obtain a document image corresponding to the target document;

[0076] The document image and the user identifier corresponding to the document image are sent to a server so that the server identifies a target document type corresponding to the document image; keyword recognition is performed on the document image based on the target document type to obtain a target keyword set corresponding to the target document; a historical keyword set corresponding to historical documents matching the user identifier is obtained; similarity between the target document and the historical documents is calculated based on the target keyword set and the historical keyword set; a target association relationship is established between the target document and the historical documents based on the similarity; and document display data is generated based on the target association relationship. The document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on the historical images corresponding to the historical documents.

[0077] The document display data returned by the server is obtained, and the document is displayed on a document display interface based on the document display data.

[0078] The document processing and document display method, apparatus, computer device, and storage medium described above obtain a document image corresponding to a target document and a user identifier corresponding to the document image, perform keyword recognition on the document image based on the target document type corresponding to the document image, obtain a target keyword set corresponding to the target document, obtain a historical keyword set corresponding to historical documents that match the user identifier, and calculate the similarity between the target document and the historical documents based on the target keyword set and the historical keyword set. A target association relationship is established between the target document and the historical documents based on the similarity, and the target association relationship is used for document display. In this way, the text information of the document can be obtained from the document image, thereby simultaneously storing the text information of documents corresponding to different document providers, breaking down information silos between different document providers and helping to improve information acquisition efficiency. Furthermore, the similarity between documents corresponding to different document providers is calculated, and an association relationship between documents corresponding to different document providers is established based on the similarity. Documents can be displayed based on the association relationship, so there is no need to access the databases corresponding to different document providers to view different documents separately, which effectively improves the efficiency of information acquisition. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 A diagram of an application environment of a document processing and document display method in one embodiment;

[0080] Figure 2 1 is a flow chart of a document processing method in one embodiment;

[0081] Figure 3is a schematic diagram of a medical record form in one embodiment;

[0082] Figure 4A Schematic diagram of a process for performing keyword recognition on a document image in one embodiment;

[0083] Figure 4B A schematic diagram of calculating similarity between documents in one embodiment;

[0084] Figure 5A This is a schematic diagram of a document display interface in one embodiment;

[0085] Figure 5B This is a schematic diagram of a document display interface in another embodiment;

[0086] Figure 6A A schematic diagram of a process for displaying document details in one embodiment;

[0087] Figure 6B This is a schematic diagram of a details display interface in one embodiment;

[0088] Figure 7A A schematic diagram of a process for displaying a key indicator change chart in a details display interface in one embodiment;

[0089] Figure 7B This is a schematic diagram of a detail display interface in another embodiment;

[0090] Figure 8 Schematic diagram of a flow chart of a document processing method in another embodiment;

[0091] Figure 9 Schematic diagram of a process for displaying a document in one embodiment;

[0092] Figure 10 A flowchart of a document processing interface in one embodiment;

[0093] Figure 11 1 is a flow chart of a medical record processing method according to an embodiment;

[0094] Figure 12 is a structural block diagram of a document processing device in one embodiment;

[0095] Figure 13 It is a structural block diagram of a document processing device in another embodiment;

[0096] Figure 14 This is a structural block diagram of a document display device in one embodiment;

[0097] Figure 15 is a diagram of the internal structure of a computer device in one embodiment;

[0098] Figure 16 FIG. 4 is a diagram showing the internal structure of a computer device in another embodiment. DETAILED DESCRIPTION

[0099] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0100] Artificial Intelligence (AI) refers to the theories, methods, techniques, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that seeks to understand the essence of intelligence and produce new intelligent machines that can respond in a manner similar to human intelligence. AI also involves studying the design principles and implementation methods of various intelligent machines, enabling them to possess the capabilities of perception, reasoning, and decision-making.

[0101] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0102] Computer vision (CV) is the science of making machines "see." Specifically, it involves using cameras and computers to replace the human eye in identifying and measuring objects, and then further processing them to produce images more suitable for human observation or transmission to instruments. As a scientific discipline, computer vision studies related theories and technologies, attempting to build artificial intelligence systems capable of extracting information from images or multidimensional data. Computer vision technologies typically include image processing, image recognition, image semantic understanding, image retrieval, optical character recognition (OCR), video processing, video semantic understanding, video content / behavior recognition, three-dimensional object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous localization and mapping, and common biometric recognition technologies such as facial recognition and fingerprint recognition.

[0103] Natural language processing (NLP) is a key area of ​​research in computer science and artificial intelligence. It studies the theories and methods that enable effective communication between humans and computers using natural language. Natural language processing (NLP) integrates linguistics, computer science, and mathematics. Therefore, research in this field involves natural language—the language we use in everyday life—and is closely linked to the study of linguistics. Natural language processing technologies typically include text processing, semantic understanding, machine translation, robotic question answering, and knowledge graphs.

[0104] Machine learning (ML) is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is at the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications span all areas of AI. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0105] The solutions provided in the embodiments of this application involve artificial intelligence computer vision technology, natural language processing, machine learning and other technologies, which are specifically illustrated by the following embodiments:

[0106] The document processing and document display methods provided in this application can be applied to Figure 1 In the application environment shown, the terminal 102 communicates with the server 104 via a network. The terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices. The server 104 can be implemented as an independent server or a server cluster consisting of multiple servers or a cloud server.

[0107] Both the terminal 102 and the server 104 can be used independently to execute the document processing and method provided in the embodiments of the present application.

[0108] For example, the terminal obtains a document image corresponding to the target document and a user identifier corresponding to the document image. The terminal identifies the target document type corresponding to the document image, performs keyword recognition on the document image based on the target document type, and obtains a target keyword set corresponding to the target document. The terminal obtains a historical keyword set corresponding to historical documents that match the user identifier. The document set consisting of the target document and historical documents includes documents corresponding to different document providers. The historical keyword set is obtained by performing keyword recognition on historical images corresponding to historical documents. The terminal calculates the similarity between the target document and the historical document based on the target keyword set and the historical keyword set, and establishes a target association relationship between the target document and the historical document based on the similarity. The target association relationship is used for document display.

[0109] The server obtains a document image corresponding to the target document, which carries the user ID. The server identifies the target document type corresponding to the document image and performs keyword recognition on the document image based on the target document type, obtaining a target keyword set corresponding to the target document. The server also obtains a historical keyword set corresponding to historical documents that match the user ID. The document set consisting of the target document and historical documents includes documents from different document providers. Based on the target keyword set and the historical keyword set, the server calculates the similarity between the target document and the historical documents. Based on the similarity, the server establishes a target association relationship between the target document and the historical documents. The target association relationship is used for document display on the terminal.

[0110] The terminal 102 and the server 104 can also be used in conjunction to execute the document processing method and the document display method provided in the embodiments of the present application.

[0111] For example, the server receives a document display request sent by a terminal, the document display request carrying a target document identifier and a user identifier, and obtains a target association relationship between the target document corresponding to the target document identifier and the historical document corresponding to the user identifier. Based on the target association relationship, the server generates document display data corresponding to the document display request and returns the document display data to the terminal, so that the terminal displays the document according to the document display data. The document set consisting of the target document and the historical documents includes documents corresponding to different document providers. The target association relationship is established by the server based on the similarity between the target document and the historical documents. The similarity is calculated by the server based on a target keyword set corresponding to the target document and a historical keyword set corresponding to the historical documents. The target keyword set or the historical keyword set is obtained by the server through keyword recognition of the current document image based on the current document type corresponding to the current document image. The current document image is either the document image corresponding to the target document or the document image corresponding to the historical document.

[0112] The terminal obtains a document collection instruction generated by a trigger operation on the document processing interface, performs image collection on the target document according to the document collection instruction, and obtains a document image corresponding to the target document. The terminal then sends the document image and the user identifier corresponding to the document image to the server, allowing the server to identify the target document type corresponding to the document image. Based on the target document type, the terminal performs keyword recognition on the document image to obtain a target keyword set corresponding to the target document. The terminal then obtains a historical keyword set corresponding to historical documents that match the user identifier. The document set consisting of the target document and historical documents includes documents corresponding to different document providers. Based on the target keyword set and the historical keyword set, the terminal calculates the similarity between the target document and the historical documents. Based on the similarity, a target association relationship is established between the target document and the historical documents. Document display data is generated based on the target association relationship. The terminal obtains the document display data returned by the server and displays the document on the document display interface based on the document display data.

[0113] In one embodiment, Figure 2 As shown, a document processing method is provided, which is applied to Figure 1 The computer device in the embodiment is used as an example for explanation. It is understandable that the computer device can be the terminal 102 or the server 104. In this embodiment, the document processing method includes the following steps:

[0114] Step S202: Acquire a document image corresponding to the target document and a user identifier corresponding to the document image.

[0115] A document is a data record document that summarizes, organizes, and comprehensively analyzes collected data, written according to specified formats and requirements. Documents can be electronic or paper. Paper documents can be printed or handwritten. Documents come in a variety of forms, and different application scenarios require different documents. For example, in healthcare, a document might be a medical record. In transportation, a document might be a cargo manifest or a courier delivery note. In finance, a document might be an invoice, etc.

[0116] A document image refers to an image corresponding to a document. A document image can be a photograph of an electronic or paper document, a scan of an electronic or paper document, or a screenshot of an electronic document. It is understood that the document image corresponding to the target document can be at least one image.

[0117] A user ID is an identifier used to uniquely identify a user. Specifically, it may include a string of at least one of letters, numbers, and symbols, such as a user account number or mobile phone number. The user ID corresponding to the document image is the user ID corresponding to the user who uploaded the document image. For example, the user ID may specifically be the user account of the current document processing application.

[0118] Specifically, a user can log in to a document processing application or a document processing webpage on a terminal and upload a document image corresponding to a target document on the document processing application or webpage. The computer device can then obtain the document image corresponding to the target document and use the user ID of the currently logged-in user as the user ID corresponding to the document image. The document image uploaded by the user can be obtained in various ways, for example, by photographing the paper document with a camera, scanning the paper document with a scanning device, or taking a screenshot of an electronic document. For example, a user can access the official account of Hospital A on a social networking application, search for their medical records on the official account, and take a screenshot of the electronic medical records displayed on the official account to obtain the document image. It will be understood that when the computer device is a terminal, the terminal obtains the document image and locally performs data processing steps such as keyword recognition, similarity calculation, and association establishment on the document image. When the computer device is a server, the server obtains the document image from the terminal and locally performs data processing steps such as keyword recognition, similarity calculation, and association establishment on the document image. The server may generate document display data according to the association relationship, and send the document display data to the terminal, so that the terminal displays the document according to the document display data.

[0119] Step S204 : identifying the target document type corresponding to the document image, performing keyword recognition on the document image based on the target document type, and obtaining a target keyword set corresponding to the target document.

[0120] The target document type refers to the document type corresponding to the target document. Different methods can be used for keyword recognition for different types of documents to improve keyword recognition efficiency. Documents can be classified according to their purpose. Taking medical records as an example, document types include inspection documents, test documents, laboratory test documents, report documents, etc. Taking cargo lists as an example, document types include daily necessities cargo lists, production supplies cargo lists, office supplies cargo lists, etc. Document types can also be further classified according to the document provider (document ownership organization) and purpose. Taking medical records as an example, document types include hospital A + test documents, hospital A + inspection documents, hospital B + report documents, etc. Taking cargo lists as an example, document types include company A + daily necessities cargo lists, company B + daily necessities cargo lists, etc. Document types can be further classified according to the document provider (document to which it belongs), department, and purpose. Taking medical records as an example, document types include hospital A + department 1 + test documents, hospital A + department 1 + examination documents, hospital B + department 1 + report documents, etc. Taking cargo lists as an example, document types include company A + department 1 + daily necessities cargo list, company B + department 1 + daily necessities cargo list, etc.

[0121] Keyword recognition refers to identifying keywords in text information within an image. The target keyword set refers to the keyword set corresponding to the target document. The keyword set corresponding to a document can include keywords from at least one dimension, and a dimension can have at least one keyword. For example, a keyword set can specifically include keywords corresponding to the time dimension, spatial dimension, indicator dimension, and feedback dimension. Keywords in the time dimension refer to the time when the document was generated. Keywords corresponding to the spatial dimension refer to the location where the document was generated, such as the document provider or department. Keywords in the indicator dimension refer to test indicators, such as height and weight in an inspection document, white blood cell count in an inspection document, quantity of goods in a cargo manifest, and volume of goods in a cargo manifest. Keywords in the feedback dimension refer to feedback information, such as the inspection and analysis results of a medical record or the cargo acceptance results in a cargo manifest.

[0122] Specifically, after the computer device acquires a document image, it can identify the target document type corresponding to the document image. For example, the document title generally reflects the document type. The computer device can identify the document title from the document image and determine the target document type corresponding to the target document based on the document title. Of course, the terminal can also prompt the user to enter document type information and obtain the document type entered or selected by the user as the target document type corresponding to the document image. To improve keyword recognition efficiency, the computer device can perform keyword recognition on the document image based on the target document type, extracting key text information from the image to obtain a target keyword set corresponding to the target document. For example, different document types are generally generated using different document templates. Therefore, the computer device can perform keyword recognition on the document image based on the target document template corresponding to the target document type. Different keyword recognition models can be trained for different document types, and the keyword recognition models are machine learning models. The computer device can also perform keyword recognition on the document image based on the target keyword recognition model corresponding to the target document type. The input data of the keyword recognition model is the document image, and the output data is the keyword set corresponding to the document.

[0123] Step S206: Obtain a historical keyword set corresponding to the historical document matching the user identifier. The document set consisting of the target document and the historical documents includes documents corresponding to different document providers. The historical keyword set is obtained by performing keyword recognition on the historical images corresponding to the historical documents.

[0124] The same user ID can correspond to multiple documents, and the multiple documents can be documents corresponding to different document providers. For example, the medical records corresponding to user A can include medical records issued by hospital A, medical records issued by hospital B, and medical records issued by hospital C. The cargo orders corresponding to user A can include cargo orders issued by supplier A, cargo orders issued by supplier B, and cargo orders issued by supplier C. Historical documents refer to documents obtained before the target document. The historical keyword set refers to the keyword set corresponding to the historical documents. The historical image refers to the document image corresponding to the historical document. It can be understood that, similar to the method of obtaining the target keyword set, keyword recognition of the document image corresponding to the historical document can obtain the historical keyword set corresponding to the historical document.

[0125] The document provider refers to the document-owning organization for a document. A document-owning organization stores and manages multiple documents belonging to it. For example, different hospitals independently manage their own medical records. Hospital A manages medical record a, which it provides, and medical record a belongs to hospital A. Different suppliers independently manage invoices for their own products. Supplier B manages invoice b for its own products, and invoice b belongs to supplier B. Different courier companies independently manage their own delivery orders. A document-owning organization can use electronic devices to collect and analyze data, obtain analysis results, and send these results to the user's terminal. For example, a hospital can use medical equipment to collect and analyze a user's vital signs data, obtain the user's medical record, and provide the generated medical record to the user. A supplier can use monitoring equipment to collect and analyze product production and supply data, obtain invoices, and provide the generated invoices to the user.

[0126] Specifically, the computer device can locally retrieve a set of historical keywords corresponding to historical documents that match the user identifier. It will be understood that the historical keyword set is generated in the same manner as the target keyword set, and can be generated based on document images corresponding to the historical documents. There can be at least one historical document, and the document set consisting of the target document and historical documents includes documents corresponding to different document providers. In other words, the document set consisting of the target document and historical documents can include documents corresponding to at least two document providers. In this way, by performing image analysis on the document images, the computer device can obtain and store keyword sets for documents corresponding to different document providers, breaking down information silos between different document providers. Later, if the user wishes to view documents corresponding to different document providers, the computer device can directly retrieve the corresponding keywords from the local database and display them, eliminating the need for the user to log in to the document systems corresponding to different document providers to view the corresponding documents separately. Furthermore, document data corresponding to different user identifiers can be stored separately locally, effectively protecting user privacy. Users can only view their own documents and cannot view documents of other users.

[0127] Step S208: Calculate the similarity between the target document and the historical documents based on the target keyword set and the historical keyword set.

[0128] Specifically, after the computer device obtains the target keyword set and the historical keyword set, the computer device may calculate the similarity between the target document and the historical documents based on the target keyword set and the historical keyword set. If there are multiple historical documents, the computer device may calculate the similarity between the target document and each historical document based on the target keyword set and the historical keyword set corresponding to each historical document.

[0129] The computer device can calculate the similarity between a target document and historical documents using a machine learning model. The computer device can input a target keyword set and a historical keyword set into a trained document similarity calculation model to obtain the similarity between the target document and the historical documents. Furthermore, to reduce the computational complexity of the machine learning model, the computer device can also obtain keywords in a target dimension from a keyword set as a data source for similarity calculation, input the keywords in the target dimension from the target keyword set and the historical keyword set into a trained document similarity calculation model, and obtain the similarity between the target document and the historical documents. For example, the computer device can obtain keywords in an indicator dimension from a keyword set corresponding to a medical record as a data source for similarity calculation, input the keywords in the indicator dimension from two medical records into a trained medical record similarity calculation model, and obtain the similarity between the two medical records. In one embodiment, the computer device can use natural language processing technology to characterize the keywords, abstract the keywords into feature vectors, and use the feature vectors corresponding to the two documents as input data for the document similarity calculation model. The output data of the document similarity calculation model is the similarity between the two documents.

[0130] The computer device may also calculate the similarity between the target document and the historical documents based on the target keyword set and the historical keyword set using a similarity calculation algorithm. The computer device may calculate the keyword differences between the target keywords and the historical keywords corresponding to the same dimension, and perform a weighted sum of the keyword differences corresponding to each dimension to obtain the similarity between the target document and the historical documents.

[0131] Step S210: establishing a target association relationship between the target document and the historical documents based on the similarity, and the target association relationship is used for document display.

[0132] The target association relationship refers to the association between the target document and historical documents. Associations are established based on the similarity between documents. There are at least two types of associations, with different associations corresponding to different similarities. Associations can represent the degree of association between documents. For example, associations can be strong or weak. A strong association indicates a strong association between documents and a high degree of similarity, while a weak association indicates a weak association between documents and a low degree of similarity. Associations can ultimately be used for document display.

[0133] Document display is used to showcase the specific content of documents. Documents with the same association can be displayed together in the same area, while documents with different associations can be displayed separately in different areas. Furthermore, documents with different associations can be displayed differently, using different display templates and displaying different dimensions of document information.

[0134] Specifically, the computer device can establish a target association relationship between the target document and the historical documents based on the similarity corresponding to the target document and the historical documents. For example, if the similarity is greater than or equal to the similarity threshold, a first association relationship (strong association) is established between the target document and the historical document. If the similarity is less than the similarity threshold, a second association relationship (weak association) is established between the target document and the historical document. The historical documents corresponding to the first association relationship and the second association relationship can be displayed in partitions. For example, each historical document corresponding to the first association relationship is displayed in area A, and each historical document corresponding to the second association relationship is displayed in area B. It can be understood that if there are multiple historical documents, the target association relationship between the target document and each historical document is established based on each similarity. In addition, the display time point of the target document and the historical document can be set as needed. It can be displayed automatically or according to a user-triggered instruction. For example, the user triggers the document display by clicking the document display control on the terminal. In addition, the target document and the historical documents can be displayed synchronously or asynchronously. For example, after identifying the target keyword set corresponding to the target document, the target document can be displayed based on the target keyword set first. Then, after establishing the target association relationship corresponding to the target document and the historical documents, the historical document can be displayed based on the target association relationship.

[0135] In one embodiment, the document provider refers to the document ownership agency corresponding to the document. The document ownership agency is used to collect data and analyze data through electronic equipment to obtain data analysis results, and send the data analysis results to the terminal corresponding to the user identifier. The data analysis results include user vital sign data corresponding to the user identifier.

[0136] Specifically, in a medical scenario, a document may be a medical record, which may be an examination document, a test document, a laboratory test document, a report document, etc. In this case, the document provider refers to the document ownership agency corresponding to the document. The document ownership agency is used to collect and analyze data through electronic equipment to obtain data analysis results, and send the data analysis results to the terminal corresponding to the user identifier. The data analysis results include the user's vital sign data corresponding to the user identifier. The electronic equipment may specifically be various medical devices for collecting and analyzing user vital sign data, such as physical diagnostic instruments (thermometers, blood pressure monitors, microscopes, audiometers, various physiological recorders, etc.), imaging instruments (X-ray machines, CT scans, magnetic resonance imaging, B-ultrasound, etc.), analytical instruments (various types of counters, biochemical and immunoassay instruments, etc.), and electrophysiological instruments (such as electrocardiographs, electroencephalographs, electromyograms, etc.). That is, the document provider refers to the hospital to which the medical record belongs. The hospital collects various vital signs data of the user through medical equipment, and performs data analysis to obtain data analysis results including the user's vital signs data. The data analysis results can be presented in the form of a medical record, and the hospital can send the medical record to the user terminal.

[0137] The overall technical solution is illustrated using a medical record as an example. A computer device obtains a medical record image uploaded by a user. This medical record image corresponds to the current medical record and carries the user's user ID. The computer device may store medical record information for historical medical records corresponding to the user ID. This medical record information may include the medical record image corresponding to the historical medical record, a historical keyword set, similarities with other medical records, and associations. The computer device identifies the target medical record type corresponding to the medical record image and, based on the target medical record type, performs keyword recognition on the medical record image corresponding to the current medical record to obtain a target keyword set corresponding to the current medical record. Based on the user ID, the computer device obtains a historical keyword set corresponding to the user's historical medical record. The historical keyword set and the target keyword set may include data such as the medical record date, hospital, department, examination items, various indicators, and examination analysis results. Based on the target keyword set and the historical keyword set, the computer device calculates the similarity between the current medical record and the historical medical record, and establishes a target association relationship between the current medical record and the historical medical record based on this similarity. The computer device can display the medical record based on the target association relationship.

[0138] In the document processing method described above, a document image corresponding to a target document is obtained, the document image carrying a user identifier. Keyword recognition is performed on the document image based on the target document type corresponding to the document image to obtain a target keyword set corresponding to the target document. A historical keyword set corresponding to historical documents matching the user identifier is obtained. The document set consisting of the target document and historical documents includes documents corresponding to different document providers. Similarity between the target document and the historical documents is calculated based on the target keyword set and the historical keyword set. A target association relationship is established between the target document and the historical documents based on the similarity. This target association relationship is used for document display. In this way, document text information can be obtained from the document image, thereby simultaneously storing text information from documents corresponding to different document providers. This breaks down information silos between different document providers and helps improve information acquisition efficiency. Furthermore, similarity between documents corresponding to different document providers is calculated, and an association relationship between documents corresponding to different document providers is established based on the similarity. Document display is based on this association relationship, eliminating the need to access the databases corresponding to different document providers to view different documents separately, effectively improving information acquisition efficiency.

[0139] In one embodiment, identifying the target document type corresponding to the document image includes:

[0140] Perform text recognition on a preset image area in the document image to obtain a target title corresponding to the target document; and determine the target document type based on the target title.

[0141] The target title refers to the document title corresponding to the target document. The document title is generally located on the first line of the document, or in the area with the largest font size. The preset image area can specifically be the area corresponding to the first line of the document image, or in the area with the largest font size.

[0142] Specifically, the computer device can perform text recognition on a predetermined image area within the document image to obtain a target title corresponding to the target document. The document title is typically used to indicate the document type, such as the institution to which the document belongs or the purpose of the document. Therefore, the computer device can determine the target document type corresponding to the document image based on the target title.

[0143] refer to Figure 3 The computer device performs text recognition on the header image area in the medical record image to obtain the target title corresponding to the target document. The target title is "A Hospital Test Report". According to the target title, it can be determined that the target document type is the test report of Hospital A.

[0144] In this embodiment, the target title corresponding to the target document can be obtained by performing text recognition on a preset image area in the document image, and the target document type corresponding to the target document can be quickly determined based on the target title of the target document.

[0145] In one embodiment, Figure 4A As shown, keyword recognition is performed on the document image based on the target document type to obtain a target keyword set corresponding to the target document, including:

[0146] Step S402 : performing image segmentation on the document image based on the target document template corresponding to the target document type to obtain a plurality of candidate image regions corresponding to the document image.

[0147] Step S404 : performing text extraction on each candidate image region to obtain text information corresponding to each candidate image region.

[0148] The target document template refers to the document template corresponding to the target document type. The document template is a template used to generate documents. It can be understood that documents of the same document type are generated based on the same document template, but the specific document data is different. The document template has a fixed and standardized structure layout, which is composed of multiple information units for recording different information. For example, refer to Figure 3 , Figure 3 The medical record shown is composed of a header for recording the user's attribute information, a body for recording the user's vital signs data, and a footer for recording auxiliary information (time, doctor).

[0149] Specifically, when performing keyword recognition, the computer device can perform image segmentation on the document image based on the target document template corresponding to the target document type, subdivide the document image into image areas that record different information, and obtain multiple candidate image areas corresponding to the document image. It can be understood that the document template specifies the position and size of each information unit, and each information unit is used to record document data of different dimensions. Therefore, based on the document template, the document image can be segmented to cut out the image area corresponding to each information unit. Furthermore, the computer device can perform text extraction on each candidate image area respectively, extract the text information in the image area, and thus obtain the text information corresponding to each candidate image area. For example, the computer device performs text extraction on the candidate image area through the OCR (Optical Character Recognition) recognition algorithm to obtain the text information corresponding to the candidate image area.

[0150] Step S406: perform keyword recognition on each text information to obtain target keywords corresponding to each candidate image region.

[0151] Step S408: obtaining a target keyword set based on each target keyword.

[0152] Specifically, the computer device can perform keyword recognition on the text information corresponding to each candidate image region to obtain target keywords corresponding to each candidate image region. The computer device can perform keyword recognition on the text information based on a keyword dictionary to obtain keywords in the text information. Specifically, the computer device can perform keyword recognition on each text information based on a comprehensive keyword dictionary to obtain target keywords corresponding to each candidate image region. To improve recognition accuracy, the computer device can also establish a keyword dictionary corresponding to each candidate image region. The computer device performs keyword recognition on the corresponding text information based on the keyword dictionary corresponding to each candidate image region to obtain target keywords corresponding to each candidate image region. Ultimately, each target keyword constitutes a target keyword set.

[0153] In this embodiment, the text recognition area can be refined through image segmentation, thereby improving the efficiency and accuracy of text extraction, and further improving the efficiency and accuracy of keyword recognition.

[0154] In one embodiment, each candidate image region has a corresponding keyword dictionary, and keyword recognition is performed on each text information to obtain target keywords corresponding to each candidate image region, including:

[0155] Obtain a current keyword dictionary corresponding to the current candidate image area; perform keyword recognition on text information corresponding to the current candidate image area based on the current keyword dictionary to obtain a target keyword corresponding to the current candidate image area.

[0156] Among them, the keyword dictionary is a dictionary manually established based on expert knowledge and experience, and the keyword dictionary includes multiple candidate keywords. The computer device can update the keyword dictionary regularly and continuously update the keyword dictionary based on the accumulation of document data to achieve more accurate keyword recognition results. When updating the dictionary, the computer device can count the recognition frequency corresponding to each candidate keyword in the keyword dictionary and filter out candidate keywords with a recognition frequency less than a preset frequency threshold. In addition, the computer device can also add new candidate keywords to the keyword dictionary. The new candidate keywords can be new words created and popularized with the development of technology and network dissemination. The current keyword dictionary refers to the keyword dictionary corresponding to the current candidate image area. It can be understood that different application scenarios correspond to different keyword dictionaries. For example, in the medical scenario, the keyword dictionary mainly includes medical-related terms, and in the financial scenario, the keyword dictionary mainly includes economic-related terms.

[0157] Specifically, different candidate image regions are typically used to record different types of text information, and different candidate image regions include keyword information of different dimensions. Therefore, to improve keyword recognition efficiency and accuracy, the computer device can establish corresponding keyword dictionaries specifically for different candidate image regions. Therefore, when performing keyword recognition, the computer device can obtain a current keyword dictionary corresponding to the current candidate image region and perform keyword recognition on the text information corresponding to the current candidate image region based on the current keyword dictionary to obtain a target keyword corresponding to the current candidate image region. It will be understood that if the current candidate image region is a spatial image region, the current keyword dictionary may record candidate keywords representing multiple spatial dimensions, and the target keywords ultimately identified are keywords corresponding to the spatial dimensions. For example, a medical record may represent a hospital or department, while a cargo list may represent a supplier or department. If the current candidate image region is an indicator image region, the current keyword dictionary may record keywords representing multiple indicator dimensions, i.e., key indicators. The target keywords ultimately identified are keywords corresponding to the indicator dimensions. For example, a medical record may represent various key test indicators, while a cargo list may represent various key product indicators. Furthermore, although document images of different document types may all include indicator image areas, the indicator image areas corresponding to different document types may correspond to different keyword dictionaries. For example, the keyword dictionary corresponding to the indicator image area of ​​an inspection document includes multiple key inspection indicators, and the keyword dictionary corresponding to the indicator image area of ​​a test document includes multiple key test indicators. If the current candidate image area is a time image area, the current keyword dictionary may record multiple keywords corresponding to the time dimension, and the target keyword ultimately identified is the keyword corresponding to the time dimension, for example, the data sampling time, data analysis result generation time, and data analysis result verification time corresponding to the medical record, and the data sampling time, data analysis result generation time, and data analysis result verification time corresponding to the cargo list, etc.

[0158] In this embodiment, different image regions use different keyword dictionaries for keyword recognition, which can improve the accuracy and efficiency of keyword recognition.

[0159] In one embodiment, the current candidate image region is an indicator image region, and the target keyword corresponding to the indicator image region is the target keyword corresponding to the indicator dimension. Keyword recognition is performed on the text information corresponding to the current candidate image region based on the current keyword dictionary to obtain the target keyword corresponding to the current candidate image region, including:

[0160] Match the text information corresponding to the indicator image area with the corresponding keyword dictionary to obtain the initial keyword corresponding to the indicator dimension; the initial keyword includes the key indicator type and the initial key indicator data, and the initial key indicator data includes the key indicator initial unit; obtain the key indicator reference unit corresponding to the key indicator type; in the initial keyword corresponding to the indicator dimension, replace the key indicator initial unit based on the key indicator reference unit to obtain the target keyword corresponding to the indicator dimension.

[0161] The indicator image area refers to the image area that records the text information corresponding to the indicator dimension. The computer device can determine the indicator image area in the document image according to the document template. For example, Figure 3 , Figure 3 The middle image area of ​​the medical record shown is the indicator image area. Accordingly, the target keyword identified from the indicator image area is the target keyword corresponding to the indicator dimension. Of course, the indicator image area can also be located at the beginning, end, left, or right of the document image; the specific location can be determined based on the document template.

[0162] Compared to keywords in other dimensions, the data composition of keywords corresponding to the indicator dimension in documents is more complex. Different types of indicator data correspond to different data units. Accurate data units are required for keywords corresponding to indicator dimensions to accurately calculate the similarity between documents, so ensuring the accuracy of data units is crucial. Keywords in documents typically consist of fields and field values. For example, the keyword "Name: A" contains "Name" and "A" as the field value. Similarly, keywords corresponding to the indicator dimension include key indicator types and initial key indicator data. Key indicator types represent fields, while key indicator data represent field values. Key indicator types refer to the specific field types of keywords corresponding to the indicator dimension. For example, in medical records, key indicator types may include total protein count, blood protein count, weight, blood pressure, and pulse. Key indicator data refer to the specific field values ​​of keywords corresponding to the indicator dimension. For example, in medical records, the indicator data corresponding to the weight indicator type is 50 kg, and the indicator data corresponding to the total protein count indicator type is 65.5 g / L.

[0163] Initial key indicator data refers to the key indicator data obtained by initial identification. Initial key indicator data includes the initial unit of the key indicator. The initial unit of the key indicator refers to the key indicator unit obtained by initial identification. The key indicator unit is usually composed of letters, numbers, and symbols. For example, the key indicator unit corresponding to the weight indicator type is kg. Since the key indicator unit occupies a relatively small area in the image area, letters, numbers, and symbols are easily confused, and there may be cases where the initial unit of the key indicator is not accurately identified. However, compared to the key indicator unit, the key indicator type is usually composed of Chinese characters. The area occupied by Chinese characters in the image area is larger than the area occupied by the unit, and compared to letters, numbers, and symbols, Chinese characters are not easily confused. Therefore, in order to improve the recognition accuracy of the keywords corresponding to the indicator dimension, the computer device can determine the corresponding key indicator unit according to the key indicator type, without worrying about whether the initial unit of the key indicator obtained by identification is accurate, and directly replace the initial unit of the key indicator. The key indicator reference unit refers to the accurate key indicator unit corresponding to the key indicator type.

[0164] Specifically, if the current candidate image region is an indicator image region, keyword recognition can be performed on the indicator image region to obtain a target keyword corresponding to the indicator dimension. When performing keyword recognition on the indicator image region, the computer device can first match the text information corresponding to the indicator image region with the corresponding keyword dictionary to obtain an initial keyword corresponding to the indicator dimension. The initial keyword includes a key indicator type and initial key indicator data, and the initial key indicator data includes an initial unit of the key indicator. It is understood that the identified initial unit of the key indicator may contain a certain degree of recognition error. In this case, the computer device can correct the initial unit of the key indicator based on the identified key indicator type. The computer device pre-stores the accurate key indicator units corresponding to each key indicator type, namely, the key indicator reference units corresponding to each key indicator type. After identifying the key indicator type, the computer device can obtain the key indicator reference unit corresponding to the key indicator type and, in the initial keyword corresponding to the indicator dimension, replace the initial unit of the key indicator with the accurate unit corresponding to the key indicator type based on the key indicator reference unit, thereby obtaining the target keyword corresponding to the indicator dimension.

[0165] For example, in the medical record, the indicator data corresponding to the total protein count indicator type is 65.5g / L. The recognition result obtained through keyword recognition is that the key indicator type is "total protein count", then the computer device can obtain the key indicator reference unit "g / L" corresponding to "total protein count", and directly use the key indicator reference unit "g / L" as the unit of the indicator data "65.5". At this time, even if the recognition result of the key indicator unit in the initial key indicator data (that is, the initial unit of the key indicator) is "kg / L", the recognition is wrong, because the computer device can automatically replace the recognition result of the key indicator unit with the key indicator reference unit "g / L", it has effectively avoided the situation where the wrong data unit is used for similarity calculation in the future, and to a certain extent improved the accuracy of the similarity calculation.

[0166] In one embodiment, the computer device can also pre-store the reference indicator data range corresponding to each key indicator type. The computer device compares the initial key indicator data and the reference indicator data range corresponding to the same key indicator type, and can distinguish normal indicators from abnormal indicators from the target keywords corresponding to the indicator dimension. For example, the reference indicator data range corresponding to the total protein count indicator type is 65-85g / L. If the initial key indicator data identified from the medical record image is 60g / L, through data comparison, the computer device can determine that the total protein count in the medical record is low, which is an abnormal indicator. Subsequently, when displaying the document, the abnormal indicator can be highlighted, for example, using eye-catching colors or larger fonts to highlight the abnormal indicator.

[0167] In this embodiment, in the initial keywords corresponding to the indicator dimension, the key indicator initial unit is replaced by the key indicator reference unit corresponding to the key indicator type, so that a more accurate target keyword corresponding to the indicator dimension can be obtained, thereby helping to improve the accuracy of the similarity technology.

[0168] In one embodiment, the target keyword set and the historical keyword set both include keywords of multiple dimensions. Calculating the similarity between the target document and the historical document based on the target keyword set and the historical keyword set includes:

[0169] In the target keyword set and the historical keyword set, the keyword differences are calculated based on the target keywords and historical keywords corresponding to the same dimension to obtain the keyword differences corresponding to each dimension; the target attention information corresponding to each dimension is obtained; and the similarity between the target document and the historical document is calculated based on the keyword differences and attention information corresponding to each dimension.

[0170] Attention information refers to attention weights. Keyword differences across different dimensions can correspond to different attention weights. Target attention information refers to accurate weights obtained through training and calculation. Computers can periodically update the attention weights corresponding to each dimension to achieve more accurate similarity calculations.

[0171] Specifically, both the target keyword set and the historical keyword set include keywords of multiple dimensions. When calculating the similarity between two documents, the computer device can calculate the keyword difference based on the target keywords and historical keywords corresponding to the same dimension, thereby obtaining the keyword difference corresponding to each dimension. Then, the computer device obtains the target attention information corresponding to each dimension, and performs weighted summation of the keyword differences corresponding to each dimension based on the target attention information corresponding to each dimension to obtain the similarity of the two documents. For example, the target keyword set includes time A1 and indicator B1, and the historical keyword set includes time A2 and indicator B2. The computer device can calculate the keyword difference △A corresponding to the time dimension based on time A1 and time A2, and calculate the keyword difference △B corresponding to the indicator dimension based on indicator B1 and indicator B2. The final similarity between the target document and the historical document is △A*a+△B*b. At this time, a represents the target attention information corresponding to the time dimension, and b represents the target attention information corresponding to the indicator dimension.

[0172] It is understood that if the keywords corresponding to the same dimension include keywords of multiple types, when calculating the keyword differences corresponding to a dimension, the computer device can calculate keyword sub-differences based on the target keywords and historical keywords corresponding to the same type, thereby obtaining keyword sub-differences corresponding to each type, and obtaining the keyword differences corresponding to the dimension based on each keyword sub-difference. For example, under the indicator dimension, the target keyword set includes indicator A1, indicator B1, and indicator C1, and the historical keyword set includes indicator A2, indicator B3, and indicator C3. Indicators A1 and A2 are keywords corresponding to indicator type A, indicators B1 and B2 are keywords corresponding to indicator type B, and indicators C1 and C2 are keywords corresponding to indicator type C. The computer device can calculate the keyword sub-difference △A corresponding to indicator type A based on indicators A1 and A2, the keyword difference △B corresponding to indicator type B based on indicators B1 and B2, and the keyword difference △C corresponding to indicator type C based on indicators C1 and C2. The keyword difference corresponding to the indicator dimension between the target document and the historical document is then △A*a+△B*b+△C*c, where a represents the attention information corresponding to indicator type A, b represents the attention information corresponding to indicator type B, and c represents the attention information corresponding to indicator type C. It will be understood that when calculating keyword sub-differences, keywords of the same type need to be converted into data with the same units to unify the calculation standard.

[0173] In one embodiment, using the keyword of the time dimension as an important data for calculating similarity can improve the accuracy of similarity calculation. Figure 4B The keyword set corresponding to a medical record can include examination time, key indicators, and examination subjects. The similarity between the target medical record and the historical medical record can be calculated based on the differences in examination time, key indicators, and examination subjects between the target and historical medical records. The similarity calculated based on examination subjects and key indicators may only indicate a weak correlation between the two records. However, if the examination times of the target and historical records are close, the time dimension information can increase the similarity between the target and historical records, ultimately making the correlation between the target and historical records strong. For example, a routine blood test document and a chromosome examination document may not have much correlation in terms of examination content and key indicators. However, if the examination times of the two documents are close together, a doctor can determine whether the patient has kidney or liver problems by combining the routine blood test and chromosome examination.

[0174] In this embodiment, based on the attention information corresponding to each dimension, a weighted calculation is performed on the keyword differences corresponding to each dimension, so that an accurate similarity between the target document and the historical document can be obtained.

[0175] In one embodiment, obtaining target attention information corresponding to each dimension includes:

[0176] Obtain the training keyword differences corresponding to the training document pairs in each dimension, and obtain the training similarities corresponding to the training document pairs; obtain the initial attention information corresponding to each dimension; calculate the initial similarities corresponding to the training document pairs based on the training keyword differences and the initial attention information corresponding to each dimension; calculate the similarity loss value based on the initial similarities corresponding to the training document pairs and the training similarities, and adjust the initial attention information corresponding to each dimension based on the similarity loss value until the convergence conditions are met, thereby obtaining the target attention information corresponding to each dimension.

[0177] A training document pair consists of a pair of training documents used to train attention information. Multiple training document pairs are possible. Initial attention information refers to the attention information to be trained for each dimension. Training keyword differences are calculated based on the keywords corresponding to the same dimension in the two training documents. Training similarity refers to the similarity between the training document pairs and can be obtained through expert comparative analysis of the training document pairs. Training similarity provides a relatively authoritative and accurate similarity result.

[0178] Specifically, the computer device can perform machine learning on the attention information based on the training data to obtain target attention information, and the training data specifically includes relevant data of the training document pair. The computer device can specifically obtain the training keyword differences corresponding to the training document pair in each dimension, obtain the initial attention information corresponding to each dimension, and perform weighted summation of the training keyword differences based on the initial attention information corresponding to each dimension to obtain the initial similarity corresponding to the training document pair. The computer device can calculate the similarity loss value based on the initial similarity corresponding to the training document pair and the training similarity, and adjust the initial attention information corresponding to each dimension based on the similarity loss value until the convergence condition is met, thereby obtaining the target attention information corresponding to each dimension. The convergence condition can specifically be that the similarity loss value is less than a preset loss value, the number of adjustments to the similarity loss value reaches a preset number, etc. In this way, through deep learning, the calculated initial similarity can gradually approach the training similarity, thereby obtaining relatively accurate target attention information. Subsequently, based on the initial attention information corresponding to each dimension, the target similarity between the target document and the historical document can be accurately calculated.

[0179] It can be understood that in the same dimension, the target attention information corresponding to each type can also be trained in a similar way.

[0180] In one embodiment, establishing a target association relationship between a target document and a historical document based on similarity includes:

[0181] Obtain association relationship mapping information, the association relationship mapping information including reference similarities corresponding to at least two candidate association relationships; different candidate association relationships correspond to different document display modes; match the similarity with the reference similarity in the association relationship mapping information, and determine a target association relationship from various candidate association relationships based on the matching result.

[0182] The association mapping information is used to determine the association between two documents based on their similarity. The association mapping information includes reference similarities corresponding to at least two candidate associations. Different candidate associations correspond to different reference similarities. The reference similarity can be a numerical similarity or a similarity range.

[0183] The document display mode refers to the unified display format for all historical documents belonging to the same candidate association relationship. Different candidate association relationships can correspond to different document display modes, meaning that historical documents corresponding to different candidate association relationships can adopt different display formats. For example, historical documents corresponding to candidate association relationship A may be displayed using display format 1, while historical documents corresponding to candidate association relationship B may be displayed using display format 2. Different document display modes may specifically differ in at least one of the following: data arrangement order, displayed data type, and display area.

[0184] Specifically, the computer device can obtain the association relationship mapping information and match the similarity corresponding to the target document and the historical document with the reference similarity in the association relationship mapping information. If the similarity corresponding to the target document and the historical document falls within a certain reference similarity, the candidate association relationship corresponding to the reference similarity is used as the target association relationship between the target document and the historical document, and the target association relationship between the target document and the historical document is established. For example, the association relationship mapping information includes a first association relationship and a second association relationship. The reference similarity corresponding to the first association relationship is that the similarity is greater than or equal to the similarity threshold, and the reference similarity corresponding to the second association relationship is that the similarity is less than the similarity threshold. Therefore, if the similarity corresponding to the target document and the historical document is greater than or equal to the similarity threshold, the target association relationship between the target document and the historical document is the first association relationship. If the similarity corresponding to the target document and the historical document is less than the similarity threshold, the target association relationship between the target document and the historical document is the second association relationship.

[0185] In this embodiment, the association relationship between the target document and the historical document can be quickly determined based on the association relationship mapping information.

[0186] In one embodiment, there are multiple historical documents, and the method further includes:

[0187] Obtain a document display instruction; the document display instruction carries a target document identifier corresponding to the target document; obtain target association relationships between the target document and each historical document according to the document display instruction; display historical documents corresponding to the same target association relationship in a centralized area, and display historical documents corresponding to different target association relationships in partitioned areas.

[0188] The document display instruction is used to trigger document display. The document display instruction can be automatically triggered after a user uploads a document image. In this case, the computer device can determine the target association relationship between the target document and the historical documents and automatically display the historical documents based on the target association relationship. The document display instruction can also be triggered by a user operating a document display control on the interface. In this case, the computer device can first determine the target association relationship between the target document and the historical documents and then, after the user triggers the document display control, display the historical documents based on the target association relationship.

[0189] Displaying historical documents corresponding to the same target relationship in a centralized area means that all historical documents corresponding to the same target relationship are displayed in the same area. Displaying historical documents corresponding to different target relationships in a partitioned area means that all historical documents corresponding to different target relationships are displayed in different areas.

[0190] Specifically, a computer device obtains a document display instruction, which carries a target document identifier corresponding to a target document. The document display instruction can be used to instruct the computer device to display historical documents corresponding to the target document. In response to the document display instruction, the computer device can obtain target association relationships between the target document and each historical document. Based on the target association relationships, each historical document corresponding to the same target association relationship is displayed in a centralized area, and historical documents corresponding to different target association relationships are displayed in partitioned areas. For example, the historical documents corresponding to the target document include Historical Document 1, Historical Document 2, and Historical Document 3. The target association relationships between the target document and Historical Document 1 and Historical Document 2 are both first association relationships. The target association relationship between the target document and Historical Document 3 is a second association relationship. The computer device can then display Historical Document 1 and Historical Document 2 in Area A, and Historical Document 3 in Area B. It will be appreciated that when displaying historical documents, the target document can be displayed simultaneously. For example, the computer device may display the target document, historical document 1, and historical document 2 in area A, and display the target document and historical document 3 in area B.

[0191] In one embodiment, the computer device can also filter the various historical documents corresponding to the same target association relationship, and only display a preset number of historical documents when displaying them in a centralized area. The computer device can specifically filter the various historical documents corresponding to the same target association relationship based on similarity to determine the historical documents that need to be displayed. For example, only the three historical documents with the highest similarity to the target document are displayed in area A, or only the historical documents with a similarity to the target document greater than a preset threshold are displayed in area A. Furthermore, when displaying historical documents, the computer device can determine the display order of the historical documents based on the similarity. For example, in area A, the historical documents with higher similarity to the target document will be displayed at the front.

[0192] It is understood that within the same centralized area, different types of historical records can be further categorized and displayed. For example, within area A, historical medical records for examination documents, inspection documents, laboratory tests, and reports can be further categorized and displayed.

[0193] In this embodiment, historical documents corresponding to the same target association relationship are displayed in a centralized area, and historical documents corresponding to different target association relationships are displayed in partitioned areas, which can facilitate user viewing and thus improve document information acquisition efficiency.

[0194] In one embodiment, the target association relationship includes a first association relationship and a second association relationship. The historical documents corresponding to the same target association relationship are displayed in a centralized area, and the historical documents corresponding to different target association relationships are displayed in partitions, including:

[0195] Based on the historical keyword set of the historical documents corresponding to the first association relationship, the historical documents corresponding to the first association relationship are displayed; based on the association keywords of the historical documents corresponding to the second association relationship, the historical documents corresponding to the second association relationship are displayed; the similarity of the historical documents corresponding to the first association relationship is greater than the similarity of the historical documents corresponding to the second association relationship, and the association keywords are obtained by keyword matching the target keyword set and the historical keyword set of the historical documents corresponding to the second association relationship.

[0196] Here, associated keywords refer to keywords of similar or identical types between the historical document and the target document. The computer device can perform keyword matching on the target keyword set corresponding to the target document and the historical keyword set of the historical document corresponding to the second association relationship, searching the historical keyword set for historical keywords of similar or identical types to the target keyword as associated keywords corresponding to the historical document. For example, if a historical document includes weight data A of user A, and the target document includes weight data B of user A, both weight data A and weight data B are used to describe weight and are keywords of the same type, but the specific weight values ​​may differ. In this case, the associated keyword corresponding to the historical document may be weight data A.

[0197] Specifically, if the similarity of the historical documents corresponding to the first association relationship is greater than the similarity of the historical documents corresponding to the second association relationship, the first association relationship can be considered to be a strong association, and the second association relationship can be considered to be a weak association. If the historical document and the target document are strongly associated, then it means that the similarity between the historical document and the target document is high, and the historical keyword set corresponding to the historical document and the target keyword set corresponding to the target document are relatively similar. For the target document, the reference value of most of the keywords in the historical keyword set is relatively high. Therefore, if the target association relationship corresponding to the historical document is the first association relationship, the computer device can display the historical document corresponding to the first association relationship based on the historical keyword set of the historical document corresponding to the first association relationship. That is, when the computer device displays strongly associated historical documents, it can display each historical keyword corresponding to the historical document as a label corresponding to the historical document. Reference Figure 5A Clicking the "Strong Association" tab takes the user to a centralized display area corresponding to the first association relationship. This area displays historical documents A, B, and C, which are strongly associated with the target document. Each historical document's corresponding sub-area displays the document ID and document tag. The document tags can be the historical keywords associated with the historical document.

[0198] If the historical document is weakly associated with the target document, it means that the similarity between the historical document and the target document is low, and the historical keyword set corresponding to the historical document and the target keyword set corresponding to the target document are not very similar. For the target document, the reference value of most of the keywords in the historical keyword set is relatively low. Therefore, if the target association relationship corresponding to the historical document is the second association relationship, the computer device can display the historical document corresponding to the second association relationship based on the associated keywords of the historical document corresponding to the second association relationship. That is, when displaying weakly associated historical documents, the computer device only displays the same or similar types of keywords between the historical keyword set and the target keyword set, and only displays the key information in the historical document that has reference value to the target document. For example, target document A and historical document B are weakly associated, target document A includes 10 target keywords, historical document B includes 8 keywords, and the same or similar types of keywords between target document A and historical document B only include blood sugar values. Then, when displaying historical document B, only the blood sugar value corresponding to historical document B can be displayed, and other information with low reference value in historical document B does not need to be displayed. Reference Figure 5B Clicking the "Weak Association" tab takes the user to a centralized display area corresponding to the second association relationship. This area displays historical documents F and G, which are weakly associated with the target document. Each historical document's corresponding display sub-area displays the document ID and document tag. In this case, the document tag can be the association keyword between the historical document and the target document. The association keywords between historical document F and the target document are tags F1 and F6, while the association keyword between historical document G and the target document is tag G3.

[0199] I understand. Figure 5A and Figure 5B The search box in is used to search for corresponding documents based on keywords.

[0200] In this embodiment, the historical documents corresponding to the first association relationship can be displayed based on the historical keyword set, and the historical documents corresponding to the second association relationship can be displayed based on the associated keywords. This display method can improve the effectiveness of the information display of the historical documents corresponding to the target documents, so that users can quickly obtain document information with higher reference value.

[0201] In one embodiment, Figure 6A As shown, the method further includes:

[0202] Step S602: Acquire a document detail display instruction generated by a trigger operation on the document display interface. The document detail display instruction carries a document identifier corresponding to the document to be displayed.

[0203] Step S604: Obtain a target display template corresponding to the document type of the document to be displayed.

[0204] Step S606: Fill the keyword set corresponding to the document to be displayed into the target display template to obtain a detail display interface corresponding to the document to be displayed.

[0205] Step S608: display the details display interface.

[0206] The document details display instruction is used to display the detailed document content of a particular document. This instruction can be generated based on a user triggering an action within a document's display sub-area. For example, if a user clicks anywhere within the display sub-area of ​​historical document A, a document details display instruction carrying the document identifier corresponding to historical document A is generated. The triggering action can be a touch operation, cursor operation, key operation, or voice operation.

[0207] The document to be displayed can be a target document or a historical document. The target display template is the document display template corresponding to the document type to be displayed. A document display template is a data visualization template for a document, used to display detailed document content. Data visualization templates can be pre-designed for different document types, allowing users to quickly access key information from the detailed display interface. For example, inspection documents use the inspection document display template, test documents use the test document display template, and laboratory documents use the laboratory document display template. Report documents use the report document display template.

[0208] Specifically, the computer device obtains a document detail display instruction generated by a user through a trigger operation on the document display interface. The document detail display instruction carries the document identifier corresponding to the document to be displayed. Based on the document identifier corresponding to the document to be displayed, the computer device can determine the document type corresponding to the document to be displayed, and then obtain the target display template corresponding to the document type of the document to be displayed. The target display template is provided with fill-in areas corresponding to various keywords. The computer device only needs to fill each keyword in the keyword set corresponding to the document to be displayed into the corresponding area in the target display template to obtain the detail display interface corresponding to the document to be displayed. Finally, the computer device displays the detail display interface to the user. It is understandable that the reference indicator data range corresponding to each key indicator type, as well as the data analysis results of whether each key indicator data is abnormal, can also be further displayed in the detail display interface.

[0209] refer to Figure 6B , Figure 6BThe following is a schematic diagram of the interface of the details display interface corresponding to the medical record in an embodiment. The details display interface displays the basic attribute information of the medical record, such as the hospital name, department, test content, doctor, and test time. The details display interface also displays the key indicators of the medical record, and the key indicators of the medical record can be further divided into normal key indicators and abnormal key indicators, and the normal key indicators and abnormal key indicators are displayed in different areas. Figure 6B In (a), the user clicks the "Normal Key Indicators" tab to enter the normal key indicator display area, refer to Figure 6B In (b), the user clicks the "Abnormal Key Indicators" tab to enter the abnormal key indicator display area.

[0210] In this embodiment, in addition to collectively displaying historical documents with the same target association relationship, the document details of a certain document can also be displayed separately, so that the user can obtain detailed information of a certain document.

[0211] In one embodiment, both the target keyword set and the historical keyword set include keywords corresponding to the indicator dimension and the time dimension. The keywords corresponding to the indicator dimension include key indicator types and key indicator data. Figure 7A As shown, the method further includes:

[0212] Step S702 : Match the keyword set corresponding to the document to be displayed with the keyword sets of other documents to obtain a common key indicator type.

[0213] Step S704 : generating a key indicator change chart corresponding to the common key indicator type based on the key indicator data corresponding to the common key indicator type and the keyword corresponding to the time dimension.

[0214] Step S706: Display a key indicator change chart in the details display interface.

[0215] The target keyword set and the historical keyword set both include keywords corresponding to the indicator dimension and the time dimension. Keywords corresponding to the indicator dimension include key indicator types and key indicator data. Shared key indicator types refer to key indicator types shared between the document to be displayed and other documents. For example, if both the target document and the historical documents include heart rate data, heart rate is a shared key indicator type for the target document and the historical documents, and the heart rate values ​​in the target document and the historical documents are key indicator data. It is understood that there can be at least one shared key indicator type. The key indicator data corresponding to the same key indicator type in different documents can be different.

[0216] A key indicator change chart is a chart that statistically analyzes key indicator data for the same key indicator type across multiple documents. It displays how key indicator data for the same key indicator type changes over time. Key indicator change charts can be displayed in various graphic formats, such as pie charts, line charts, and bar charts. If there are multiple shared key indicator types, you can generate separate key indicator change charts for each shared key indicator type.

[0217] Specifically, in the details display interface corresponding to the document to be displayed, in addition to displaying the document's own information, a data change chart can also be displayed in conjunction with information from other documents, allowing viewers of the document to quickly compare data. The computer device can match the keyword set corresponding to the document to be displayed with the keyword set of other documents to determine the key indicator types shared by the document to be displayed and the other documents. Furthermore, based on the key indicator data corresponding to the shared key indicator types and the time information corresponding to each key indicator data, the computer device can generate a chart showing the key indicator data corresponding to the shared key indicator types over time. Finally, the computer device displays the key indicator change chart in the details display interface. It is understood that the key indicator change chart can be displayed automatically, for example, automatically displaying each key indicator change chart upon a user entering the document details display interface corresponding to the document to be displayed. If there are multiple key indicator change charts, each key indicator change chart can be automatically displayed in a scrolling manner. The key indicator change chart can also be displayed based on the trigger operation performed by the user on the document details display interface. For example, after the user enters the document details display interface corresponding to the document to be displayed, he clicks on a key indicator to trigger the display of the key indicator change chart corresponding to the key indicator.

[0218] refer to Figure 7B By clicking "Basophil Percentage," users can trigger a bar chart showing changes in the corresponding key indicator, basophil percentage. The horizontal axis of the bar chart represents time, and the vertical axis represents percentage information. Doctors or patients can view this bar chart to see how the patient's basophil percentage has changed over the past several tests.

[0219] In this embodiment, a key indicator change chart is displayed in the detail display interface corresponding to the document to be displayed, which can facilitate users to quickly compare key indicator data corresponding to common key indicator types in multiple documents. In this way, users do not need to jump back and forth between the detail display interfaces corresponding to different documents.

[0220] In one embodiment, Figure 8 As shown, a document processing method is provided, which is applied to Figure 1 The following steps are used as an example to illustrate the server in the example:

[0221] Step S802: Acquire a document display request sent by the terminal; the document display request carries a target document identifier and a user identifier.

[0222] Step S804, obtaining a target association relationship between a target document corresponding to the target document identifier and a historical document corresponding to the user identifier; the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the target association relationship is established based on the similarity between the target document and the historical documents, and the similarity is calculated based on a target keyword set corresponding to the target document and a historical keyword set corresponding to the historical documents, and the target keyword set or the historical keyword set is obtained by performing keyword recognition on the current document image based on the current document type corresponding to the current document image, and the current document image is a document image corresponding to the target document or a document image corresponding to the historical document.

[0223] Step S806: Generate document display data corresponding to the document display request based on the target association relationship, and return the document display data to the terminal so that the terminal can display the document according to the document display data.

[0224] The document display request is used to request the server to return document display data, so that the terminal can display the document based on the document display data. The target document identifier refers to the document identifier corresponding to the target document. The document display data is used to render data and generate a document display interface. The document display data may include at least one of display data corresponding to the target document and display data of at least one historical document corresponding to the target document. For example, the document display data may include only display data of at least one historical document corresponding to the target document, or may include both display data of the target document and display data of at least one historical document corresponding to the target document.

[0225] Specifically, before receiving a document display request from a terminal, the server may obtain the current document image sent by the terminal, identify the current document type corresponding to the current document image, and perform keyword recognition on the current document image based on the current document type to obtain a current keyword set corresponding to the current document. It will be understood that the current document image can be a historical document or a target document. After obtaining the target keyword set corresponding to the target document, the server may calculate the similarity between the target document and the historical document based on the target keyword set corresponding to the target document and the historical keyword set corresponding to the historical document. Based on the calculated similarity, the server may establish a target association relationship between the target document and the historical document, and store the target association relationship between the target document and the historical document.

[0226] Upon receiving a document display request from a terminal, the server can, based on the target document identifier carried in the document display request, obtain the target association relationship between the target document corresponding to the target document identifier and the historical document corresponding to the user identifier, and generate document display data corresponding to the document display request based on the target association relationship. The document display data may specifically include information such as the document display template, keyword set, target association relationship, associated keywords, and key indicator change charts corresponding to the document to be displayed. The server then sends the document display data to the terminal, which then renders the document display data to generate a document display page, thereby displaying the document. Of course, the document display data can also be further rendered to generate a detailed display interface.

[0227] It is understood that the specific process of the server processing the document image data can refer to the methods described in the aforementioned relevant embodiments. The specific process of the terminal displaying the document can refer to the methods described in the aforementioned relevant embodiments, for example, displaying historical documents corresponding to the same target association in a centralized area, and displaying historical documents corresponding to different target associations in partitioned areas. This will not be further described here.

[0228] The above-described document processing method can extract the text information of a document from a document image, thereby simultaneously storing the text information of documents corresponding to different document providers. This breaks down the information silos between different document providers and helps improve information acquisition efficiency. Furthermore, the similarity between documents corresponding to different document providers is calculated, and based on this similarity, a relationship is established between the documents corresponding to different document providers. Based on this relationship, documents can be displayed, eliminating the need to access the databases corresponding to different document providers to view different documents separately, effectively improving information acquisition efficiency.

[0229] In one embodiment, Figure 9 As shown, a document display method is provided, which is applied to Figure 1 The following steps are used as an example to illustrate the terminal in the figure:

[0230] Step S902: Acquire a document collection instruction generated by a trigger operation on the document processing interface, perform image collection on the target document according to the document collection instruction, and obtain a document image corresponding to the target document.

[0231] The document processing interface refers to the interface used to trigger document processing. The document collection instruction refers to the instruction used to trigger document image collection. The triggering operation can specifically be a touch operation, cursor operation, key operation, or voice operation.

[0232] Specifically, a user can log in to a document processing application or webpage on a terminal to access the document processing interface. The terminal can monitor user trigger operations on the document processing interface and generate document collection instructions based on the trigger operations. If the terminal receives the document collection instruction, it captures an image of the target document according to the document collection instruction and then sends the captured document image and the corresponding user identifier to the server. The server then processes the document image to obtain document display data. The terminal then receives the document display data returned by the server and displays the document on the document display interface based on the document display data.

[0233] refer to Figure 10 The document processing interface displays a "Scan Your Report" control 1002 for triggering a document collection command. When the terminal detects a user clicking on "Scan Your Report" control 1002, the document collection interface is entered, and the terminal activates the camera to capture an image of the target document. When the terminal detects a user clicking on the capture control 1004, the terminal captures the target document with the camera, obtaining a document image. The capture area can be the report scanning area 1006 in the document collection interface.

[0234] Step S904: Send the document image and the user identifier corresponding to the document image to the server so that the server can identify the target document type corresponding to the document image, perform keyword recognition on the document image based on the target document type, obtain a target keyword set corresponding to the target document, obtain a historical keyword set corresponding to the historical documents matching the user identifier, calculate the similarity between the target document and the historical documents based on the target keyword set and the historical keyword set, establish a target association relationship between the target document and the historical documents based on the similarity, generate document display data based on the target association relationship, the document set composed of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on the historical images corresponding to the historical documents.

[0235] Specifically, after the terminal collects the document image corresponding to the target document, the terminal can send the document image and the user identifier corresponding to the document image to the server, so that the server can perform data processing on the document image corresponding to the target document, and then calculate the similarity and establish an association relationship between the target document and the historical documents, thereby obtaining the document display data based on the target association relationship between the target document and the historical documents.

[0236] It can be understood that the specific process of the server processing the document image and generating the document display data can refer to the methods described in the aforementioned relevant embodiments, and will not be repeated here.

[0237] In one embodiment, once the server obtains a target keyword set through keyword recognition, it can first generate document display data corresponding to the target document (i.e., target display data) based on the target keyword set and return the target display data to the terminal. After the terminal obtains the target display data, it first displays the target document on the document display interface based on the target display data. Subsequently, after the terminal obtains the target document and historical documents or the document display data corresponding to the historical documents, it then displays the document on the document display interface based on the document display data.

[0238] Step S906: Acquire the document display data returned by the server, and display the document on the document display interface based on the document display data.

[0239] Specifically, after the terminal obtains the document display data returned by the server, the terminal can render the document display data and display the document on the document display interface. The terminal can directly display the document on the document processing interface. At this time, the document processing interface is the document display interface. Of course, the terminal can also display the document display interface entrance on the document processing interface. If the terminal detects that the user triggers the document display interface entrance, it will jump to the document display interface and display the document on the document display interface. After obtaining the document display data, the terminal can also automatically jump from the document processing interface to the document display interface and display the document on the document display interface. On the document display interface, the target document can be displayed first, and then the historical document corresponding to the target document can be displayed. The target document and historical documents can also be displayed at the same time.

[0240] It can be understood that the specific method of displaying documents on the document display interface can refer to the methods described in the aforementioned relevant embodiments. For example, historical documents corresponding to the same target association relationship are displayed in a centralized area, and historical documents corresponding to different target association relationships are displayed in partitions. No further details will be given here.

[0241] The above-mentioned document display method can obtain the text information of the document from the document image, thereby simultaneously storing the text information of documents corresponding to different document providers, breaking down the information silos between different document providers and helping to improve information acquisition efficiency. Furthermore, the similarity between documents corresponding to different document providers is calculated, and based on this similarity, a relationship between documents corresponding to different document providers is established. Document display can be based on this relationship, eliminating the need to access the databases corresponding to different document providers to view different documents separately, effectively improving information acquisition efficiency.

[0242] This application also provides an application scenario, which applies the above-mentioned document processing and document display method. Specifically, the application of the document processing and document display method in this application scenario is as follows:

[0243] In healthcare scenarios, different hospitals typically maintain their own medical records independently, creating information silos between them. Traditionally, patients have been required to bring their medical records from various hospitals with them every time they visit, allowing doctors to review and understand their past medical conditions. However, paper medical records are easily lost or forgotten, while electronic medical records require access to multiple hospital systems, resulting in inefficient information acquisition.

[0244] This application can perform OCR and keyword recognition on medical record images provided by different hospitals to obtain the keyword set corresponding to each medical record. Based on the keyword set of the medical record, an association relationship is established between multiple medical records of the same patient, forming an association relationship between one medical record and multiple medical records. Subsequently, the medical records of different hospitals can be visualized on the patient's terminal, thereby improving the doctor's medical record review efficiency.

[0245] refer to Figure 11 , the data processing of medical records can specifically include the following steps:

[0246] 1. Create a keyword dictionary

[0247] Professionals can research and organize various types of medical records to build keyword dictionaries. There can be multiple keyword dictionaries, and keyword dictionaries can be updated regularly.

[0248] 2. Perform OCR recognition on the medical record image to obtain the text information of the medical record, and perform keyword recognition on the text information based on the keyword dictionary to obtain the keyword set of the medical record

[0249] 2-1. Upload medical record images

[0250] Patients can log in to the medical record processing application on the terminal and take photos of their previous medical records and upload them. The terminal can then send the captured medical record images to the server.

[0251] 2-2. OCR recognition

[0252] The server can perform OCR on the medical record image to obtain the text information of the medical record. Specifically, the server performs text recognition on a preset image area in the medical record image to obtain the corresponding medical record title, and then determines the corresponding medical record type based on the medical record title. The server then performs image segmentation on the medical record image based on the medical record template corresponding to the medical record type, obtaining multiple candidate image regions corresponding to the medical record image. The server then performs text extraction on each candidate image region to obtain the text information corresponding to each candidate image region.

[0253] 2-3. Keyword Identification

[0254] The server can perform keyword recognition on the text information based on the keyword dictionary to obtain a keyword set corresponding to the medical record. Specifically, each candidate image region has a corresponding keyword dictionary. The server can perform keyword recognition on the text information corresponding to the current candidate image region based on the current keyword dictionary corresponding to the current candidate image region to obtain the keyword corresponding to the current candidate image region. This further obtains the keywords corresponding to each candidate image region, and each keyword constitutes the keyword set corresponding to the medical record. The keyword set corresponding to the medical record may include information such as examination subject, examination time, examination indicators, and examination content.

[0255] 3. Calculate the similarity between medical records based on the keyword set, and establish the association relationship between medical records based on the similarity

[0256] After data processing, the server can obtain the keyword sets corresponding to each of the patient's previous medical records. Based on the keyword sets between the two medical records, the server can calculate the similarity between the two medical records. Furthermore, based on the similarity between the medical records, the server can establish an association relationship between the medical records.

[0257] If the similarity between two medical records is greater than or equal to the similarity threshold, a strong association is established between the two records. If the similarity between two medical records is less than the similarity threshold, a weak association is established between the two records. Whenever the server obtains a new medical record, it can establish an association relationship between the new medical record and each historical medical record based on the similarity.

[0258] 4. Display medical records based on the relationship between them

[0259] Based on the association relationship between medical records, the server can generate medical record display data for the historical medical record corresponding to the current medical record and send it to the terminal. The terminal can display the historical medical record based on the medical record display data. For example, historical medical records that are strongly associated with the current medical record can be centrally displayed in one area, and historical medical records that are weakly associated with the current medical record can be centrally displayed in another area. When performing centralized display, historical medical records can be displayed based on the medical record identifier and the medical record overview. For historical medical records that are strongly associated with the current medical record, the medical record overview can be the various keywords corresponding to the historical medical record. For historical medical records that are weakly associated with the current medical record, the medical record overview can be keywords in the historical medical record that are similar or identical to those in the current medical record.

[0260] In addition, when displaying multiple medical records at the same time, different types of medical records can be further classified and displayed, so that doctors can quickly identify and find the required medical records from a macro perspective. Different types of medical records can match different types of display methods. When displaying a medical record, the content and indicators in the medical record are displayed with corresponding data visualization, so that doctors can quickly view the indicator content contained in the medical record, thereby improving the efficiency of doctors' outpatient consultations. When displaying a medical record, corresponding reference values ​​and data units can be configured for different types of indicator content, so that doctors can quickly identify the required key indicators and abnormal indicators as a whole, as well as the corresponding reference range of each indicator. When displaying a medical record, an indicator change chart corresponding to a certain type of indicator content can also be generated, so that doctors can quickly compare the historical data and current data of a certain type of indicator content.

[0261] Furthermore, according to field research, in the traditional hospital consultation process, after completing the examination in the clinic, taking the examination documents to the department to see the doctor for consultation, the time required is as follows: ① In large hospitals, during peak hours, the waiting time is usually about 10-20 minutes; ② After entering the department, waiting for the doctor to record the patient's key indicators and understand their medical history and recent condition, which takes about 3-5 minutes; ③ The doctor gives a diagnosis and prescribes medication, which takes about 3-5 minutes. The entire process takes a total of 16-30 minutes.

[0262] Experiments have shown that after applying the technical solution of this application, patients can complete the photo scanning of medical records while waiting in line, which can be shortened by ① to 5-8 minutes (it is understandable that providing electronic medical records of different hospitals to the current hospital can avoid some repeated examinations); after entering the department, they can directly show the doctor their past medical history and recent physical condition information, which can be shortened by ② to 1-2 minutes; finally, the doctor makes a diagnosis and treatment judgment based on the visual results and gives targeted medication recommendations based on previous medication records, which can be shortened by ③ to 1-2 minutes. The whole process can be completed in 7-12 minutes, which improves efficiency by more than 50%, greatly shortening the patient's consultation and stay time in the hospital, greatly improving the utilization rate of medical resources, and greatly promoting the development of smart medical care.

[0263] It can be understood that the document processing and document display methods of the present application can also be applied to process documents in other fields, such as cargo lists and express delivery orders in the transportation field, invoices in the financial field, etc.

[0264] It should be understood that although Figure 2 、 4AThe steps in the flowcharts of 6A, 7A, 8, and 9 are shown in sequence 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 restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 2 、 4A At least part of the steps in 6A, 7A, 8, and 9 may include multiple steps or multiple stages. These steps or stages do not necessarily have to be performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily one by one, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0265] In one embodiment, Figure 12 As shown, a document processing device is provided. The device can be a software module or a hardware module, or a combination of the two to form a part of a computer device. The device specifically includes: an image acquisition module 1202, a keyword recognition module 1204, a historical information acquisition module 1206, a similarity calculation module 1208 and an association relationship establishment module 1210, wherein:

[0266] The image acquisition module 1202 is used to acquire a document image corresponding to the target document and a user identifier corresponding to the document image.

[0267] The keyword recognition module 1204 is used to identify the target document type corresponding to the document image, perform keyword recognition on the document image based on the target document type, and obtain a target keyword set corresponding to the target document.

[0268] The historical information acquisition module 1206 is used to acquire a historical keyword set corresponding to historical documents matching the user identifier. The document set consisting of the target document and historical documents includes documents corresponding to different document providers.

[0269] The similarity calculation module 1208 is configured to calculate the similarity between the target document and the historical documents based on the target keyword set and the historical keyword set.

[0270] The association relationship establishing module 1210 is used to establish a target association relationship between the target document and the historical documents based on similarity. The target association relationship is used for document display.

[0271] The document processing device, by acquiring document text information from document images, can simultaneously store the text information of documents corresponding to different document providers, breaking down information silos between different document providers and helping to improve information acquisition efficiency. Furthermore, the device calculates the similarity between documents corresponding to different document providers and establishes a relationship between them based on the similarity. Documents can then be displayed based on this relationship, eliminating the need to access the databases corresponding to different document providers to view each document separately, effectively improving information acquisition efficiency.

[0272] In one embodiment, the keyword identification module includes:

[0273] The document type recognition unit is used to perform text recognition on a preset image area in the document image, obtain a target title corresponding to the target document, and determine the target document type based on the target title.

[0274] In one embodiment, the keyword identification module includes:

[0275] The image segmentation unit is used to perform image segmentation on the document image based on the target document template corresponding to the target document type to obtain multiple candidate image regions corresponding to the document image.

[0276] The text extraction unit is used to extract text from each candidate image region to obtain text information corresponding to each candidate image region.

[0277] The keyword recognition unit is used to perform keyword recognition on each text information to obtain the target keyword corresponding to each candidate image area.

[0278] The keyword set determining unit is configured to obtain a target keyword set based on each target keyword.

[0279] In one embodiment, each candidate image area has a corresponding keyword dictionary, and the keyword recognition unit is further used to obtain the current keyword dictionary corresponding to the current candidate image area, perform keyword recognition on the text information corresponding to the current candidate image area based on the current keyword dictionary, and obtain the target keyword corresponding to the current candidate image area.

[0280] In one embodiment, the current candidate image region is an indicator image region, and the target keyword corresponding to the indicator image region is the target keyword corresponding to the indicator dimension. The keyword recognition unit is further configured to match the text information corresponding to the indicator image region with the corresponding keyword dictionary to obtain an initial keyword corresponding to the indicator dimension, the initial keyword including a key indicator type and initial key indicator data, the initial key indicator data including an initial unit of the key indicator, obtain a key indicator reference unit corresponding to the key indicator type, and replace the initial unit of the key indicator in the initial keyword corresponding to the indicator dimension based on the key indicator reference unit to obtain the target keyword corresponding to the indicator dimension.

[0281] In one embodiment, the target keyword set and the historical keyword set both include keywords of multiple dimensions. The similarity calculation module includes:

[0282] The keyword difference calculation unit is used to calculate keyword differences in the target keyword set and the historical keyword set based on the target keywords and historical keywords corresponding to the same dimension to obtain keyword differences corresponding to each dimension.

[0283] The attention information acquisition unit is used to obtain the target attention information corresponding to each dimension.

[0284] The similarity calculation unit is used to calculate the similarity between the target document and the historical document based on the keyword differences and attention information corresponding to each dimension.

[0285] In one embodiment, the attention information acquisition unit is also used to obtain the training keyword differences corresponding to the training document pairs in each dimension, obtain the training similarities corresponding to the training document pairs, obtain the initial attention information corresponding to each dimension, calculate the initial similarities corresponding to the training document pairs based on the training keyword differences and the initial attention information corresponding to each dimension, calculate the similarity loss value based on the initial similarities and training similarities corresponding to the training document pairs, adjust the initial attention information corresponding to each dimension based on the similarity loss value until the convergence conditions are met, and obtain the target attention information corresponding to each dimension.

[0286] In one embodiment, the association relationship establishment module is also used to obtain association relationship mapping information, which includes reference similarities corresponding to at least two candidate association relationships. Different candidate association relationships correspond to different document display modes. The similarity is matched with the reference similarity in the association relationship mapping information, and the target association relationship is determined from various candidate association relationships based on the matching results.

[0287] In one embodiment, there are multiple historical documents, and the document processing device further includes:

[0288] The document display module is used to obtain document display instructions. The document display instructions carry the target document identifier corresponding to the target document. According to the document display instructions, the target association relationship between the target document and each historical document is obtained. The historical documents corresponding to the same target association relationship are displayed in a centralized area, and the historical documents corresponding to different target association relationships are displayed in partitions.

[0289] In one embodiment, the target association relationship includes a first association relationship and a second association relationship. The document display module is further configured to display historical documents corresponding to the first association relationship based on a historical keyword set of the historical documents corresponding to the first association relationship, and to display historical documents corresponding to the second association relationship based on the association keywords of the historical documents corresponding to the second association relationship, wherein the similarity of the historical documents corresponding to the first association relationship is greater than the similarity of the historical documents corresponding to the second association relationship, and the association keywords are obtained by keyword matching the target keyword set and the historical keyword set of the historical documents corresponding to the second association relationship.

[0290] In one embodiment, the document display module is also used to obtain a document detail display instruction generated by a trigger operation on the document display interface. The document detail display instruction carries a document identifier corresponding to the document to be displayed, obtains a target display template corresponding to the document type to be displayed, fills the keyword set corresponding to the document to be displayed into the target display template, obtains a detail display interface corresponding to the document to be displayed, and displays the detail display interface.

[0291] In one embodiment, both the target keyword set and the historical keyword set include keywords corresponding to the indicator dimension and the time dimension, and the keywords corresponding to the indicator dimension include key indicator types and key indicator data. The document display module is further configured to match the keyword set corresponding to the document to be displayed with the keyword sets of other documents to obtain a shared key indicator type, generate a key indicator change chart corresponding to the shared key indicator type based on the key indicator data corresponding to the shared key indicator type and the keywords corresponding to the time dimension, and display the key indicator change chart in the details display interface.

[0292] In one embodiment, the document provider refers to the document ownership agency corresponding to the document. The document ownership agency is used to collect data and analyze data through electronic equipment to obtain data analysis results, and send the data analysis results to the terminal corresponding to the user identifier. The data analysis results include user vital sign data corresponding to the user identifier.

[0293] In one embodiment, Figure 13As shown, a document processing device is provided. The device can be a software module or a hardware module, or a combination of the two to form a part of a computer device. The device specifically includes: a request acquisition module 1302, an association relationship acquisition module 1304 and a display data determination module 1306, wherein:

[0294] The request acquisition module 1302 is used to acquire the document display request sent by the terminal; the document display request carries the target document identifier and the user identifier.

[0295] The association relationship acquisition module 1304 is used to obtain the target association relationship between the target document corresponding to the target document identifier and the historical document corresponding to the user identifier; the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the target association relationship is established based on the similarity between the target document and the historical documents, and the similarity is calculated based on the target keyword set corresponding to the target document and the historical keyword set corresponding to the historical documents. The target keyword set or the historical keyword set is obtained by keyword recognition of the current document image based on the current document type corresponding to the current document image. The current document image is the document image corresponding to the target document or the document image corresponding to the historical document.

[0296] The display data determination module 1306 is configured to generate document display data corresponding to the document display request based on the target association relationship, and return the document display data to the terminal so that the terminal performs document display according to the document display data.

[0297] In one embodiment, Figure 14 As shown, a document display device is provided. The device can be a software module or a hardware module, or a combination of the two to form a part of a computer device. The device specifically includes: an image acquisition module 1402, a request sending module 1404 and a document display module 1406, wherein:

[0298] The image acquisition module 1402 is used to obtain a document acquisition instruction generated by a trigger operation on the document processing interface, and to acquire an image of a target document according to the document acquisition instruction to obtain a document image corresponding to the target document;

[0299] Request sending module 1404 is configured to send the document image and the user identifier corresponding to the document image to the server, so that the server can identify the target document type corresponding to the document image, perform keyword recognition on the document image based on the target document type, obtain a target keyword set corresponding to the target document, obtain a historical keyword set corresponding to historical documents matching the user identifier, calculate the similarity between the target document and the historical document based on the target keyword set and the historical keyword set, establish a target association relationship between the target document and the historical document based on the similarity, and generate document display data based on the target association relationship. The document set consisting of the target document and the historical documents includes documents corresponding to different document providers. The historical keyword set is obtained by performing keyword recognition on the historical images corresponding to the historical documents.

[0300] The document display module 1406 is used to obtain the document display data returned by the server and display the document on the document display interface based on the document display data.

[0301] The document processing and display device described above, by acquiring document text information from document images, can simultaneously store the text information of documents corresponding to different document providers, breaking down information silos between different document providers and helping to improve information acquisition efficiency. Furthermore, similarities between documents corresponding to different document providers are calculated, and based on these similarities, associations between documents corresponding to different document providers are established. Documents can then be displayed based on these associations, eliminating the need to access the databases corresponding to different document providers to view different documents separately, effectively improving information acquisition efficiency.

[0302] The specific definitions of the document processing and document display apparatus can be found in the definitions of the document processing and document display methods above and will not be repeated here. Each module in the document processing apparatus described above may be implemented in whole or in part via software, hardware, or a combination thereof. Each of the modules described above may be embedded in or independent of a processor in a computer device in hardware form, or may be stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0303] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 15As shown. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as a historical keyword set corresponding to historical documents, a target keyword set corresponding to target documents, the similarity between the target document and the historical document, and the target association relationship. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a document processing and document display method is implemented.

[0304] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 16 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a document processing and document display method is implemented. The display screen of the computer device can be a liquid crystal display or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0305] Those skilled in the art will understand that Figure 15 、 16 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0306] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0307] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above-mentioned method embodiments when executed by a processor.

[0308] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of each of the above-described method embodiments.

[0309] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0310] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, 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.

[0311] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A document processing method, characterized in that: The method comprises: Obtaining a document image corresponding to a target document and a user identifier corresponding to the document image; Identifying a target document type corresponding to the document image, performing keyword recognition on the document image based on the target document type, and obtaining a target keyword set corresponding to the target document; Obtaining a historical keyword set corresponding to a historical document matching the user identifier, wherein the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on historical images corresponding to the historical documents; In the target keyword set and the historical keyword set, keyword differences are calculated based on the target keywords and historical keywords corresponding to the same dimension to obtain keyword differences corresponding to each dimension; The similarity between the target document and the historical document is calculated based on the keyword differences corresponding to the respective dimensions and the target attention information corresponding to the respective dimensions; the target attention information corresponding to the respective dimensions is obtained by calculating the initial similarity corresponding to the training document pair based on the training keyword differences corresponding to the respective dimensions and the initial attention information corresponding to the respective dimensions, calculating a similarity loss value based on the initial similarity corresponding to the training document pair and the training similarity, and adjusting the initial attention information corresponding to the respective dimensions based on the similarity loss value until a convergence condition is met; A target association relationship is established between the target document and the historical document based on the similarity, and the target association relationship is used for document display; different association relationships correspond to different document display modes.

2. The method according to claim 1, characterized in that The performing keyword recognition on the document image based on the target document type to obtain a target keyword set corresponding to the target document includes: performing image segmentation on the document image based on a target document template corresponding to the target document type to obtain a plurality of candidate image regions corresponding to the document image; Perform text extraction on each candidate image region to obtain text information corresponding to each candidate image region; Perform keyword recognition on each text information to obtain the target keyword corresponding to each candidate image area; The target keyword set is obtained based on each target keyword.

3. The method according to claim 2, characterized in that Each candidate image region has a corresponding keyword dictionary, and performing keyword recognition on each text information to obtain target keywords corresponding to each candidate image region includes: Get the current keyword dictionary corresponding to the current candidate image area; Keyword recognition is performed on text information corresponding to the current candidate image region based on the current keyword dictionary to obtain a target keyword corresponding to the current candidate image region.

4. The method according to claim 3, characterized in that The current candidate image area is an indicator image area, and the target keyword corresponding to the indicator image area is a target keyword corresponding to the indicator dimension; The performing keyword recognition on the text information corresponding to the current candidate image region based on the current keyword dictionary to obtain a target keyword corresponding to the current candidate image region includes: Matching the text information corresponding to the indicator image area with the corresponding keyword dictionary to obtain the initial keyword corresponding to the indicator dimension; the initial keyword includes the key indicator type and initial key indicator data, and the initial key indicator data includes the key indicator initial unit; Obtain the key indicator reference unit corresponding to the key indicator type; In the initial keywords corresponding to the indicator dimension, the key indicator initial unit is replaced based on the key indicator reference unit to obtain the target keyword corresponding to the indicator dimension.

5. The method according to claim 1, wherein There are multiple historical documents, and the method further includes: Obtaining a document display instruction; the document display instruction carries a target document identifier corresponding to the target document; Obtaining target association relationships between the target document and each historical document according to the document display instruction; The historical documents corresponding to the same target relationship are displayed in a centralized area, and the historical documents corresponding to different target relationships are displayed in partitions.

6. The method according to claim 5, characterized in that The target association relationship includes a first association relationship and a second association relationship; The display of historical documents corresponding to the same target association relationship in a centralized area and display of historical documents corresponding to different target association relationships in partitioned areas includes: Based on the historical keyword set of the historical documents corresponding to the first association relationship, display the historical documents corresponding to the first association relationship; Based on the associated keywords of the historical documents corresponding to the second association relationship, the historical documents corresponding to the second association relationship are displayed; the similarity of the historical documents corresponding to the first association relationship is greater than the similarity of the historical documents corresponding to the second association relationship, and the associated keywords are obtained by keyword matching the target keyword set and the historical keyword set of the historical documents corresponding to the second association relationship.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: Obtaining a document detail display instruction generated by a trigger operation on a document display interface, wherein the document detail display instruction carries a document identifier corresponding to the document to be displayed; Obtaining a target display template corresponding to the document type of the document to be displayed; Fill the keyword set corresponding to the document to be displayed into the target display template to obtain a detail display interface corresponding to the document to be displayed; The details display interface is displayed.

8. The method according to claim 7, characterized in that The target keyword set and the historical keyword set both include keywords corresponding to an indicator dimension and a time dimension, the keywords corresponding to the indicator dimension include a key indicator type and key indicator data, and the method further includes: Match the keyword set corresponding to the document to be displayed with the keyword set of other documents to obtain a common key indicator type; Based on the key indicator data corresponding to the common key indicator type and the keywords corresponding to the time dimension, a key indicator change chart corresponding to the common key indicator type is generated; The key indicator change chart is displayed in the details display interface.

9. The method according to claim 1, characterized in that The identifying the target document type corresponding to the document image includes: Performing text recognition on a preset image area in the document image to obtain a target title corresponding to the target document; The target document type is determined based on the target title.

10. The method according to claim 1, characterized in that The establishing of a target association relationship between the target document and the historical document based on the similarity includes: Acquire association relationship mapping information, wherein the association relationship mapping information includes reference similarities corresponding to at least two candidate association relationships; different candidate association relationships correspond to different document display modes; The similarity is matched with the reference similarity in the association relationship mapping information, and the target association relationship is determined from various candidate association relationships based on the matching result.

11. The method according to any one of claims 1 to 6, characterized in that The document provider refers to the document ownership agency corresponding to the document. The document ownership agency is used to collect data and analyze data through electronic equipment to obtain data analysis results, and send the data analysis results to the terminal corresponding to the user identifier. The data analysis results include the user vital sign data corresponding to the user identifier.

12. A document processing method, characterized in that: The method comprises: Obtaining a document display request sent by a terminal; the document display request carries a target document identifier and a user identifier; Obtain a target association relationship between a target document corresponding to the target document identifier and a historical document corresponding to the user identifier; the document set composed of the target document and the historical documents includes documents corresponding to different document providers, and the target association relationship is established based on the similarity between the target document and the historical documents, and the similarity is calculated based on the target keywords and historical keywords corresponding to the same dimension in the target keyword set corresponding to the target document and the historical keyword set corresponding to the historical document, and is calculated based on the keyword differences corresponding to each dimension and the target attention information corresponding to each dimension, and the target keyword set or the historical document is obtained. The historical keyword set is obtained by performing keyword recognition on the current document image based on the current document type corresponding to the current document image, where the current document image is the document image corresponding to the target document or the document image corresponding to the historical document; the target attention information corresponding to each dimension is obtained by calculating the initial similarity corresponding to the training document pair based on the training keyword difference corresponding to each dimension and the initial attention information corresponding to each dimension, calculating a similarity loss value based on the initial similarity corresponding to the training document pair and the training similarity, and adjusting the initial attention information corresponding to each dimension based on the similarity loss value until a convergence condition is met; Generating document display data corresponding to the document display request based on the target association relationship, and returning the document display data to the terminal, so that the terminal performs document display according to the document display data.

13. A document display method, characterized in that: The method comprises: Acquire a document collection instruction generated by a trigger operation on a document processing interface, perform image collection on a target document according to the document collection instruction, and obtain a document image corresponding to the target document; The document image and the user identification corresponding to the document image are sent to the server so that the server can identify the target document type corresponding to the document image, perform keyword recognition on the document image based on the target document type, obtain the target keyword set corresponding to the target document, obtain the historical keyword set corresponding to the historical documents matching the user identification, calculate the keyword difference based on the target keyword and historical keyword corresponding to the same dimension in the target keyword set and the historical keyword set, obtain the keyword difference corresponding to each dimension, calculate the similarity between the target document and the historical document based on the keyword difference corresponding to each dimension and the target attention information corresponding to each dimension, and establish the target document based on the similarity. The target association relationship between the target document and the historical document is used to generate document display data based on the target association relationship. The document set consisting of the target document and the historical document includes documents corresponding to different document providers. The historical keyword set is obtained by performing keyword recognition on historical images corresponding to the historical documents. The target attention information corresponding to each dimension is obtained by calculating the initial similarity corresponding to the training document pair based on the training keyword difference corresponding to each dimension and the initial attention information corresponding to each dimension, calculating the similarity loss value based on the initial similarity corresponding to the training document pair and the training similarity, and adjusting the initial attention information corresponding to each dimension based on the similarity loss value until a convergence condition is met. The document display data returned by the server is obtained, and the document is displayed on a document display interface based on the document display data.

14. A document processing device, characterized in that: The device comprises: An image acquisition module, configured to acquire a document image corresponding to a target document and a user identifier corresponding to the document image; a keyword recognition module, configured to identify a target document type corresponding to the document image, perform keyword recognition on the document image based on the target document type, and obtain a target keyword set corresponding to the target document; a historical information acquisition module, configured to acquire a historical keyword set corresponding to historical documents matching the user identifier, wherein the document set consisting of the target document and the historical documents includes documents corresponding to different document providers, and the historical keyword set is obtained by performing keyword recognition on historical images corresponding to the historical documents; a similarity calculation module for calculating keyword differences between the target keyword set and the historical keyword set based on the target keyword and the historical keyword corresponding to the same dimension, obtaining keyword differences corresponding to each dimension, and calculating the similarity between the target document and the historical document based on the keyword differences corresponding to each dimension and the target attention information corresponding to each dimension; the target attention information corresponding to each dimension is obtained by calculating the initial similarity corresponding to the training document pair based on the training keyword differences corresponding to each dimension and the initial attention information corresponding to each dimension, calculating a similarity loss value based on the initial similarity corresponding to the training document pair and the training similarity, and adjusting the initial attention information corresponding to each dimension based on the similarity loss value until a convergence condition is met; The association relationship establishing module is used to establish a target association relationship between the target document and the historical document based on the similarity, and the target association relationship is used for document display.

15. A document processing device, characterized in that: The device comprises: A request acquisition module, configured to acquire a document display request sent by a terminal; the document display request carries a target document identifier and a user identifier; An association relationship acquisition module is used to acquire a target association relationship between a target document corresponding to the target document identifier and a historical document corresponding to the user identifier; the document set composed of the target document and the historical document includes documents corresponding to different document providers, and the target association relationship is established based on the similarity between the target document and the historical document, and the similarity is calculated based on the target keyword and the historical keyword corresponding to the same dimension in the target keyword set corresponding to the target document and the historical keyword set corresponding to the historical document, and is calculated based on the keyword difference corresponding to each dimension and the target attention information corresponding to each dimension, and the target keyword The set or the historical keyword set is obtained by performing keyword recognition on the current document image based on the current document type corresponding to the current document image, and the current document image is the document image corresponding to the target document or the document image corresponding to the historical document; the target attention information corresponding to each dimension is obtained by calculating the initial similarity corresponding to the training document pair based on the training keyword difference corresponding to each dimension and the initial attention information corresponding to each dimension, calculating the similarity loss value based on the initial similarity corresponding to the training document pair and the training similarity, and adjusting the initial attention information corresponding to each dimension based on the similarity loss value until the convergence condition is met; A display data determination module is used to generate document display data corresponding to the document display request based on the target association relationship, and return the document display data to the terminal so that the terminal performs document display according to the document display data.

16. A document display device, characterized in that: The device comprises: An image acquisition module is used to obtain a document acquisition instruction generated by a trigger operation on the document processing interface, and to acquire an image of a target document according to the document acquisition instruction to obtain a document image corresponding to the target document; A request sending module is used to send the document image and the user identification corresponding to the document image to the server, so that the server can identify the target document type corresponding to the document image, perform keyword recognition on the document image based on the target document type, obtain the target keyword set corresponding to the target document, obtain the historical keyword set corresponding to the historical documents matching the user identification, calculate the keyword difference based on the target keyword and the historical keyword corresponding to the same dimension in the target keyword set and the historical keyword set, obtain the keyword difference corresponding to each dimension, calculate the similarity between the target document and the historical document based on the keyword difference corresponding to each dimension and the target attention information corresponding to each dimension, and establish a similarity based on the similarity. The target association relationship between the target document and the historical documents is generated based on the target association relationship. The document set consisting of the target document and the historical documents includes documents corresponding to different document providers. The historical keyword set is obtained by performing keyword recognition on historical images corresponding to the historical documents. The target attention information corresponding to each dimension is obtained by calculating the initial similarity corresponding to the training document pair based on the training keyword difference corresponding to each dimension and the initial attention information corresponding to each dimension. A similarity loss value is calculated based on the initial similarity corresponding to the training document pair and the training similarity. The initial attention information corresponding to each dimension is adjusted based on the similarity loss value until a convergence condition is met. The document display module is used to obtain the document display data returned by the server and display the document on the document display interface based on the document display data.

17. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 13 are implemented.

18. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 13 are implemented.

19. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 13 are implemented.

Images