Insurance policy information processing method and system, electronic device, and storage medium

Abstract

The present application relates to the field of artificial intelligence technology, and in particular to a method for processing insurance policy information and its system, electronic device, and storage medium. The insurance policy information processing method of the present application first obtains an insurance policy image. Then, the table elements of the insurance policy image are identified to obtain the table layout information in the insurance policy image. In addition, the text content of the insurance policy image is extracted to obtain the insurance text information. Further, based on a natural language recognition model, the insurance text information is semantically identified with reference to the table layout information to obtain insurance key fields. Finally, the target service options corresponding to the insurance key fields are matched from the candidate protection database. The insurance policy information processing method of the present application is based on a natural language recognition model, and the insurance text information is semantically identified with reference to the table layout information in the insurance policy image to obtain insurance key fields, and then the target service options corresponding to the insurance key fields are matched, thereby improving the efficiency of insurance policy information processing.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a policy information processing method and its system, electronic device, and storage medium. Background Art

[0002] Insurance policy is the abbreviation of insurance policy, which refers to the written proof of the insurance contract signed between the insurer and the policyholder. The insurance policy clearly and completely records the rights and obligations of the two parties to the insurance. Its contents mainly include the name of the insured, the name of the insured object and its storage place or status, the insurance amount, the insurance period, the insurance premium and other explanations of the two parties on matters related to the insurance object. Its contents may also include insurance clauses or other conditions agreed upon by both parties and additional conditions such as policy changes, transfers and cancellations.

[0003] After an insurance broker signs an insurance policy with an insured, he or she will be responsible for managing the policy content. As the number of insureds under the insurance broker's name increases, the number of policies that the insurance broker needs to manage also increases, so it is easy for errors and omissions to occur, resulting in low information processing efficiency. Therefore, how to improve the efficiency of policy information processing has become an urgent problem to be solved in the industry. Summary of the invention

[0004] The present application aims to solve at least one of the technical problems existing in the prior art. To this end, the present application proposes a policy information processing method and system, electronic device, and storage medium, which can improve the efficiency of policy information processing.

[0005] According to the first aspect of the present application, a method for processing insurance policy information includes:

[0006] Obtaining an insurance policy picture, wherein the insurance policy picture reflects the insurance information of the insured;

[0007] Performing table element recognition on the insurance policy picture to obtain table layout information in the insurance policy picture, wherein the table layout information reflects the shape and position of the table in the insurance policy picture;

[0008] Extracting text content from the insurance policy image to obtain insurance text information;

[0009] Based on the natural language recognition model, the insurance text information is semantically recognized with reference to the table layout information to obtain insurance key fields;

[0010] The target service options corresponding to the insurance key fields are obtained by matching from the candidate protection database.

[0011] According to some embodiments of the present application, the step of performing table element recognition on the insurance policy image to obtain table layout information in the insurance policy image includes:

[0012] Scanning the insurance policy picture horizontally to obtain horizontal layout information of horizontal straight lines in the insurance policy picture;

[0013] Scan the insurance policy picture vertically to obtain vertical layout information of vertical straight lines in the insurance policy picture;

[0014] The table layout information is obtained according to the horizontal layout information and the vertical layout information.

[0015] According to some embodiments of the present application, obtaining the table layout information according to the horizontal layout information and the vertical layout information includes:

[0016] Determining positions of respective intersection points of the horizontal straight line and the vertical straight line according to the horizontal layout information and the vertical layout information;

[0017] Based on the multiple eight-way connected regions corresponding to the intersections, determining the groups of intersections having adjacent positional relationships;

[0018] Connecting each group of intersections having adjacent positional relationships to determine the position and shape of each cell in the insurance policy image;

[0019] The table layout information is obtained according to the position and shape of each of the cells.

[0020] According to some embodiments of the present application, the semantic recognition of the insurance text information is performed based on the natural language recognition model with reference to the table layout information to obtain the insurance key fields, including:

[0021] Referring to the table layout information, performing a first text segmentation process on the insurance text information to obtain insurance text fields;

[0022] Vectorize the insurance text field to obtain an insurance text vector corresponding to the insurance text field;

[0023] Acquiring a preset normative vector based on a preset normative corpus, wherein the preset normative corpus is an insurance vocabulary corpus that complies with a general norm;

[0024] Matching the insurance text vector with the preset normative vector, and calculating the Euclidean distance between the insurance text vector and the preset normative vector;

[0025] When the Euclidean distance is less than a preset threshold, obtaining a preset standard field corresponding to the preset standard vector in the preset standard corpus;

[0026] The preset specification field is determined as the insurance key field.

[0027] According to some embodiments of the present application, obtaining a preset standard vector based on a preset standard corpus includes:

[0028] Performing a second text segmentation process on the preset standard corpus to obtain the preset standard fields;

[0029] The preset specification field is vectorized to obtain the preset specification vector corresponding to the preset specification field.

[0030] According to some embodiments of the present application, extracting text content from the insurance policy image to obtain insurance text information includes:

[0031] Binarizing the insurance policy image;

[0032] Generate a pixel distribution histogram based on the insurance policy image after binarization processing;

[0033] The text content of the insurance policy image is segmented based on the pixel distribution histogram, and the insurance text information is extracted.

[0034] According to some embodiments of the present application, before obtaining the target service option corresponding to the insurance key field from the candidate protection database through matching, the method further includes:

[0035] Acquire multiple types of preset insurance information, and configure corresponding preset service options for each type of the preset insurance information to form the candidate protection database.

[0036] According to the second aspect of the present application, the insurance policy information processing system includes:

[0037] The picture acquisition module is used to acquire the policy picture, which reflects various insurance information of the insured;

[0038] A text extraction module is used to extract text content from the insurance policy image to obtain insurance text information;

[0039] A table recognition module, used to recognize table elements on the policy image to obtain table layout information in the policy image, wherein the table layout information reflects the shape and position of each table in the policy image;

[0040] A semantic recognition module, for performing semantic recognition on the insurance text information based on a natural language recognition model and referring to the table layout information to obtain insurance key fields;

[0041] The service matching module is used to match the target service options corresponding to the insurance key fields from the candidate protection database.

[0042] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the insurance policy information processing method as described in any one of the embodiments of the first aspect of the present application.

[0043] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement a policy information processing method as described in any one of the embodiments of the first aspect of the present application.

[0044] The insurance policy information processing method and system, electronic device, and storage medium according to the embodiments of the present application have at least the following beneficial effects:

[0045] According to the insurance policy information processing method of the present application, the insurance policy picture is first obtained, wherein the insurance policy picture reflects the insurance information of the insured. Then, the table element recognition is performed on the insurance policy picture to obtain the table layout information in the insurance policy picture, wherein the table layout information reflects the shape and position of the table in the insurance policy picture. In addition, the text content of the insurance policy picture is extracted to obtain the insurance text information. Further, based on the natural language recognition model, the insurance text information is semantically recognized with reference to the table layout information to obtain the insurance key fields. Finally, the target service options corresponding to the insurance key fields are matched from the candidate protection database. The insurance policy information processing method of the present application is based on the natural language recognition model, and the insurance text information is semantically recognized with reference to the table layout information in the insurance policy picture to obtain the insurance key fields, and then the target service options corresponding to the insurance key fields are matched, thereby improving the efficiency of the insurance policy information processing.

[0046] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0048] Figure 1 A flowchart of a method for processing insurance policy information provided in an embodiment of the present application;

[0049] Figure 2 Another flowchart of the insurance policy information processing method provided in the embodiment of the present application;

[0050] Figure 3 Another flowchart of the insurance policy information processing method provided in the embodiment of the present application;

[0051] Figure 4Another flowchart of the insurance policy information processing method provided in the embodiment of the present application;

[0052] Figure 5 Another flowchart of the insurance policy information processing method provided in the embodiment of the present application;

[0053] Figure 6 Another flowchart of the insurance policy information processing method provided in the embodiment of the present application;

[0054] Figure 7 A schematic diagram of the insurance policy information processing system module provided in an embodiment of the present application;

[0055] Figure 8 Schematic diagram of an electronic device for executing the insurance policy information processing method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as limiting the present application.

[0057] In the description of this application, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed", etc. are understood to exclude the number itself, and "above", "below", "within", etc. are understood to include the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0058] In the description of the present application, it should be understood that descriptions involving orientation, such as up, down, left, right, front, back, etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0059] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0060] In the description of this application, it should be noted that, unless otherwise clearly defined, the terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in this application in combination with the specific content of the technical solution. In addition, the identification of specific steps below does not necessarily represent a limitation on the order of steps and execution logic. The execution order and execution logic between each step should be understood and inferred with reference to the contents described in the embodiment.

[0061] Artificial Intelligence (AI): It is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. AI is a branch of computer science. AI attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing and expert systems. AI can simulate the information process of human consciousness and thinking. AI is also the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. The basic technologies of AI generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. AI software technology mainly includes computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology and machine learning / deep learning.

[0062] The underwriting method provided in the embodiment of the present application relates to the field of artificial intelligence technology. The underwriting method provided in the embodiment of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the underwriting method, etc., but is not limited to the above forms.

[0063] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0064] It should be noted that in each specific implementation of the present application, when it comes to the need to perform relevant processing based on personal information, personal behavior data, personal historical data, personal location information, and other data related to personal identity or characteristics, the individual's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with the relevant laws, regulations, and standards of the relevant countries and regions. In addition, when the embodiment of the present application needs to obtain sensitive personal information of an individual, the individual's separate permission or separate consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the individual's separate permission or separate consent, the necessary personal-related data used to enable the normal operation of the embodiment of the present application will be obtained.

[0065] Insurance policy is the abbreviation of insurance policy, which refers to the written proof of the insurance contract signed between the insurer and the policyholder. The insurance policy clearly and completely records the rights and obligations of the two parties to the insurance. Its contents mainly include the name of the insured, the name of the insured object and its storage place or status, the insurance amount, the insurance period, the insurance premium and other explanations of the two parties on matters related to the insurance object. Its contents may also include insurance clauses or other conditions agreed upon by both parties and additional conditions such as policy changes, transfers and cancellations.

[0066] After an insurance broker signs an insurance policy with an insured, he or she will be responsible for managing the policy content. As the number of insureds under the insurance broker's name increases, the number of policies that the insurance broker needs to manage also increases, so it is easy for errors and omissions to occur, resulting in low information processing efficiency. Therefore, how to improve the efficiency of policy information processing has become an urgent problem to be solved in the industry.

[0067] The present application aims to solve at least one of the technical problems existing in the prior art. To this end, the present application proposes a policy information processing method and system, electronic device, and storage medium, which can improve the efficiency of policy information processing.

[0068] Further explanation is given below based on the accompanying drawings.

[0069] Reference Figure 1 According to the first aspect of the present application, the insurance policy information processing method includes:

[0070] Step S101, obtaining an insurance policy picture, which reflects the insurance information of the insured;

[0071] It should be noted that the policy picture refers to a picture that records the insurance information of the insured. In some embodiments of the present application, after the insured enters his or her insurance information into the policy, the policy picture that records the insurance information of the insured can be obtained immediately. It should be understood that there are various ways to obtain the insurance picture, which may include but are not limited to obtaining from a removable storage medium and obtaining through a network transmission method. In some more specific embodiments, after obtaining the policy picture, the insurance information of the insured in the policy picture can be archived according to the name of the insured or the identity identification number (Identity Document, ID) of the insured, and included in the database for unified management by adding information, deleting information, changing information or querying information.

[0072] Step S102, identifying table elements on the policy picture to obtain table layout information in the policy picture, where the table layout information reflects the shape and position of the table in the policy picture;

[0073] According to some embodiments provided by the present application, the insurance policy image contains a variety of insurance information, wherein the constituent elements of the insurance information may include, but are not limited to, tables, texts, color identifiers, and other types. In some more specific embodiments of the present application, since the table divides the insurance policy image into a number of cells, and each cell contains insurance information such as text, color identifiers, and icons that are independent of each other in content, the shape and position of each table in the insurance policy image is determined based on the table layout information, and the shape and position of the cells in each table can be further determined. Therefore, the insurance information is further identified and extracted based on the shape and position of the cells in each table, which means that the insured's insurance policy information can be processed more effectively and more specifically, thereby improving the efficiency of processing the insurance policy information.

[0074] Step S103, extracting text content from the policy image to obtain insurance text information;

[0075] It should be noted that extracting text content from the policy image refers to extracting information from the text content in the policy image to obtain the insurance text information. Information extraction refers to a text processing technology that extracts factual information such as entities, relationships, events, etc. of a specified type from natural language text and forms a structured data output. Information extraction is a technology that extracts specific information from text data. Text data is composed of some specific units, such as sentences, paragraphs, and chapters, and text information is composed of some small specific units, such as characters, words, phrases, sentences, paragraphs, or a combination of these specific units. Extracting noun phrases, names, place names, etc. from text data are all text information extraction, among which the information extracted by the text information extraction technology can also be various other types of information. It should be understood that the insurance text information refers to the insurance information of the text content in the policy image. Since the insurance policy image reflects the insurance information of the insured, the insurance text information is obtained by processing the text content of the insurance policy image, which can facilitate the subsequent semantic recognition of the insurance text information based on the natural language recognition model and reference to the table layout information to obtain the insurance key fields. It should be noted that the identification of specific steps in the embodiments of the present application does not necessarily represent a limitation on the order of steps and execution logic. For example, there is no necessary execution logic between step S102 and step S103. It should be understood that step S102 can be executed before step S103, step S102 can also be executed after step S103, and step S102 can also be executed simultaneously with step S103.

[0076] According to some embodiments provided by the present application, the policy image contains a variety of insurance information, wherein the components of the insurance information may include, but are not limited to, tables, texts, color identifiers, and other types. Extracting text content from the policy image to obtain the insurance text information refers to extracting the text content elements in the policy image. According to some embodiments provided by the present application, extracting text content from the policy image to obtain the insurance text information can be done in a variety of ways, including but not limited to extracting text content from the policy image through a neural network image recognition model, extracting text content from the policy image through an optical character recognition (OCR) model, and extracting text content from the policy image through an image projection method. Considering that in the process of extracting text content from the policy image to obtain the insurance text information, it is only necessary to map the image area in the policy image to the corresponding character, so there is no very strict requirement for recognition accuracy. Therefore, in order to save computing power resources, a more preferred embodiment of the present application uses an OCR model to extract text content from the policy image to obtain the insurance text information. It should be understood that extracting text content from the insurance policy image to obtain insurance text information may include, but is not limited to, the specific embodiments listed above.

[0077] Step S104, based on the natural language recognition model, semantic recognition is performed on the insurance text information with reference to the table layout information to obtain insurance key fields;

[0078] It should be noted that the insurance key field is the field in the insurance text information that can reflect the insurance needs of the insured. It should be understood that since the insurance information in the policy picture will vary depending on the different insurance needs of the insured, there will be many differences in the insurance information in multiple policy pictures. It is difficult to efficiently obtain the corresponding insurance key fields for multiple policy pictures by relying solely on manual processing. Therefore, according to some embodiments provided by the present application, based on a natural language recognition model, the insurance text information is semantically recognized with reference to the table layout information to obtain the insurance key field, which can improve the efficiency of processing the policy information. It should be noted that natural language processing (Natural Language Processing, NLP) refers to the use of computers to process, understand and use human languages ​​(such as Chinese, English, etc.). Natural language processing is a branch of artificial intelligence. It is an interdisciplinary subject of computer science and linguistics, and is often referred to as computational linguistics. Natural language processing includes grammatical analysis, semantic analysis, paragraph understanding, etc. Natural language processing is commonly used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis and opinion mining. It involves data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research and linguistic research related to language processing. The natural language recognition model in some embodiments of the present application is the artificial intelligence model used for natural language processing. In the related art, commonly used natural language recognition models include but are not limited to the Elmo model (Embeddings from Language Models, ELMo), the pre-trained generative model (Generative Pre-Training, GPT) and the Bert model and many other types of models. It should be emphasized that the key insurance fields are the fields in the insurance text information that can reflect the insurance needs of the insured, such as the policy number, name of the insured, effective date, payment period, payment method, beneficiary name, insurance type name, coverage, etc. Therefore, the key insurance fields in the policy picture will vary due to the different insurance needs of each insured, and there will be many differences in the key insurance fields in multiple policy pictures.

[0079] Step S105, obtaining target service options corresponding to the insurance key fields from the candidate protection database.

[0080] It should be noted that the candidate protection database refers to a service information database that is pre-set and matches various insurance needs. It should be understood that in the process of processing the policy information, it is necessary to match the target service options that are compatible with the insurance needs of the insured from the candidate protection database according to the respective insurance needs of the insured. Therefore, according to some embodiments provided by the present application, after obtaining the insurance key fields, it is necessary to first screen out the insurance mapping relationship that matches the insurance key fields from the candidate protection database, and then determine the respective insurance needs of the insured based on the insurance mapping relationship to obtain the target service options corresponding to the insurance key fields. It should be understood that the form of the insurance mapping relationship can be in the form of "preset key fields-preset service options", or in the form of "preset key fields-preset service methods-preset service options", or in other types of forms. When the content of any link in the insurance mapping relationship is obtained, the other contents in the corresponding insurance mapping relationship can be locked and obtained based on the link. It should be noted that the target service option is the service plan obtained according to the insurance needs of the insured during the policy information processing process. The target service option can output different target service options for different insurance information based on the insurance needs of the insured. For example, some insureds have different requirements for the payment period, so the candidate protection database includes a series of preset service options corresponding to the payment period, such as lump sum payment (paying all insurance premiums at one time), lifetime payment, payment by age, and annual payment. If it is identified that the insurance key field of a certain insured shows that the insured's insurance demand is "annual payment", then the target service option corresponding to "annual payment" can be matched from the candidate protection database to meet the customer's insurance needs. It should be emphasized that the target service options corresponding to the insurance key field matched from the candidate protection database may include, but are not limited to, the specific embodiments listed above.

[0081] According to the insurance policy information processing method of the present application, the insurance policy picture is first obtained, wherein the insurance policy picture reflects the insurance information of the insured. Then, the table element recognition is performed on the insurance policy picture to obtain the table layout information in the insurance policy picture, wherein the table layout information reflects the shape and position of the table in the insurance policy picture. In addition, the text content of the insurance policy picture is extracted to obtain the insurance text information. Further, based on the natural language recognition model, the insurance text information is semantically recognized with reference to the table layout information to obtain the insurance key fields. Finally, the target service options corresponding to the insurance key fields are matched from the candidate protection database. The insurance policy information processing method of the present application is based on the natural language recognition model, and the insurance text information is semantically recognized with reference to the table layout information in the insurance policy picture to obtain the insurance key fields, and then the target service options corresponding to the insurance key fields are matched, thereby improving the efficiency of the insurance policy information processing.

[0082] Reference Figure 2According to some embodiments of the present application, the table elements are identified on the insurance policy picture to obtain the table layout information in the insurance policy picture, including:

[0083] Step S201, horizontally scanning the insurance policy picture to obtain horizontal layout information of horizontal straight lines in the insurance policy picture;

[0084] It should be noted that the horizontal layout information is used to describe the horizontal distribution of the insurance information in the form of a table in the insurance policy image. It should be understood that the table style in the insurance policy image is usually horizontal and vertical. Therefore, according to some preferred embodiments provided by the present application, by horizontally scanning the insurance policy image, the horizontal layout information of the horizontal straight lines in the insurance policy image can be obtained, thereby obtaining the position and shape of the horizontal dimension of the table elements in the insurance policy image.

[0085] Step S202, vertically scanning the insurance policy image to obtain vertical layout information of vertical straight lines in the insurance policy image;

[0086] It should be noted that the vertical layout information is used to describe the vertical distribution of the insurance information in the form of a table in the insurance policy image. It should be understood that the table style in the insurance policy image is usually horizontal and vertical. Therefore, according to some preferred embodiments provided by the present application, by vertically scanning the insurance policy image, the vertical layout information of the vertical straight lines in the insurance policy image can be obtained, thereby obtaining the position and shape of the vertical dimension of the table elements in the insurance policy image.

[0087] Step S203, obtaining table layout information according to the horizontal layout information and the vertical layout information.

[0088] According to some embodiments provided by the present application, for table recognition, since the table style in the policy image is usually horizontal and vertical, the table layout information composed of the horizontal layout information and the vertical layout information in the policy image can be obtained by scanning the policy image horizontally and vertically, so the position and shape of each cell of each table in the policy image can be further identified to determine the table wireframe corresponding to the table layout information. It should be emphasized that the direction of horizontal scanning and vertical scanning is determined based on the orientation or position relationship shown in the policy image, and "horizontal" and "vertical" are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that obtaining the table layout information must have a specific orientation or be operated in a specific orientation.

[0089] Reference Figure 3 According to some embodiments of the present application, table layout information is obtained according to the horizontal layout information and the vertical layout information, including:

[0090] Step S301, determining the positions of each intersection point of a horizontal straight line and a vertical straight line according to the horizontal layout information and the vertical layout information;

[0091] According to some embodiments provided by the present application, for the recognition of tables, since the table style in the insurance policy image is usually horizontal and vertical, the table layout information composed of the horizontal layout information and the vertical layout information in the insurance policy image can be obtained by scanning the insurance policy image horizontally and vertically, so the position and shape of each cell of each table in the insurance policy image can be further identified to determine the table wireframe corresponding to the table layout information. According to some more specific embodiments of the present application, the table layout information is obtained according to the horizontal layout information and the vertical layout information. The positions of the intersections of the horizontal straight lines and the vertical straight lines in each table in the insurance policy image can be determined according to the horizontal layout information and the vertical layout information, and then the groups of intersections with adjacent positional relationships are determined based on the multiple eight-way connected areas corresponding to the intersections, and then the groups of intersections with adjacent positional relationships are connected to determine the positions and shapes of each cell in the insurance policy image, so that the table layout information can be obtained according to the positions and shapes of each cell.

[0092] Step S302, determining groups of intersections having adjacent positional relationships based on multiple eight-way connected regions corresponding to each intersection;

[0093] It should be noted that the eight-way connected area is also called the eight-way connected area, which refers to the area formed by starting from the positions of the intersections of the horizontal and vertical lines in the policy image and combining the positions of the pixels in eight directions of each intersection, where the eight directions refer to the eight directions of up, down, left, right, upper left, upper right, lower left, and lower right. It should be understood that based on the multiple eight-way connected areas corresponding to each intersection, each group of intersections with adjacent positional relationships can be determined among the intersections of the horizontal and vertical lines.

[0094] Step S303, connecting each group of intersection points having adjacent positional relationships to determine the position and shape of each cell in the insurance policy image;

[0095] According to some embodiments provided in the present application, based on the multiple eight-way connected areas corresponding to each intersection point, each group of intersection points with adjacent positional relationships among each intersection point of the horizontal straight line and the vertical straight line can be determined. Furthermore, each group of intersection points with adjacent positional relationships can be connected in sequence to obtain the minimum enclosing rectangle in each table, wherein the minimum enclosing rectangle is the cell that constitutes each table. Further, by repeatedly calling until convergence, it is possible to converge from the initial table outer frame to the final rectangular frame of each minimum unit, so that the position and shape of each cell in each table in the insurance policy image can be found.

[0096] Step S304, obtaining table layout information according to the position and shape of each cell.

[0097] According to some embodiments provided in the present application, after determining the position and shape of each cell in each table in the insurance policy image, the table layout information of each table in the insurance policy image can be obtained based on the position and shape of each cell.

[0098] Reference Figure 4 According to some embodiments of the present application, based on the natural language recognition model, the insurance text information is semantically recognized with reference to the table layout information to obtain the insurance key fields, including:

[0099] Step S401, referring to the table layout information, performing a first text segmentation process on the insurance text information to obtain the insurance text field;

[0100] According to some embodiments provided in the present application, in order to facilitate semantic recognition of insurance text information, it is necessary to perform a first text segmentation process on the insurance text information before performing semantic recognition based on a natural language recognition model to obtain insurance text fields, and then perform semantic recognition based on the insurance text fields to facilitate further text matching. It should be noted that the reason for performing the first text segmentation processing on the insurance text information with reference to the table layout information is that, according to some exemplary embodiments provided in the present application, the insurance policy image contains a variety of insurance information, wherein the constituent elements of the insurance information may include, but are not limited to, tables, texts, color identifiers and other types. Since the table divides the insurance policy image into several cells, and each cell contains insurance information such as text, color identifiers, icons, etc. that are independent of each other in content, the insurance text information is first segmented based on the table layout information to determine the shape and position of each table in the insurance policy image. Then, the shape and position of the cells in each table can be further determined. Therefore, the insurance information is further identified and extracted based on the shape and position of the cells in each table, which means that the insured's policy information can be processed more effectively and more specifically, thereby improving the efficiency of processing the policy information.

[0101] Step S402, vectorizing the insurance text field to obtain an insurance text vector corresponding to the insurance text field;

[0102] According to some embodiments provided by the present application, after the first text segmentation processing is performed on the insurance text information, some insurance text fields will be obtained, such as the policy number corresponding to the insured, the name of the insured, the effective date, the payment period, the payment method, the name of the beneficiary, the name of the insurance type, the coverage liability, and a series of insurance text fields. Further, the insurance text field is vectorized to obtain the insurance text vector corresponding to the insurance text field. It should be understood that the insurance text vector is the feature vector of the text type of the insurance text field, and the insurance text vector is used to compare with the preset standard vector corresponding to the preset standard corpus, so as to obtain the insurance key field. It should be emphasized that the insurance key field is the field in the insurance text information that can reflect the insurance needs of the insured.

[0103] Step S403, obtaining a preset normative vector based on a preset normative corpus, where the preset normative corpus is an insurance vocabulary corpus that complies with a general norm;

[0104] It should be noted that the preset standard corpus refers to an insurance vocabulary corpus that conforms to the general insurance field terminology standards, wherein the preset standard corpus can be a pre-organized text collection containing various insurance field terms, or a standard dictionary containing various insurance field terms, or other types of insurance field vocabulary corpus. Among them, the selection of the first national standard for the insurance industry, "Insurance Terminology (GB / T 36687-2018)", as the preset standard corpus is a more preferred embodiment of the present application. In some exemplary embodiments of the present application, the form of the preset standard vector corresponds to the form of the insured text vector, and based on the insured text vector obtained by vectorizing the insured text field, a matching query can be performed on the preset standard vector corresponding to the preset standard corpus.

[0105] It should be understood that the reason for obtaining the preset standard vector based on the preset standard corpus is to match the preset standard vector extracted from the preset standard corpus with the insurance text vector extracted from the insurance text field in the subsequent steps to obtain the insurance key field.

[0106] Step S404, matching the insurance text vector with a preset standard vector, and calculating the Euclidean distance between the insurance text vector and the preset standard vector;

[0107] Step S405, when the Euclidean distance is less than a preset threshold, obtaining a preset standard field corresponding to the preset standard vector in the preset standard corpus;

[0108] According to some embodiments provided in the present application, the policy number, insured name, effective date, payment period, payment method, beneficiary name, insurance type name, coverage liability and other types of words in the insurance field may represent completely different meanings with a difference of one character, and the insurance vocabulary specified in the preset standard corpus is a finite set. Therefore, in some more preferred embodiments, the insurance text information is more accurately matched with the content of the preset standard corpus. By matching the insurance text vector with the preset standard vector, the Euclidean distance between the insurance text vector and the preset standard vector is calculated. When the Euclidean distance between an insurance text vector and a preset standard vector is less than a preset threshold, the preset standard field corresponding to the preset standard vector in the preset standard corpus can be obtained, thereby determining the preset standard field as the insurance key field.

[0109] Step S406: determine the preset standard field as the insurance key field.

[0110] It should be noted that in some more specific embodiments of the present application, the insurance terms specified in the preset standard collection can be numbered in sequence, for example, policy number = 10, insurance type name = 11, policyholder name = 12, beneficiary name = 14, key content field = 15xx. It should be noted that the key content field is the field of key content in the insurance text information that can reflect the insured's insurance needs, such as the specific policy number, the insurance type name corresponding to the insured, and other information, and "xx" is a secondary number, representing the number of the subdivided type of the key content field. If all types of key content fields need to be included, the key content field can be represented by the number "1500". Therefore, the insurance text vector {insured person's name, policy number, 123456} is matched with the preset standard vector {insured person's name, policy number, key content field}. Under the action of the natural language recognition model to perform semantic recognition on "123456", it can be determined that "123456" is a key content field, and this key content field represents the specific policy number corresponding to the insured. Then it can be converted to calculate the Euclidean distance between {12, 10, 1510} and {12, 10, 1500} corresponding to the insurance text vector, where "1510" can represent the key content field of the specific policy number type. It should be understood that the provisions for the preset standard fields in the preset standard corpus can include, but are not limited to the specific embodiments listed above.

[0111] In some embodiments of the present application, when the Euclidean distance between the insured text vector and the preset normative vector is less than a preset threshold, it can be determined that the semantic difference between the insured text vector and the preset normative vector is small, and the preset normative field corresponding to the preset normative vector can be further determined as the insurance key field. If the insured text vector cannot match the preset normative vector whose Euclidean distance is less than the preset threshold, the insurance text field referred to by the insured text vector is determined as an unknown new word or a non-insurance word and is directly skipped. It should be understood that the reason why the preset normative vector is obtained based on the preset normative corpus, the preset normative vector extracted from the preset normative corpus is matched with the insured text vector extracted from the insurance text field to obtain the insurance key field is that in various types of insurance policy pictures, the insurance policy information filled in by multiple insured persons may have the problem of insufficient standardization in expression, thereby affecting the judgment of the insured person's insurance needs during the insurance policy information processing process, and further affecting the efficiency of the insurance policy information processing method. Therefore, through steps S401 to S406, the insurance text field can be further standardized and invalid fields can be eliminated based on the preset standard vector in the preset standard corpus, so as to obtain the insurance key field. The insurance text field of the insured's insurance information can be extracted from the insurance key field with higher efficiency based on the natural language recognition model, and finally the target service option corresponding to the insurance information can be matched from the candidate protection database.

[0112] According to some embodiments of the present application, before obtaining the target service option corresponding to the insurance key field from the candidate protection database through matching, the method further includes:

[0113] Acquire multiple types of preset insurance information, and configure corresponding preset service options for each type of preset insurance information to form a candidate protection database.

[0114] It should be noted that the preset insurance information is the insurance information pre-set based on the various insurance needs that the insured may have, and the preset service options refer to the service plans pre-set based on the various insurance needs that the insured may have. The preset service options can configure a variety of preset service options based on the various insurance needs of the insured. For example, some insureds have different requirements for the payment period, so the candidate protection database includes a series of preset service options corresponding to the payment period, such as lump sum payment (paying all insurance premiums at one time), lifetime payment, payment by age, and annual payment. If it is identified that the insurance key field of a certain insured shows that the insured's insurance demand is "annual payment", then the preset service options corresponding to "annual payment" can be matched from the candidate protection database to meet the user's insurance needs. For another example, according to various types of preset insurance information, preset service options such as "reminding users to renew before the policy expires", "reminding users on the day the benefits are issued that there are benefits to be collected", and even "sending birthday wishes on the user's birthday" can also be matched from the candidate protection database. It should be understood that the preset service options corresponding to each type of preset insurance information configuration may include, but are not limited to, the specific embodiments listed above.

[0115] Reference Figure 5 According to some embodiments of the present application, obtaining a preset normative vector based on a preset normative corpus includes:

[0116] Step S501, performing a second text segmentation process on the preset standard corpus to obtain preset standard fields;

[0117] Step S502, vectorize the preset standard field to obtain a preset standard vector corresponding to the preset standard field.

[0118] According to some embodiments provided by the present application, to obtain a preset specification vector based on a preset specification corpus, it is necessary to first perform a second text segmentation process on the preset specification corpus to obtain preset specification fields. After performing a second text segmentation process on the preset specification corpus, some preset specification fields will be obtained, such as the policy number, policyholder name, effective date, payment period, payment method, beneficiary name, insurance type name, and insurance liability that appear in the preset specification corpus, as well as the field types of the key content fields corresponding to the above-mentioned types of words. It should be noted that the key content field is the key content field in the insurance text information that can reflect the insurance needs of the policyholder, such as the specific policy number, the insurance type name corresponding to the policyholder, and other information. Further, by vectorizing the preset specification field, the preset specification vector corresponding to the preset specification field can be obtained. For example, if the policy number = 10, the insurance type name = 11, the insured person's name = 12, the beneficiary's name = 14, and the key content field = 15xx, then the preset specification vector {insured person's name, policy number, key content field} is {12, 10, 15xx}. It should be noted that "xx" is a secondary number representing the number of the subdivided type of the key content field. If all types of key content fields need to be included, the key content field can be represented by the number "1500". It should be understood that obtaining the preset specification vector based on the preset specification corpus is not limited to the specific embodiments listed above.

[0119] Reference Figure 6 According to some embodiments of the present application, text content is extracted from the policy image to obtain insurance text information, including:

[0120] Step S601, binarization processing is performed on the insurance policy image;

[0121] Step S602, generating a pixel distribution histogram based on the policy image after binarization processing;

[0122] Step S603, segmenting the text content of the insurance policy image based on the pixel distribution histogram, and extracting the insurance text information.

[0123] According to some embodiments provided by the present application, the text content of the insurance policy image is extracted to obtain the insurance text information in a variety of ways, including but not limited to extracting the text content of the insurance policy image through a neural network image recognition model, extracting the text content of the insurance policy image through an optical character recognition (OCR) model, and extracting the text content of the insurance policy image through an image projection method. In a more specific embodiment of the present application, the image projection method is selected to extract the text content of the insurance policy image to obtain the insurance text information. It should be noted that the image projection method, in principle, uses the distribution histogram of the pixels of the binary image to analyze, so as to find the dividing point of adjacent characters for segmentation, thereby extracting the insurance text information. It should be understood that when segmenting the insurance policy image, a suitable method should be selected. For example, if the layout of some insurance policy images is left-right type, it is suitable to use the vertical projection method, and if it is up-down type, it is suitable to use the horizontal projection method. If the characters in the image are criss-crossed, it is necessary to first perform vertical projection segmentation and then horizontal segmentation, or use the connected domain segmentation method to extract the character range. It should be understood that extracting text content from the insurance policy image to obtain insurance text information may include, but is not limited to, the specific embodiments listed above.

[0124] Reference Figure 7 According to the second aspect of the present application, the insurance policy information processing system 700 includes:

[0125] The picture acquisition module 701 is used to acquire the policy picture, which reflects the various insurance information of the insured;

[0126] The text extraction module 702 is used to extract text content from the insurance policy image to obtain insurance text information;

[0127] The table recognition module 703 is used to recognize table elements in the policy picture and obtain table layout information in the policy picture. The table layout information reflects the shape and position of each table in the policy picture.

[0128] Semantic recognition module 704, used to perform semantic recognition on the insurance text information based on the natural language recognition model and with reference to the table layout information to obtain insurance key fields;

[0129] The service matching module 705 is used to match the target service options corresponding to the insurance key fields from the candidate protection database.

[0130] According to the insurance policy information processing system 700 of the present application, the insurance policy information processing system 700 includes an image acquisition module 701, a text extraction module 702, a table recognition module 703, a semantic recognition module 704, and a service matching module 705. The insurance policy information processing system 700 of the present application first obtains the insurance policy image through the image acquisition module 701, wherein the insurance policy image reflects the insurance information of the insured. Then, the table element recognition of the insurance policy image is performed through the table recognition module 703 to obtain the table layout information in the insurance policy image, wherein the table layout information reflects the shape and position of the table in the insurance policy image. In addition, the text content of the insurance policy image is extracted by the text extraction module 702 to obtain the insurance text information. Further, based on the natural language recognition model of the semantic recognition module 704, the insurance text information is semantically recognized with reference to the table layout information to obtain the insurance key fields. Finally, the service matching module 705 is used to match the target service options corresponding to the insurance key fields from the candidate protection database. The insurance policy information processing system of the present application is based on the natural language recognition model of the semantic recognition module 704. It performs semantic recognition on the insurance text information with reference to the table layout information in the insurance policy image to obtain the insurance key fields, and then matches the target service options corresponding to the insurance key fields, thereby improving the efficiency of insurance policy information processing.

[0131] Figure 8 The electronic device 800 provided in the embodiment of the present application is shown. The electronic device 800 includes: a processor 801, a memory 802, and a computer program stored in the memory 802 and executable on the processor 801, and the computer program is used to execute the above-mentioned insurance policy information processing method when it is executed.

[0132] The processor 801 and the memory 802 may be connected via a bus or other means.

[0133] The memory 802 is a non-transitory computer-readable storage medium that can be used to store non-transitory software programs and non-transitory computer executable programs, such as the insurance policy information processing method described in the embodiment of the present application. The processor 801 implements the above-mentioned insurance policy information processing method by running the non-transitory software program and instructions stored in the memory 802.

[0134] The memory 802 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function. The data storage area may store and execute the above-mentioned insurance policy information processing method. In addition, the memory 802 may include a high-speed random access memory 802, and may also include a non-volatile memory 802, such as at least one storage device storage device, a flash memory device or other non-volatile solid-state storage device. In some embodiments, the memory 802 may optionally include a memory 802 remotely arranged relative to the processor 801, and these remote memories 802 may be connected to the electronic device 800 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network and combinations thereof.

[0135] The non-transient software programs and instructions required to implement the above-mentioned insurance policy information processing method are stored in the memory 802. When executed by one or more processors 801, the above-mentioned insurance policy information processing method is executed, for example, Figure 1 Steps S101 to S105 of the method, Figure 2 Steps S201 to S203 of the method, Figure 3 Steps S301 to S304 of the method, Figure 4 Steps S401 to S406 of the method, Figure 5 Steps S501 to S502 of the method, Figure 6 The method steps S601 to S603 in FIG.

[0136] The embodiment of the present application also provides a computer-readable storage medium storing computer-executable instructions, and the computer-executable instructions are used to execute the above-mentioned insurance policy information processing method.

[0137] In one embodiment, the computer readable storage medium stores computer executable instructions, which are executed by one or more control processors, for example, Figure 1 Steps S101 to S105 of the method, Figure 2 Steps S201 to S203 of the method, Figure 3 Steps S301 to S304 of the method, Figure 4 Steps S401 to S406 of the method, Figure 5 Steps S501 to S502 of the method, Figure 6 The method steps S601 to S603 in FIG.

[0138] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0139] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed method above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, storage device storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. In addition, it is well known to those skilled in the art that communication media generally include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium. It should also be understood that the various implementations provided in the embodiments of the present application can be arbitrarily combined to achieve different technical effects.

[0140] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above implementation mode. Technical personnel familiar with the field can also make various equivalent modifications or substitutions under the shared conditions without violating the spirit of the present application. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A method for processing insurance policy information, characterized in that: include: Obtaining an insurance policy picture, wherein the insurance policy picture reflects the insurance information of the insured; Scanning the insurance policy picture horizontally to obtain horizontal layout information of horizontal straight lines in the insurance policy picture; Scan the insurance policy picture vertically to obtain vertical layout information of vertical straight lines in the insurance policy picture; Determining positions of respective intersection points of the horizontal straight line and the vertical straight line according to the horizontal layout information and the vertical layout information; Based on the multiple eight-way connected regions corresponding to the intersections, determining the groups of intersections having adjacent positional relationships; Connecting each group of intersections having adjacent positional relationships to determine the position and shape of each cell in the insurance policy image; According to the position and shape of each of the cells, table layout information is obtained, wherein the table layout information reflects the shape and position of the table in the insurance policy image; Extracting text content from the insurance policy image to obtain insurance text information; Based on the natural language recognition model, referring to the table layout information, the first text segmentation processing is performed on the insurance text information to obtain the insurance text field; the insurance text field is vectorized to obtain the insurance text vector corresponding to the insurance text field; based on the preset standard corpus, a preset standard vector is obtained, and the preset standard corpus is an insurance vocabulary corpus that conforms to the general standard; the insurance text vector is matched with the preset standard vector, and the Euclidean distance between the insurance text vector and the preset standard vector is calculated; when the Euclidean distance is less than a preset threshold, the preset standard field corresponding to the preset standard vector in the preset standard corpus is obtained; and the preset standard field is determined as the insurance key field; The target service options corresponding to the insurance key fields are obtained by matching from the candidate protection database.

2. The method according to claim 1, characterized in that: The obtaining of a preset standard vector based on a preset standard corpus includes: Performing a second text segmentation process on the preset standard corpus to obtain the preset standard fields; The preset specification field is vectorized to obtain the preset specification vector corresponding to the preset specification field.

3. The method according to any one of claims 1 to 2, characterized in that: The extracting of text content from the insurance policy image to obtain insurance text information includes: Binarizing the insurance policy image; Generate a pixel distribution histogram based on the insurance policy image after binarization processing; The text content of the insurance policy image is segmented based on the pixel distribution histogram, and the insurance text information is extracted.

4. The method according to any one of claims 1 to 2, characterized in that: Before obtaining the target service option corresponding to the insurance key field from the candidate protection database through matching, the method further includes: Acquire multiple types of preset insurance information, and configure corresponding preset service options for each type of the preset insurance information to form the candidate protection database.

5. A policy information processing system, characterized in that: include: The picture acquisition module is used to acquire the policy picture, which reflects various insurance information of the insured; A text extraction module is used to extract text content from the insurance policy image to obtain insurance text information; a table recognition module, configured to scan the insurance policy image horizontally to obtain horizontal layout information of horizontal straight lines in the insurance policy image, scan the insurance policy image vertically to obtain vertical layout information of vertical straight lines in the insurance policy image, determine positions of each intersection of the horizontal straight lines and the vertical straight lines based on the horizontal layout information and the vertical layout information, determine groups of the intersections having adjacent positional relationships based on a plurality of eight-way connected regions corresponding to each of the intersections, connect the groups of the intersections having adjacent positional relationships, determine positions and shapes of each cell in the insurance policy image, and obtain the table layout information based on the positions and shapes of each cell, wherein the table layout information reflects the shapes and positions of each table in the insurance policy image; A semantic recognition module is used to perform a first text segmentation process on the insurance text information based on a natural language recognition model and with reference to the table layout information to obtain an insurance text field; vectorize the insurance text field to obtain an insurance text vector corresponding to the insurance text field; obtain a preset standard vector based on a preset standard corpus, wherein the preset standard corpus is an insurance vocabulary corpus that conforms to general standards; match the insurance text vector with the preset standard vector, and calculate the Euclidean distance between the insurance text vector and the preset standard vector; when the Euclidean distance is less than a preset threshold, obtain a preset standard field corresponding to the preset standard vector in the preset standard corpus; and determine the preset standard field as an insurance key field; The service matching module is used to match the target service options corresponding to the insurance key fields from the candidate protection database.

6. An electronic device, characterized in that: include: A memory and a processor, wherein the memory stores a computer program, and the processor implements the policy information processing method as described in any one of claims 1 to 4 when executing the computer program.

7. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement the insurance policy information processing method as described in any one of claims 1 to 4.

Images