KR102310450B1 - Computer program for providing a method to analysis insurance documents - Google Patents

Abstract

In one embodiment of the present disclosure for solving the above-described problems, a computer program stored in a computer-readable storage medium executable by one or more processors is disclosed. A computer program stored in the computer-readable storage medium, wherein the computer program, when executed by one or more processors, performs the following operations for analyzing a document for insurance subscription, the operations comprising: a chatbot interface of a user terminal Determining whether to start an insurance-related procedure based on the user's chat record received through the; receiving an insurance subscription document input through the chatbot interface from the user terminal; inputting the insurance subscription document into a document analysis model and analyzing the insurance subscription document using the document analysis model; and determining insurance subscription based on the analysis result of the insurance subscription document.

Description

COMPUTER PROGRAM FOR PROVIDING A METHOD TO ANALYSIS INSURANCE DOCUMENTS

The present disclosure relates to a computer program for providing information processing using a computer, and more particularly, to a computer program for providing an analysis method for a document.

In recent years, as the demand for insurance increases, the automation of insurance information and subscription processes using a chatbot in the terminal is underway in order to provide insurance efficiently. However, human intervention is still essential in the procedure and review for insurance. In particular, it can be said that it is the process of determining the degree of risk or loss that can occur to an individual, and the contract review (underwriting), which is the process of evaluating the expected mortality rate or the expected disease incidence rate, is a contract review task manager (underwriter, underwriter). The insurance company decides whether to accept the policyholder's (subscriber's) insurance subscription through the result of the contract review, classifies the insured's risk level, and can use it to adjust the limit of premiums or insurance money according to the degree of risk. are leaving

However, the contract review conducted by the person in charge of contract review or the mandatory review conducted by other experts will be postponed to the next business day following the review and approval of insurance subscriptions received on the same day as a holiday.

Accordingly, there may be a demand in the industry for a method of automating the insurance subscription process including insurance guidance and contract review through chatbots without time constraints.

Korean Patent Publication KR10-2019-0074356

The present disclosure has been made in response to the above-described background art, and is intended to provide a computer program for providing a document analysis method related to insurance subscription.

In one embodiment of the present disclosure for solving the above-described problems, a computer program stored in a computer-readable storage medium executable by one or more processors is disclosed. A computer program stored in the computer-readable storage medium, wherein the computer program, when executed by one or more processors, performs the following operations for analyzing a document for insurance subscription, the operations comprising: a chatbot interface of a user terminal Determining whether to start an insurance-related procedure based on the user's chat record received through the; receiving an insurance subscription document input through the chatbot interface from the user terminal; inputting the insurance subscription document into a document analysis model and analyzing the insurance subscription document using the document analysis model; and determining insurance subscription based on the analysis result of the insurance subscription document.

Alternatively, the insurance subscription document may include an image of a document related to insurance subscription.

Alternatively, the operation of receiving the insurance subscription document from the user terminal through the chatbot interface may include: transmitting insurance guide information to the user terminal through the chatbot interface provided to the user terminal; receiving a user confirmation signal for the insurance guide information of the user through a chatbot interface provided to the user terminal; and receiving the insurance subscription document based on a user's upload input for the insurance subscription document through a chatbot interface provided to the user terminal.

Alternatively, the operation of inputting the insurance subscription document into a document analysis model and analyzing the insurance subscription document using the document analysis model may include: one or more characters included in the insurance subscription document using the document analysis model; identifying the signature; and performing natural language processing on the one or more characters using the document analysis model to read the structure of the insurance subscription document.

Alternatively, the operation of reading the structure of the insurance subscription document by performing natural language processing on the one or more characters using the document analysis model may include: based on a result of natural language processing for the insurance subscription document, the insurance subscription document identifying one or more items included in the ; and reading the structure of the insurance subscription document based on a context analysis result for the one or more items.

Alternatively, the reading of the structure of the insurance subscription document based on the context analysis result for the one or more items may include reading the structure of the insurance subscription document by comparing the context analysis result with an existing insurance subscription document structure. movement; or reading the structure of the insurance subscription document by inputting the context analysis result into the structural analysis submodel; It may include at least one operation of

Alternatively, the structural analysis submodel is learned based on the analysis result of the imaged insurance subscription document, and uses a policy network to compare structural information included in the selected essential document with the structural information of the imaged insurance subscription document. to determine whether to select another essential document by processing it, calculate a reward based on the structural information contained in the other essential document and the structural information of the imaged insurance policy document, and select other essential documents that can maximize the reward. can be learned to choose.

Alternatively, the determining of the insurance subscription based on the analysis result of the insurance subscription document may include: recognizing at least one of a type or content of the insurance subscription document based on the analysis result of the insurance subscription document; and determining an insurance subscription based on at least one of a type or content of the insurance subscription document.

Alternatively, the determining of insurance subscription based on at least one of the type or content of the insurance subscription document may include generating an examination result for insurance subscription based on at least one of the type or content of the insurance subscription document may include.

Alternatively, generating an examination result for insurance subscription based on at least one of the type or content of the insurance subscription document is an operation of generating underwriting supplementary information based on a recognition result for each item included in the insurance subscription document may include.

Alternatively, generating the examination result for insurance subscription based on at least one of the type or content of the insurance subscription document is an operation of generating conditional approval information based on a recognition result for each item included in the insurance subscription document may include.

Alternatively, the determining of insurance subscription based on the analysis result of the insurance subscription document may include: generating document supplemental information for insurance subscription based on the analysis result of the insurance subscription document; and transmitting the supplementary information to the user terminal.

Alternatively, the generating of the document supplement information for insurance subscription based on the analysis result of the insurance subscription document may include: determining whether a missing item exists in the insurance subscription document; and determining whether an anomaly exists in an item included in the insurance subscription document.

A method for providing a document analysis method related to insurance subscription according to another embodiment of the present disclosure is disclosed. The method includes the steps of: determining whether to initiate an insurance-related procedure based on a chat record of a user received through a chatbot interface of a user terminal; receiving an insurance subscription document input through the chatbot interface from a user terminal; inputting the insurance subscription document into a document analysis model and analyzing the insurance subscription document using the document analysis model; and determining insurance subscription based on a result of analyzing the insurance subscription document.

A server for providing a document analysis method related to insurance subscription according to another embodiment of the present disclosure is disclosed. The server includes a processor including one or more cores; Memory; and a network unit for transmitting and receiving data to and from the user terminal, wherein the processor determines whether to initiate an insurance-related procedure based on the user's chat record received through the chatbot interface of the user terminal, and receives the chatbot interface from the user terminal receiving an insurance subscription document input through You can decide to purchase insurance.

The present disclosure may provide a computer program for providing a method related to insurance subscription more efficiently using a document analysis model.

Various aspects are now described with reference to the drawings, in which like reference numbers are used to refer to like elements collectively. In the following examples, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of one or more aspects. It will be evident, however, that such aspect(s) may be practiced without these specific details.
1 shows a conceptual diagram illustrating a system in which various aspects of a server for providing a document analysis method related to insurance subscription of the present disclosure can be implemented.
2 is a block diagram of a server for providing a document analysis method related to insurance subscription according to an embodiment of the present disclosure.
3 is a schematic diagram illustrating a network function underlying a structure analysis model for determining the structure of a document according to an embodiment of the present disclosure.
4 is a diagram exemplarily illustrating a document analysis method related to insurance subscription according to an embodiment of the present disclosure.
5 is a flowchart illustrating a method of analyzing a document related to insurance subscription according to an embodiment of the present disclosure.
6 illustrates logic for implementing a document analysis method related to insurance subscription according to an embodiment of the present disclosure.
7 depicts a simplified, general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure may be implemented.

Various embodiments are now described with reference to the drawings. In this specification, various descriptions are presented to provide an understanding of the present disclosure. However, it is apparent that these embodiments may be practiced without these specific descriptions.

As used herein, the terms “component,” “model,” “system,” and the like refer to a computer-related entity, hardware, firmware, software, a combination of software and hardware, or execution of software. For example, a component can be, but is not limited to being, a process running on a processor, a processor, an object, a thread of execution, a program, and/or a computer. For example, both an application running on a computing device and the computing device may be a component. One or more components may reside within a processor and/or thread of execution. A component may be localized within one computer. A component may be distributed between two or more computers. In addition, these components can execute from various computer readable media having various data structures stored therein. Components may communicate via a network such as the Internet with another system, for example via a signal having one or more data packets (eg, data and/or signals from one component interacting with another component in a local system, distributed system, etc.) may communicate via local and/or remote processes depending on the data being transmitted).

In addition, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or.” That is, unless otherwise specified or clear from context, "X employs A or B" is intended to mean one of the natural implicit substitutions. That is, X employs A; X employs B; or when X employs both A and B, "X employs A or B" may apply to either of these cases. It should also be understood that the term “and/or” as used herein refers to and includes all possible combinations of one or more of the listed related items.

Also, the terms "comprises" and/or "comprising" should be understood to mean that the feature and/or element in question is present. However, it should be understood that the terms "comprises" and/or "comprising" do not exclude the presence or addition of one or more other features, elements and/or groups thereof. Also, unless otherwise specified or unless the context is clear as to designating a singular form, the singular in the specification and claims should generally be construed to mean "one or more."

Those skilled in the art will further appreciate that the various illustrative logical blocks, configurations, models, circuits, means, logics, and algorithm steps described in connection with the embodiments disclosed herein may be implemented in electronic hardware, computer software, or combinations of both. It should be recognized that they can be implemented with To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, configurations, means, logics, models, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application. However, such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

The scope of the actions in the claims of the present disclosure is generated by the functions and features described in the respective actions, and unless a precedence of the order in each action is specified, the claims are The order of description of each operation is not affected. For example, in a claim described as an action including action A and action B, even if action A is described before action B, the scope of rights is not limited to that action A must precede action B.

Network functions and neural networks may often be used interchangeably in this disclosure.

1 shows a conceptual diagram illustrating a system in which various aspects of a server for providing a document analysis method related to insurance subscription of the present disclosure can be implemented.

A system according to embodiments of the present disclosure may include a document analysis server 100 , a user terminal 10 , an external server 20 , and a network. The document analysis server 100 , the user terminal 10 and the external server 20 according to embodiments of the present disclosure may mutually transmit and receive data for the system according to embodiments of the present disclosure through a network. .

2 is a block diagram of a server for providing a document analysis method related to insurance subscription according to an embodiment of the present disclosure.

The server 100, which provides a method for insurance subscription using a chatbot according to an embodiment of the present disclosure, responds to a customer's inquiry in a one-on-one conversation with a user for one of consultation, inquiry, or customer service related to personal financial products can be used to Here, the personal financial product may mean a general product related to finance, such as insurance, loan, annuity, and fund.

The document analysis server 100 using a chatbot according to an embodiment of the present disclosure may include a network unit 110 , a processor 120 , and a memory 130 .

According to an embodiment of the present disclosure, the network unit 110 may transmit and receive information with the user terminal 10 and the external server 20 . Specifically, the network unit 110 may receive the user's chatting record generated through the chatbot interface provided to the user terminal 10 . The network unit 110 may transmit an answer of the chatbot generated by the processor 120 based on the input signal of the user terminal 10 to the user terminal 10 . Accordingly, the user may have a conversational interaction with the chatbot through the user terminal 10 , and based on this, the user's chatting record may be generated. In addition, the network unit 110 may receive the user's confirmation signal from the user terminal 10, and may receive the user's personal information from the external server (20). Also, the network unit 110 may receive the insurance subscription document based on the user's upload input for the insurance subscription document through the chatbot interface provided to the user terminal 10 . The insurance subscription document may include one or more documents and documents related to insurance subscription determined in advance based on insurance products and types, and may include an image of documents related to insurance subscription applied by the user.

The document analysis server 100 according to an embodiment of the present disclosure may generate an answer of the chatbot provided to the user terminal 10 . Using the chatbot interface by the user may include chatting with the chatbot through the user terminal 10 .

The processor 120 may determine whether to start the insurance-related procedure based on the user's chat record received through the chatbot interface of the user terminal. The processor 120 may determine whether the user wants a simple chatbot service or an insurance-related procedure based on the user's chat record. When it is determined that the user desires an insurance-related procedure, the processor 120 may initiate an insurance-related procedure. The processor 120 may determine which procedure among insurance-related procedures is desired to be initiated based on the user's chat record. The processor 120 may understand the context of the chat record by processing the chat record in natural language, and the processor 120 may, for example, detect a change in the user's personal information in the context of the chat record, It is possible to determine which insurance procedure should be initiated for the user through a correlation search or the like. The insurance-related procedure may include a user's insurance subscription procedure, a claim procedure, an inquiry procedure, and the like. For example, when the user inputs meaningful chatting in relation to insurance, the processor 120 may cause an insurance-related procedure to be initiated on the chatbot interface. For example, based on the chat record that the user has been to orthopedic surgery, the processor 120 may initiate an insurance claim procedure for the user, an insurance subscription procedure, an insurance subscription inducement procedure, and the like.

The processor 120 may inquire personal information of the user using the chatbot interface from the database. The processor 120 may generate a message requesting the user to allow the inquiry of personal information (eg, providing an option for consent). A message transmitted to the user terminal 10 through the network unit 110 may be displayed through a chatbot interface. When a signal allowing the inquiry of the user's personal information is received from the user terminal 10 , the processor 120 may inquire the user's personal information.

The user's personal information may include information about the current user's status, such as the user's marital status, having children, and job status, user identification information, and insurance subscription history. User personal information may be information pre-stored in a database. When the user identification information is not inquired, the processor 120 may generate a message requesting to input the user identification information. The generated request message may be transmitted to the user terminal 10 , and user identification information may be generated based on an input signal of the user terminal 10 receiving the request message.

The processor 120 according to an embodiment of the present disclosure may generate an answer of the chatbot for recommending an insurance product based on the user's personal information and chat record. Specifically, the processor 120 may receive the user's personal information inquired from the database through the network unit 110 and chat using the user's chatbot interface.

In addition, the processor 120 may recognize the user's personal information change based on the received chatting record. For example, when it is determined that the user's marriage-related content from among the chatting records is determined, the processor 120 may recognize that there is a change in the user's personal information. Accordingly, the processor 120 may generate an answer of the chatbot that asks the user in relation to the change of personal information. The generated answer may be transmitted to the user terminal 10 through the network unit 110 and may be displayed as a question related to personal information change through a chatbot interface. The user's answer to the question related to the change of personal information is based on the input signal of the user terminal 10 and may be received by the network unit 110 . Accordingly, the processor 120 may correct the user's personal information based on the user's answer. Specifically, the user's answer to the question related to personal information change may be first divided into positive and negative. Upon receiving the affirmative answer, the processor 120 may generate an answer of the chatbot requesting additional input for the changed personal information through the chatbot interface. The generated chatbot answer may be transmitted from the network unit 110 to the user terminal 10 to be displayed by the chatbot interface. The processor 120 may correct personal information based on an affirmative answer and additional input from the user terminal 10 . For example, when an affirmative answer and a change in marital status are additionally input from the user terminal 10 , the processor 120 may correct the marital status of the user's personal information based on the change.

The processor 120 according to an embodiment of the present disclosure may select one or more insurance products having a high probability of the user's subscription as the recommended insurance based on the user's personal information. Specifically, the processor 120 may select an insurance product having a high subscription probability based on personal information or an insurance product having a high subscription probability based on a change in personal information as the recommended insurance. For example, the processor 120 may select an insurance product having a high subscription probability as the recommended insurance based on an insurance necessity situation that may occur to the user and the user's ability to subscribe. The insurance requirement and the user's income may be determined by the processor 120 based on the user's personal information. For example, the processor 120 may determine that the unmarried user needs insurance as a disability and medical expenses. Accordingly, it is possible to select insurance related to medical expenses for medical loss and medical expenses for unmarried users, and diagnosis expenses for diseases requiring medical treatment as recommended insurance. As another example, the processor 120 may determine that the insurance requirement situation for the user with dependents is the death of the primary source of income. Accordingly, the processor 120 may select the death insurance as the recommended insurance for the user who has dependents. In addition, the processor 120 may select an insurance product related to one of a life insurance and a term insurance as a recommended insurance based on the user's ability to subscribe.

The processor 120 according to an embodiment of the present disclosure may select one or more insurance products without the user's insurance subscription history as recommended insurance based on the chatting record. Specifically, the processor 120 may obtain an insurance subscription history based on the user's personal information and determine an insurance product without the user's insurance subscription history as the recommended insurance. Also, the processor 120 may select an insurance required for the user as a recommended insurance based on the user's chat record. In addition, the processor 120 may select the insurance required for the user based on the user's chatting record, and select the recommended insurance by filtering through the user's subscription history. For example, the processor 120 may receive a chat record of the user's visit to a hospital. In this case, the processor 120 may determine that the insurance required for the user is actual medical expenses insurance or accident insurance, and the processor 120 determines that there is accident insurance in the user's insurance subscription history but no actual expenses insurance. Insurance products related to insurance can be selected as recommended insurance.

One or more recommended insurances selected by the processor 120 according to an embodiment of the present disclosure may be displayed through the chatbot interface of the user terminal 10 . Specifically, the processor 120 may generate an answer of the chatbot related to the recommended insurance and transmit it to the user terminal 10 through the network unit 110 . Accordingly, by displaying the answer related to the recommended insurance through the chatbot interface of the user terminal 10 , the recommended insurance with a high probability that the user will subscribe may be recommended in the form of a chatting message of the chatbot interface. In addition, the selected one or more recommended insurances may be displayed as options for the user to select through the chatbot interface. Accordingly, the processor 120 may select an insurance product having a high subscription probability as the recommended insurance based on the user's personal information and chatting contents. In addition, through the chatbot interface of the user terminal 10 , a recommended insurance with a high probability of subscription may be recommended to the user to induce the user to subscribe to the insurance.

According to an embodiment of the present disclosure, the processor 120 may transmit insurance information including recommended insurance information to the user terminal 10 through the network unit 110 . In addition, when generating insurance information including recommended insurance, the processor 120 may display the insurance information requested by the user and recommended insurance information separately. For example, the processor 120 may receive a chat record including a request for information related to death insurance from the user terminal 10 through the network unit 110 . In this case, the processor 120 may generate information related to the insurance requested by the user and recommended insurance information based on the user's request. Also, based on the fact that the user's income level is not high, the processor 120 may display information on death insurance related to term insurance among death insurance separately from information requested by the user. In addition, the processor 120 may cause the information of the recommended insurance and the insurance information requested by the user to be separately displayed to the user through the chatbot interface of the user terminal 10 .

According to an embodiment of the present disclosure, the processor 120 may perform an additional operation by recognizing the user's intention for insurance information. For example, the user's intention for insurance information may be determined based on the user's chatting record input through the chatbot interface after providing the insurance information. The processor 120 may determine the user's intention for insurance information based on the user's chatting record input through the chatbot interface after providing the insurance information. User intent for insurance information may include, for example, initiating insurance-related procedures for insurance information, requesting additional information about insurance information, requesting other insurance information, and the like.

After transmitting the insurance information according to an embodiment of the present disclosure, the processor 120 may additionally transmit a question for understanding the user's intention to the insurance information including the recommended insurance information to the user terminal. A question for understanding the user's intention for insurance information may be displayed through the chatbot interface of the user terminal 10 , and the processor 120 may receive the user's opinion based on the user's input into the chatbot interface. For example, the processor 120 that has transmitted the actual cost insurance as the recommended insurance may generate “What do you think about the recommended actual insurance?” as a message of the chatbot in order to understand the user's subscription intention. In this case, the generated chatbot message may be in the form of requesting the user's natural language input, but may also be in the form of allowing the user to select an option suggested by the chatbot message on the chatbot interface. The above-described answer of the chatbot is merely an example, and the present disclosure is not limited thereto. Accordingly, the processor 120 may recognize the user's intention from the user answer received from the user terminal 10 and generate a new chatbot answer.

Specifically, the processor 120 may receive an answer from the user who intends to request information about the recommended insurance. For example, the processor 120 may determine that the user intends to request information when receiving a question related to insurance for actual cost insurance, which is recommended insurance, through the user terminal 10 . Accordingly, the processor 120 may generate an answer including information related to insurance for actual expenses and transmit it to the user terminal 10 to be displayed through the chatbot interface.

When receiving a negative chat record from the user after delivering the insurance information to the chatbot interface, the processor 120 may identify the user's intention and suggest additional insurance-related procedures. When receiving an opinion that there is no intention to join the user terminal 10 , the processor 120 may generate a question about the reason for not intending to join as an answer of the chatbot to be displayed through the chatbot interface. In addition, the processor 120 may display a question related to the user's intention to join the chatbot interface in the form of a natural language or an option. For example, an option for a reason for not wanting to subscribe may be indicated by the insurance already purchased, the amount of coverage, and the premium. Based on the touch input of the user terminal 10 , at least one of the options may be selected for the reason that the user does not intend to subscribe. In this case, the processor 120 may recommend another insurance based on the reason that the user does not intend to subscribe or may generate an answer of the chatbot providing detailed information of the insurance. For example, based on the input signal of the user terminal 10 , when the reason for not wanting to purchase the life insurance is the premium, the processor 120 may select the recommended insurance instead of the life insurance. Accordingly, the processor 120 may determine the reason for not wanting to subscribe through the chatbot, and provide real-time feedback on insurance that is not suitable for the user. Accordingly, the processor 120 may recommend an optimal insurance to the user to induce subscription. In the case of the user, information on multiple insurances can be conveniently requested when signing up for insurance, and the chatbot can provide information whenever the user wants by performing 24-hour consultation, thus improving ease of use. In addition, in the case of an insurance company, by performing subscription consultation through a chatbot interface, the cost of manpower related to consultation can be reduced.

The processor 120 according to an embodiment of the present disclosure may generate an answer of the chatbot to the detailed information of the recommended insurance based on the reason that the user does not intend to subscribe. For example, when the recommended insurance is a term insurance related to medical expenses, the processor 120 may generate an answer of the chatbot that recommends one or more special contracts based on the reason that there is no intention to subscribe. The processor 120 may select one or more special contracts based on the user's opinion on the answer of the chatbot. Accordingly, the processor 120 may determine the insurance that the user intends to subscribe to based on the selected special contract. Accordingly, the processor 120 may select detailed conditions of the recommended insurance through the chatbot interface to provide an insurance that can satisfy the user's subscription intention. Users can conveniently add or remove multiple specials that can be added to insurance through the chatbot interface, which can improve usability in insurance subscriptions.

The processor 120 may determine the user's intention to initiate an insurance-related procedure based on the chat record. In this case, the processor 120 may determine what insurance-related procedure the user wants to process through the chatbot interface, based on the chat record, user information, and the like. The processor 120 may determine whether an insurance-related procedure that the user wants to process through the chatbot interface can be processed through the chatbot interface or a procedure to be processed by the manager. For example, the processor 120 may determine to initiate an insurance subscription procedure when the user wants to subscribe to simple daily accident insurance based on the user's chatting record. In this case, the processor 120 may determine that the corresponding procedure can be processed as a consultation through the chatbot interface according to the type of insurance-related procedure that the user wants to initiate. In other words, when the procedure the user wants to handle is low attention due to the insurance company's predetermined criteria (for example, when the enrollment procedure is simple, there are few required documents, the amount of coverage is small, the coverage period is short, etc.) , the processor 120 may determine this to be a procedure that can be processed through the chatbot interface, and determine the start of processing through the chatbot interface of the insurance-related procedure.

As described above, a user can conveniently inquire about insurance through the chatbot interface according to an embodiment of the present disclosure, and the chatbot can respond 24 hours a day, thereby improving ease of use. In addition, insurance-related information can be easily provided through a familiar chatbot interface to generations who are reluctant to make phone calls or face-to-face consultations, so usability can be improved.

The processor 120 according to an embodiment of the present disclosure may start a subscription procedure for the insurance for which the user has applied for the subscription procedure through the chatbot interface. Specifically, when receiving an opinion that there is an intention to subscribe from the user terminal 10 , the processor 120 may select an insurance product related to the intention to subscribe. In addition, after providing insurance information to the user, the processor 120 may determine an insurance product by hitting an input to the user's chatbot interface for the insurance information. The processor 120 may generate a message for performing the insurance subscription procedure on the chatbot interface and transmit it to the chatbot interface.

The processor 120 may generate the chatbot's answer to inform the user about one or more essential documents required for the subscription process in relation to the insurance to initiate the subscription process. For example, the processor 120 may generate a chatbot answer to the types of documents required for subscription. For example, the processor 120 may generate a chatbot message indicating that a medical certificate and a health checkup result sheet are required for subscription to the user through the chatbot interface. The above-mentioned answer of the chatbot is only an example, and the present disclosure is not limited thereto.

The processor 120 according to an embodiment of the present disclosure may transmit a message of the chatbot for uploading essential documents for the insurance procedure to the chatbot interface. The processor 120 may generate an answer of the chatbot to notify the user about how to upload one or more required documents. For example, the processor 120 may designate an update location for each required document, generate a chatbot message, and transmit it to the chatbot interface. For example, the user can access the document upload interface by clicking the chatbot's message, and can upload documents to the document upload interface. In addition, the processor 120 transmits a message specifying the document to be requested to the user through the chatbot interface, and a file (eg, image, pdf file, etc.) received within a predetermined time range and message range after the message specifying the document ) can be judged as the upload of the document. The processor 120 transmits the request document to the user in the form of a chatting message to the chatbot interface, and associates the file uploaded from the user with the chat message for the request document of the chatbot to determine the type of the file uploaded from the user. . For example, the processor 120 may transmit a message requesting a health checkup result sheet to the user through the chatbot interface, and the user may upload a file related to the health checkup result sheet through a file sharing interface provided in the chatbot interface. have. The processor 120 may determine the correlation between the file uploaded by the user and the chatbot message within the time range or the message range to determine that the file is a health checkup result.

The processor 120 according to an embodiment of the present disclosure may transmit a guide document that requires a user's signature for insurance subscription, and may obtain the user's signature from an input to the user's signature input interface provided in the chatbot interface. have.

Hereinafter, a method of analyzing a document related to insurance subscription according to an embodiment of the present disclosure will be described.

According to an embodiment of the present disclosure, the processor 120 may receive a document related to insurance subscription from the user terminal 10 . Specifically, the processor 120 may transmit insurance guide information to the user terminal 10 based on the type of insurance the user has applied for through a chatbot interface provided to the user terminal 10 . The insurance guide information may include information capable of providing a user with a guide related to subscription based on an insurance product. For example, the insurance guide information may include information related to a required document required for subscription of an insurance product. Accordingly, the processor 120 may receive the user's confirmation signal for the insurance guide information, and receive the insurance subscription document based on the user's upload input for the insurance subscription document through the chatbot interface provided to the user terminal. can The insurance subscription document may include one or more documents related to the subscription of the insurance for which the user has applied for subscription. For example, it may be a photographed image of a document related to insurance subscription (eg, insurance subscription application).

According to an embodiment of the present disclosure, the processor 120 may acquire information on one or more essential documents from the external server 20 based on the type of insurance for which the user has applied for subscription. Required documents may include documents necessary for insurance-related procedures, and may be existing documents stored in a database. For example, the required documents may include documents required to purchase insurance. For example, it may include a medical certificate, contract, etc. required for insurance subscription. The user's insurance subscription may be determined through at least one review process including contract review.

According to an embodiment of the present disclosure, the processor 120 may input the insurance subscription document into the document analysis model and analyze the insurance subscription document using the document analysis model. Specifically, the processor 120 may identify the insurance subscription document received from the user terminal 10 by using a document analysis model to identify one or more characters and signatures included in the insurance subscription document. The signature may include a handwritten signature and an electronic signature that can identify at least one of the policyholder and the policyholder. In addition, the document analysis model may identify one or more characters included in the imaged insurance subscription document using an optical character recognition (OCR) technology. One or more characters included in the insurance subscription document may include characters or symbols written by the user or the insured. Specifically, one or more characters included in the insurance subscription document may include characters or symbols recorded in the insurance subscription document by handwriting of the user or the insured.

In addition, the document analysis model may perform natural language processing on the identified one or more characters to read the structure of the insurance subscription document. The read structure of the insurance subscription document may be generated as structural information of the insurance subscription document through the document analysis model. The structural information of the insurance subscription document includes information on one or more items included in the insurance subscription document, and the information on the items includes information on the name of each item extracted through context analysis and item-by-item in the insurance subscription document Information about each location may be included. Specifically, when the imaged insurance subscription document is an insurance subscription application, the document analysis model may perform natural language processing on one or more characters identified through optical character recognition to primarily classify the title and content. For example, the document analysis model may determine the type of the insurance subscription document as the insurance subscription application based on the title of the insurance subscription document imaged through context analysis of the insurance subscription document. In addition, the document analysis model may determine one or more items including names such as “name” and “gender” indicating identification information of the insurance subscriber in the content of the insurance subscription document. In addition, the document analysis model may extract location information on one or more items as information about the structure of an imaged insurance subscription document, and may associate the location information with the determined name information for each item. The location information for one or more items may include information about a location relationship between items and a location of each item in a document.

The processor 120 according to an embodiment of the present disclosure may obtain, through the external server 20 , information on essential documents of the previously stored database from the external server 20 based on the type of insurance product applied through the user terminal. can The essential document is a document related to insurance subscription including one or more essential items necessary for insurance subscription based on the type of insurance the user has applied for, and at least one or more essential documents based on one or more types may be included in the database, respectively. . The information on the required document may include information on the structure of the required document, and the structure information of the required document may include name and location information of one or more essential items.

In addition, when the type of the insurance subscription document received from the user does not correspond to at least one type based on the type of one or more essential documents obtained from the external server 20, the processor 120 generates document supplement information for this. can do. Specifically, when there are 4 required documents based on the type of insurance and there are not 4 insurance subscription documents received from the user or there is no corresponding type of insurance subscription document, document supplementary information for requesting feedback from the user can create In addition, the processor 120 may generate the document supplement information based on whether characters and symbols corresponding to the structure information of the insurance subscription document generated through the document analysis model are identified.

The processor 120 according to an embodiment of the present disclosure obtains information on a document essential for analyzing the structure information of the insurance subscription document through the network unit 110, and the structure information on the items of the insurance subscription document is It may be stored in the memory 130 .

According to an embodiment of the present disclosure, the processor 120 may read the structure of the insurance subscription document by comparing the context analysis result of the document analysis model with the structure of the essential document, which is the existing insurance subscription document, and generate the structural analysis submodel. The structure of the insurance subscription document can be read through. The processor 120 may search for an essential document for efficiently determining the structure of the insurance subscription document through the structural analysis submodel learned using the reinforcement learning method. The structural analysis submodel may be a reinforcement learning model using a policy network and a value network.

The structural analysis submodel is trained based on the analysis result of the imaged insurance subscription document. The structural analysis submodel determines whether to select another essential document by processing the comparison result of the structural information included in the selected essential document and the structural information of the imaged insurance subscription document using the policy network, and It can be learned to calculate a reward based on the structure information and the structure information of the imaged insurance subscription document, and select other essential documents that can maximize the reward.

The structural analysis submodel that receives structural information for one or more items included in the insurance subscription document through the document analysis model may search for essential documents having a structure similar to the input structural information. The structural analysis submodel has the searched essential document as one state, document search as an action, and calculates a reward based on the comparison result of the searched essential document and the structural information of the input insurance subscription document can do. The structure analysis submodel may calculate a reward based on the required documents and structure information. The structure analysis submodel may determine whether to search for a new document based on a result of comparing the searched essential document with the structure information. The structure analysis submodel may determine at least one of whether to search for a new document or which new document to search by processing the searched essential document and structure information using the policy network. The processor 120 may determine whether to search for a new document or which new document to search by using the structure analysis submodel. The structural analysis submodel can search for new documents so that the rewards are maximized. The value network of the Structural Analysis submodel can be trained on which new documents to search for will maximize rewards. The structure analysis submodel may determine a reward based on the newly discovered document and structure information. The reward may be determined based on a comparison result between the searched document and the structure information, and may be a measure related to the degree of similarity between the searched document and the structure information. The processor 120 may search for an essential document having a structure most similar to the insurance subscription document using the structural analysis submodel learned using the reinforcement learning methodology.

The policy network of the Structural Analysis sub-model may decide whether to search for a new required document or which new document to search for. Required documents may include documents necessary for insurance-related procedures, and may be existing documents stored in a database. For example, the required documents may include documents required to purchase insurance. For example, it may include a medical certificate, contract, etc. required for insurance subscription. The type of a new document to be searched for may include, for example, a criterion for selecting a document in a next state based on what type of document to be searched and a group to which the newly searched document belongs. For example, when the correspondence between the currently searched essential document and the structure information is high, the policy network may be trained to determine that the search for a new required document is unnecessary. Also, for example, when the correspondence between the currently searched essential document and the structure information is low, the policy network may be trained to determine that the search for a new document is necessary. In this case, the policy network may determine a cluster of essential documents to newly search for a document based on the type of the insurance subscription document identified based on the structure information. For example, if the insurance policy document is an insurance policy application, the policy network may be trained to determine a search for a mandatory document belonging to a class corresponding to the identified type of insurance policy document when a new mandatory document search is required.

Specifically, the type and structure information of the insurance subscription document received from the user is generated through the document analysis model of the processor 120, and the document analysis model analyzes the structural information on the type and item of the insurance subscription document received from the structural analysis sub can be entered into the model. The structural analysis submodel is trained to discover a cluster of one or more essential documents corresponding to a predetermined type of insurance subscription document in the database through the document analysis model. The structural analysis submodel performs a search for one or more essential items included in any one essential document in order to select an optimal essential document for determining the structure of the insurance subscription document. The purpose of the search is to find the required documents that contain the most required items that are most similar to one or more items contained in the insurance policy document. The structural analysis submodel determines a reward according to the similarity between the structural information of the essential document and the insurance subscription document, and when the processor 120 takes an action to search for other essential documents, other essential documents newly searched through this action are currently It can be learned to evaluate whether it contains more essential items of similar structural information than essential documents through the value function. For example, when the processor 120 determines that the insurance subscription document received from the user terminal 10 is the insurance subscription application, information on the name of each item included in the insurance subscription document is input to the structural analysis submodel. can The structural analysis submodel may be a model trained to discover a cluster of required documents corresponding to a predetermined insurance application form from a database. Accordingly, the processor 120 determines that any one essential document selected from the cluster of essential documents using the structural analysis submodel is the insurance subscription document determined as the insurance application application (ie, the insurance subscription document received from the user terminal). Rewards are calculated based on how many required items that are most similar to the name for each item are included. Specifically, the processor 120 includes all of the items of the insurance application for which the essential items included in the first insurance application form randomly selected from the cluster of essential documents using the structural analysis submodel are received, and other essential documents of the cluster. When the mandatory items included in the second application for insurance include only the name of the insured, it may be determined that the reward of the first application for insurance is higher than that of the second application for insurance. In this case, the processor 120 selects the first insurance application by using the structural analysis model and calculates a reward therefor. Accordingly, the structural analysis sub-model using reinforcement learning determines the reward based on the comparison result with each required document for the document to be judged, and compares the reward for each required document with each other, so that the required document with a larger reward shows a tendency to choose After a lot of learning, you can find the required document that contains the most required items similar to the name of the item included in the insurance subscription document you receive. In addition, when using reinforcement learning to search for essential documents that contain the most items included in insurance subscription documents, it can be searched faster than other machine learning methods.

For a description of specific contents related to information retrieval using reinforcement learning for quick retrieval, the Korean Information Science Association thesis, which is incorporated by reference in its entirety in this disclosure, is information retrieval using reinforcement learning (published date: February 2002, author: Jeong Tae-jin, Jang Byung-tak) is discussed in detail.

The processor 120 according to an embodiment of the present disclosure reads the structure of the insurance subscription document based on the final selected essential document including the most names of items of the insurance subscription document through the structural analysis submodel. There may not be a structure that completely corresponds to all items of the insurance subscription document received. That is, the location and name of the essential item of the finally selected essential document and the item included in the received insurance subscription document may not correspond to each other. For example, if the mandatory document most similar to the insurance subscription document is finally selected from the database through the structural analysis submodel, and there are 7 mandatory items included in the mandatory document, the insurance subscription document may contain not 7 items. have. Alternatively, with respect to the items included in the insurance subscription document, the mandatory items included in the mandatory documents may be mandatory items that do not correspond to. Alternatively, the positions of the essential items and the positions of the items included in the insurance subscription document may not correspond to each other.

The processor 120 according to an embodiment of the present disclosure may determine the insurance document received from the user terminal 10 based on the essential document finally selected by the structural analysis submodel. Insurance documents may include any type of insurance-related documents. For example, the insurance document may include documents related to insurance subscription, claim, and the like. The insurance document may include, for example, an insurance subscription document, and may include a subscription application required for insurance subscription, a medical certificate, and the like. Specifically, the processor 120 determines that the number of essential items in the final selected essential document does not correspond to the number of items in the insurance subscription document, or that the name of at least one essential item does not correspond to the name of the item in the insurance subscription document. can judge In this case, the processor 120 may generate document supplementation information for notifying the user of the information on the non-corresponding essential item received from the user terminal 10 . For example, the first insurance application form finally selected by the structural analysis submodel includes the applicant's name, age, and insurance application details, and the insurance application received from the user includes the applicant's name and insurance application content items only, the processor 120 may generate information on the age item of the first insurance application form that does not correspond to the essential items of the selected first insurance application form as document supplementary information.

In addition, the processor 120 compares the location information of items included in the received insurance subscription document based on the final selected essential document, and when the location information is different, the processor 120 provides structural information for the non-corresponding item can be created as new structure information. Specifically, the processor 120 may determine whether items of the received insurance subscription document correspond to essential items included in the essential document by using the structural analysis submodel. The processor 120 may compare the location information of the items of the insurance subscription document received from the document analysis model based on the location information of the essential items using the structure analysis sub-model. When the location information for at least one item does not correspond to the location information for the essential item, the processor 120 may generate structure information of an insurance subscription document including the non-corresponding item as new structure information. The structural information of the insurance subscription document may be based on a result analyzed through the document analysis model, and structural information on one or more items included in the insurance subscription document may be included. For example, the position of the applicant's name item included in the received insurance application is at the top of the insurance application form, but the position of the applicant's name item in the first application for insurance, which is a required document, is located at the top of the first insurance application. It could be the bottom. Accordingly, the processor 120 uses the structural analysis submodel to recognize the applicant's name item, which is different from the position of the essential item of the required document, as new structural information of the insurance application, and includes the applicant's name item located at the top. The entire structure information of the insurance application form can be created as new structure information. Accordingly, the structure analysis submodel of the processor 120 may transmit new structure information corresponding to the names and locations of each item of the received insurance subscription document to the external server 20 . Accordingly, the new structure information may be stored in the database of the external server 20 as structure information of the essential document and transmitted to the document analysis server 100 . The transmitted new structure information may be used by the processor 120 to determine the structure of the insurance subscription document.

Also, the processor 120 may determine an insurance subscriber based on at least one of the first user information obtained from the user terminal 10 and the second user information obtained from the insurance subscription document. Accordingly, the insurance subscription for the insurance subscriber may be determined based on at least one of the type or content of the insurance subscription document based on the analysis result of the insurance subscription document. The decision to purchase insurance may be performed based on at least one of a type or content of an insurance subscription document for the insured, and may generate a review result. The result of the examination may include conditional approval information based on the recognition result for each item included in the insurance subscription document. Specifically, the processor 120 may generate conditional approval information based on the recognition result for each item included in the insurance subscription document, and the processor 120 may generate a review result for insurance subscription based on the conditional approval information. can For example, the processor 120 may generate conditional approval information including a condition for setting an exemption range for an insurance subscriber with diabetes. Alternatively, the processor 120 may generate conditional approval information by recognizing an item based on a doctor's certificate including proof that the subscriber's blood sugar has decreased by a predetermined value for the insurance subscriber with diabetes. Alternatively, the processor 120 obtains conditional approval information for the insured with diabetes based on whether or not the conditions for the recorded information have been achieved by interworking with an exercise-related application such as the Samsung Health application to achieve more than a certain number of exercises for a certain period of time. can create The processor 120 may generate the examination result for insurance subscription for the insured with diabetes based on the record information on which exercise is achieved or the medical certificate on which blood sugar reduction is recorded. The foregoing description is merely an example, and the present disclosure is not limited thereto.

The processor 120 according to an embodiment of the present disclosure may perform an insurance subscription review for determining insurance subscription based on an analysis result of the insurance subscription document. The examination of insurance subscription may include contract examination, and the processor 120 may generate underwriting supplementary information for the examination based on the recognition result for each item included in the insurance subscription document. The underwriting supplemental information may include information based on the contract examination result, and the processor 120 may generate the examination result based on the underwriting supplemental information. Accordingly, the processor 120 may request the user's feedback based on the underwriting supplementary information to the user terminal 10 or transmit a notification regarding the information of the examination result.

The document analysis server 100 according to an embodiment of the present disclosure recognizes at least one of the type or content of the insurance subscription document based on the analysis result of the insurance subscription document, and provides information on at least one of the type or content of the insurance subscription document. You can decide to purchase insurance based on this. In addition, the document analysis server 100 may perform a review on insurance subscription based on at least one of the type or content of the insurance subscription document, and generate a review result according thereto. In addition, the document analysis server 100 may generate underwriting supplemental information and conditional approval information based on the recognition result for each item included in the insurance subscription document.

Also, the processor 120 may generate document supplemental information for insurance subscription based on the analysis result of the received insurance subscription document. Document supplement information includes information on whether or not there are missing items in the insurance subscription document received from the user terminal 10, information on whether anomaly exists, whether the document needs to be supplemented, and if necessary, what kind of supplementation should be made may include information related to Accordingly, the processor 120 may determine whether at least one or more missing items exist in the insurance subscription document, and generate document supplement information based on the existing one or more missing items. Specifically, the document supplement information includes one or more pieces of information related to essential documents that require user feedback, such as errors in insurance subscription documents, additional input required, or information on required items that are missing or unrecognized. may include In addition, the document supplement information may be generated by the processor 120 based on whether an anomaly exists in an item included in the insurance subscription document. Anomalies may include, for example, cases where the value of an item for an item included in an insurance policy deviates from a general trend. Specifically, the processor 120 may determine whether one or more anomalies exist in an item included in the insurance subscription document, and generate document supplemental information based on the one or more anomalies present. For example, when the content of the weight item of the insured person is identified as “70g” in the analyzed document related to insurance subscription, the processor 120 may determine that an anomaly exists with respect to the weight item. Accordingly, the processor 120 may generate document supplement information for the weight item in order to request feedback from the user. Also, the processor 120 may determine that the anomaly exists even when the content of the same item does not match with respect to at least two or more insurance subscription documents for the corresponding type of insurance subscription. For example, the processor 120 may analyze two insurance subscription documents received from the user terminal 10 for the same type of insurance subscription, respectively. If the analyzed insurance subscription document is an insurance subscription application form and it is a doctor's certificate, if it is determined that the name of the insurance subscription application form and the doctor's diagnosis certificate are different from among the items included in the identification item of the insured in each analyzed document, In order to solicit feedback from users, it is possible to create documentation supplementary information for the insured's name. The generated document supplementation information may be transmitted to the user terminal 10 by the processor 120 .

According to an embodiment of the present disclosure, the underwriting supplemental information and the document security information may be transmitted to the user terminal 10 through the network unit 110, and feedback from the user terminal 10 based on the underwriting supplemental information and the document supplementary information information can be received. The processor 120 may request this from the user through the chatbot interface when supplementation is required for the recognition content of the document or when additional documents for underwriting are required. The processor 120 may receive a supplemental message from the user through the chatbot interface, and may perform supplementation such as document recognition or underwriting. In this case, the feedback information may be information including the corrected content of the document in correspondence with the document supplement information. Specifically, the processor 120 may generate an answer of the chatbot informing the user about the portion in the document in which the anomaly exists and the portion not recognized as a character based on the document supplement information. For example, when the processor 120 determines that the name is not recognized based on the document supplement information, the chatbot responds to this as “The name is not recognized in the document.” can be created with As another example, when the user's signature is not recognized, the processor 120 may generate "Signature not recognized" as the chatbot's answer. As another example, when the processor 120 determines that there is an anomaly in the body weight based on the document supplement information, the chatbot responds to the response “The weight is strange.” It can be created by or “Please check your weight again.” The above-mentioned answer of the chatbot is only an example, and the present disclosure is not limited thereto. In addition, the processor 120 may transmit the document to the user terminal 10 so that the user can check the part of the document notified through the chatbot based on the document supplement information through the chatbot interface.

In addition, the processor 120 may additionally generate an answer of the chatbot requesting to correct the unrecognized portion of the document through the user terminal 10 . For example, for a document whose name is not recognized, the processor 120 responds to the chatbot with “Name is not recognized in the document. Please enter your name here.” For example, when the processor 120 determines that the user's name is not recognized based on the document supplement information, it may transmit an input interface for inputting the user's name to the chatbot interface to request input of the name. For example, with respect to the document for which the signature is not recognized, the processor 120 may display “Signature not recognized. Please sign here.” as the chatbot's response. As another example, in response to the anomaly about the weight of the document, the processor 120 may say, “The weight is strange. Please re-enter your weight here.” as the chatbot's response. Accordingly, the processor 120 may determine whether to revise the part of the document based on the input signal of the user terminal 10 to the answer of the chatbot. The above-described answer of the chatbot is merely an example, and the present disclosure is not limited thereto.

As such, when user confirmation or additional input is required for document recognition, the user can conveniently request an answer through the chatbot interface, so that the document recognition accuracy can be improved, and the user can check the recognition result of the chatbot. This can increase the user's trust in the chatbot and improve usability. Users can conveniently perform supplementary work on documents through chatting with chatbots, providing convenience to generations familiar with chatting, etc., and insurance companies can improve recognition accuracy by requesting additional input from users. .

In addition, the chat record additionally input by the user may be stored in the server 100 and included in the learning data for document recognition. In the case of a document in which the user confirms the recognition result through the chatbot interface, new learning data can be generated by using the user's confirmation of the recognition result as a label for the document.

The processor 120 according to an embodiment of the present disclosure may transmit a confirmation request to the manager terminal for the modified part through the chatbot interface. In more detail, the processor 120 may transmit a document to which the modified part is applied and a request for confirmation thereof to the manager terminal based on the input signal of the user terminal 10 . In addition, the processor 120 may generate a chatbot answer informing that the manager is confirming the status before receiving the confirmation signal from the manager terminal and transmit it to the user terminal 10 . In addition, when receiving a confirmation signal for document modification from the manager terminal, the processor 120 may generate a chatbot answer informing about the manager's confirmation and transmit it to the user terminal 10 . The processor 120 may determine the revised item and the written content to be reflected in the document structure analysis result based on the manager confirmation signal. Accordingly, the processor 120 may provide the user with a function for easily correcting typos and undescribed parts of the document through the chatbot interface without having to re-upload the entire document. Also, the processor 120 may generate an answer of the chatbot informing about at least one or more missing documents based on the document supplement information. Specifically, the processor 120 may generate information on the missing document and an answer of the chatbot for requesting the missing document based on the document supplement information and transmit it to the user terminal 10 . In this way, the user can easily grasp the progress through the chatbot interface. In the case of an insurance company, since the chatting record through the chatbot interface is retained in the server 100, it can be used as evidence in case of disputes such as incomplete sales to users in the future.

The processor 120 according to an embodiment of the present disclosure may generate an answer of the chatbot asking the user about one or more documents whose types are unknown. Specifically, when the types of one or more documents analyzed through the document analysis model are not determined, the processor 120 may predict one or more types of documents based on the document analysis result. Accordingly, the processor 120 may generate a question about the type of one or more expected documents as an answer of the chatbot or as an option. The generated chatbot answers and options may be transmitted to the user terminal 10 including one or more documents and displayed through the chatbot interface. Accordingly, the processor 120 may determine the type of the document based on the user's input signal and answer. For example, when the processor 120 expects an insurance application for the first document and the second document, the chatbot answers the question “Are there any insurance applications?” Or, "Is the first document an insurance application?" Accordingly, the processor 120 may determine the type of the unclear document based on the input signal of the user terminal 10 .

The processor 120 according to an embodiment of the present disclosure may transmit the analyzed document to the user terminal 10 through the document analysis model to request the user's confirmation signal. Specifically, the processor 120 may transmit one or more text-displayed documents to the user terminal 10 through OCR of the document analysis model. For example, the processor 120 may transmit the recognized document in the form of an html document to the chatbot interface. The user can review whether the recognized document was recognized correctly in the chatbot interface. In this case, the processor 120 may highlight a specific portion of a document including one or more texts. For example, if you are applying for an insurance product that is not covered by high blood pressure, you can highlight the blood pressure-related part of the document. The document including the highlight mark may be transmitted to the user terminal 10 and displayed through the chatbot interface. Since these document recognition records can be held in the server 100 through the chatbot interface, the user can better look at the important parts of the insurance contract, and can inquire at any time through the chat record, so that the user's insurance document It can increase understanding, and the insurance company can use it in case of insurance-related disputes such as incomplete sales.

According to an embodiment of the present disclosure, the processor 120 may receive the user's confirmation signal through the chatbot interface, and transmit the insurance subscription examination result to the user terminal 10 . Specifically, the processor 120 may receive, from the user terminal 10 , a confirmation signal for the items corrected based on the underwriting supplement information and the document supplement information. Upon receiving the confirmation signal from the user terminal 10 , the processor 120 may transmit the insurance subscription examination result based on the insurance document in which the amended matters are reflected to the user terminal 10 . In addition, the processor 120 may transmit the insurance subscription examination result to the external server 20 and may select another insurance as the recommended insurance based on the examination result. Specifically, the processor 120 may modify the user's insurance subscription history based on the insurance subscription examination result. A case in which the user's insurance subscription is rejected may also be added to the insurance subscription history through the processor 120 . Accordingly, the processor 120 may select another insurance as the recommended insurance based on the reason that the user's insurance subscription is rejected. The other selected insurance may be displayed by being transmitted to the user terminal 10 through the chatbot interface. Accordingly, the document analysis server 100 may process the user's request for insurance subscription and examination on holidays and 24 hours a day. In particular, the document analysis server 100 may provide a function of performing an insurance subscription review for a working user even on weekends or holidays, and editing and reviewing within a short time.

Users can conveniently consult insurance through the chatbot interface, and deliver insurance-related documents in a convenient way like sharing files in chat. This can help insurance companies cut costs. In addition, since the user and the chatbot can supplement the recognition of insurance-related documents or request additional documents through a convenient interface of the chatbot type, the recognition accuracy can be improved, and the accuracy of insurance-related processing can be improved.

3 is a schematic diagram illustrating a network function underlying a structure analysis model for determining the structure of a document according to an embodiment of the present disclosure.

Throughout this specification, computational model, neural network, network function, and neural network may be used interchangeably. A neural network may be composed of a set of interconnected computational units, which may generally be referred to as nodes. These nodes may also be referred to as neurons. A neural network is configured by including at least one or more nodes. Nodes (or neurons) constituting the neural networks may be interconnected by one or more links.

In the neural network, one or more nodes connected through a link may relatively form a relationship between an input node and an output node. The concepts of an input node and an output node are relative, and any node in an output node relationship with respect to one node may be in an input node relationship in a relationship with another node, and vice versa. As described above, an input node-to-output node relationship may be created around a link. One or more output nodes may be connected to one input node through a link, and vice versa.

In the relationship between the input node and the output node connected through one link, the value of the output node may be determined based on data input to the input node. Here, a node interconnecting the input node and the output node may have a weight. The weight may be variable, and may be changed by the user or algorithm in order for the neural network to perform a desired function. For example, when one or more input nodes are interconnected to one output node by respective links, the output node includes values input to input nodes connected to the output node and weights set in links corresponding to the respective input nodes. An output node value may be determined based on .

As described above, in a neural network, one or more nodes are interconnected through one or more links to form an input node and an output node relationship within the neural network. The characteristics of the neural network may be determined according to the number of nodes and links in the neural network, the correlation between the nodes and the links, and the value of a weight assigned to each of the links. For example, when the same number of nodes and links exist and there are two neural networks having different weight values between the links, the two neural networks may be recognized as different from each other.

A neural network may include one or more nodes. Some of the nodes constituting the neural network may configure one layer based on distances from the initial input node. For example, a set of nodes having a distance n from the initial input node may constitute n layers. The distance from the initial input node may be defined by the minimum number of links that must be traversed to reach the corresponding node from the initial input node. However, the definition of such a layer is arbitrary for description, and the order of the layer in the neural network may be defined in a different way from the above. For example, a layer of nodes may be defined by a distance from the final output node.

The initial input node may refer to one or more nodes to which data is directly input without going through a link in a relationship with other nodes among nodes in the neural network. Alternatively, in a relationship between nodes based on a link in a neural network, it may mean nodes that do not have other input nodes connected by a link. Similarly, the final output node may refer to one or more nodes that do not have an output node in relation to other nodes among nodes in the neural network. In addition, the hidden node may mean nodes constituting the neural network other than the first input node and the last output node. The neural network according to an embodiment of the present disclosure may be a neural network in which the number of nodes in the input layer may be the same as the number of nodes in the output layer, and the number of nodes decreases and then increases again as progresses from the input layer to the hidden layer. can In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes in the input layer may be less than the number of nodes in the output layer, and the number of nodes decreases as the number of nodes progresses from the input layer to the hidden layer. have. In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes in the input layer may be greater than the number of nodes in the output layer, and the number of nodes increases as the input layer progresses to the hidden layer. can The neural network according to another embodiment of the present disclosure may be a neural network in a combined form of the aforementioned neural networks.

A deep neural network (DNN) may refer to a neural network including a plurality of hidden layers in addition to an input layer and an output layer. Deep neural networks can be used to identify the latent structures of data. In other words, it can identify the potential structure of photos, texts, videos, voices, and music (for example, what objects are in the photos, what the text and emotions are, what the texts and emotions are, etc.) . Deep neural networks include convolutional neural networks (CNNs), recurrent neural networks (RNNs), auto encoders, generative adversarial networks (GANs), and restricted Boltzmann machines (RBMs). boltzmann machine), a deep belief network (DBN), a Q network, a U network, a Siamese network, and the like. The description of the deep neural network described above is only an example, and the present disclosure is not limited thereto.

In an embodiment of the present disclosure, the network function may include an auto-encoder. The auto-encoder may be a kind of artificial neural network for outputting output data similar to input data. The auto encoder may include at least one hidden layer, and an odd number of hidden layers may be disposed between the input/output layers. The number of nodes in each layer may be reduced from the number of nodes of the input layer to an intermediate layer called the bottleneck layer (encoding), and then expanded symmetrically with reduction from the bottleneck layer to the output layer (symmetrically with the input layer). In this case, although the dimensionality reduction layer and the dimension reconstruction layer are illustrated as being symmetrical in the example of FIG. 3 , the present disclosure is not limited thereto, and the nodes of the dimension reduction layer and the dimension reconstruction layer may or may not be symmetrical. The auto-encoder can perform non-linear dimensionality reduction. The number of input layers and output layers may correspond to the number of sensors remaining after pre-processing of input data. In the auto-encoder structure, the number of nodes of the hidden layer included in the encoder may have a structure that decreases as the distance from the input layer increases. If the number of nodes in the bottleneck layer (the layer with the fewest nodes between the encoder and the decoder) is too small, a sufficient amount of information may not be conveyed, so a certain number or more (e.g., more than half of the input layer, etc.) ) may be maintained.

The neural network may be trained by at least one of supervised learning, unsupervised learning, and semi-supervised learning. The training of the neural network is to minimize the error in the output. In the training of a neural network, iteratively inputs the training data to the neural network, calculates the output of the neural network and the target error for the training data, and calculates the error of the neural network from the output layer of the neural network to the input layer in the direction to reduce the error. It is a process of updating the weight of each node in the neural network by backpropagation in the direction. In the case of teacher learning, learning data in which the correct answer is labeled in each learning data is used (ie, labeled learning data), and in the case of comparative learning, the correct answer may not be labeled in each learning data. That is, for example, the learning data in the case of teacher learning regarding data classification may be data in which categories are labeled in each of the learning data. Labeled training data is input to the neural network, and an error can be calculated by comparing the output (category) of the neural network with the label of the training data. As another example, in the case of comparison learning about data classification, an error may be calculated by comparing the input training data with the neural network output. The calculated error is back propagated in the reverse direction (ie, from the output layer to the input layer) in the neural network, and the connection weight of each node of each layer of the neural network may be updated according to the back propagation. A change amount of the connection weight of each node to be updated may be determined according to a learning rate. The computation of the neural network on the input data and the backpropagation of errors can constitute a learning cycle (epoch). The learning rate may be applied differently according to the number of repetitions of the learning cycle of the neural network. For example, in the early stage of training of a neural network, a high learning rate can be used to enable the neural network to quickly acquire a certain level of performance, thereby increasing efficiency, and using a low learning rate at the end of learning can increase accuracy.

In the training of neural networks, in general, the training data may be a subset of real data (that is, data to be processed using the trained neural network), and thus the error on the training data is reduced, but the error on the real data is reduced. There may be increasing learning cycles. Overfitting is a phenomenon in which errors on actual data increase by over-learning on training data as described above. For example, a phenomenon in which a neural network that has learned a cat by showing a yellow cat does not recognize that it is a cat when it sees a cat other than yellow may be a type of overfitting. Overfitting can act as a cause of increasing errors in machine learning algorithms. In order to prevent such overfitting, various optimization methods can be used. In order to prevent overfitting, methods such as increasing training data, regularization, or dropout in which a part of nodes in the network are omitted in the process of learning, may be applied.

4 is a diagram exemplarily illustrating a document analysis method related to insurance subscription according to an embodiment of the present disclosure.

The document analysis server 100 may determine whether to start an insurance-related procedure with the user based on the chatting message received from the user terminal 10 .

According to an embodiment of the present disclosure, the document analysis server 100 may receive a user confirmation signal for insurance guide information and a document for insurance subscription from the user terminal 10 ( 201 ). Specifically, the document analysis server 100 may transmit information on insurance subscription to the user terminal through a chatbot interface provided to the user terminal 10 and receive a user confirmation signal for the insurance guide information. The document analysis server 100 may receive the insurance subscription document based on the user's upload input for the insurance subscription document through the chatbot interface provided to the user terminal 10 . The insurance subscription document may include an image of a document related to insurance subscription.

In addition, the document analysis server 100 may obtain information on essential documents for insurance for which the user has applied for subscription from the external server 20 in order to identify the type of insurance subscription document ( 203 ). Specifically, the document analysis server 100 may obtain information on the types of one or more essential documents required for insurance subscription from the external server 20 based on the type of insurance product for which the user applies for subscription.

Also, the document analysis server 100 may analyze the imaged insurance subscription document through the document analysis model ( 205 ). Specifically, one or more characters and signatures included in the insurance subscription document may be identified using the document analysis model. In addition, the document analysis model may identify one or more characters included in the imaged insurance subscription document using an optical character recognition (OCR) technology. In addition, the document analysis model may perform natural language processing on the identified one or more characters to read the structure of the insurance subscription document. The read structure of the insurance subscription document may be generated as structural information of the insurance subscription document through the document analysis model. The structural information of the insurance subscription document includes information on one or more items included in the insurance subscription document, and the information on the items includes information on the name of each item extracted through context analysis and item-by-item in the insurance subscription document Information about each location may be included.

Also, the document analysis server 100 may read the structure of the imaged insurance subscription document through the structure analysis submodel ( 207 ). Specifically, the structural analysis submodel is learned based on the analysis result of the imaged insurance subscription document. The structural analysis submodel determines whether to select another essential document by processing the comparison result of the structural information included in the selected essential document and the structural information of the imaged insurance subscription document using the policy network, and It can be learned to calculate a reward based on the structure information and the structure information of the imaged insurance subscription document, and select other essential documents that can maximize the reward. The document analysis server 100 may determine the insurance document received from the user terminal 10 based on the essential document finally selected by the structural analysis submodel.

Also, the document analysis server 100 may generate information on the examination result for insurance subscription and document supplement information based on the analysis result of the imaged insurance subscription document ( 209 ). Specifically, the document analysis server 100 may generate an examination result for insurance subscription based on at least one of the type or content of the insurance subscription document. Also, the document analysis server 100 may generate document supplement information for insurance subscription based on the analysis result of the insurance subscription document. The document supplementary information may include information on items missing in the insurance subscription document and information on whether anomalies exist in the items included in the insurance subscription document. Also, the document analysis server 100 may generate underwriting supplementary information based on the recognition result for each item included in the insurance subscription document. Also, the document analysis server 100 may generate conditional approval information based on the recognition result for each item included in the insurance subscription document. Accordingly, the document analysis server 100 may generate an examination result based on the underwriting supplement information and the conditional approval information.

Also, the document analysis server 100 may transmit document supplement information for requesting feedback from the user to the user terminal 10 ( 211 ).

Also, the document analysis server 100 may receive feedback information based on the document supplementation information from the user terminal 10 ( 213 ). Specifically, the feedback information includes information on the item to which the user's correction is performed on the item for which the feedback is requested, and accordingly, the correction is applied. The feedback information may be received by the document analysis server 100 through the user terminal 10 .

Also, the document analysis server 100 may determine insurance subscription based on the feedback information received from the user terminal 10 and the examination result of the insurance subscription document ( 215 ). Specifically, the document analysis server 100 recognizes at least one of the type or content of the insurance subscription document based on the analysis result of the insurance subscription document, and performs insurance subscription based on at least one of the type or content of the insurance subscription document. can decide

Also, the document analysis server 100 may transmit information on the insurance subscriber and information on the insurance subscription to the external server 20 based on the insurance subscription decision ( 217 ).

Also, the document analysis server 100 may transmit information on insurance money and notification information on insurance subscription to the user terminal 10 based on insurance subscription ( 219 ). Specifically, the notification information on insurance subscription may include information on whether to subscribe to the insurance and information on the examination result.

According to an embodiment of the present disclosure, the user can conveniently input a chat message through the chatbot interface to initiate insurance-related procedures, and can conduct insurance-related consultations through chatting with the document analysis server 100, In response to a document request from the document analysis server 100 , a document may be conveniently shared through the chatbot interface. In addition, the user can supplement the contents of insurance-related documents shared through the chatbot interface through chatting with the document analysis server 100, and finally can conveniently subscribe to insurance.

5 is a flowchart illustrating a method of analyzing a document related to insurance subscription according to an embodiment of the present disclosure.

According to an embodiment of the present disclosure, based on the user's chatting record received through the chatbot interface of the user terminal, it may be determined whether to start the insurance-related procedure ( 510 ).

According to an embodiment of the present disclosure, an insurance subscription document inputted through a chatbot interface from a user terminal may be received ( 520 ).

According to an embodiment of the present disclosure, the insurance subscription document may be analyzed using the document analysis model by inputting the insurance subscription document into the document analysis model ( 530 ).

According to an embodiment of the present disclosure, insurance subscription may be determined based on the analysis result of the insurance subscription document ( 540 ).

The order of the steps illustrated in FIG. 5 described above may be changed if necessary, and at least one or more steps may be omitted or added. That is, the above-described steps are merely an embodiment of the present disclosure, and the scope of the present disclosure is not limited thereto.

6 illustrates logic for implementing a document analysis method related to insurance subscription according to an embodiment of the present disclosure.

According to an embodiment of the present disclosure, a computer program for providing a document analysis method related to insurance subscription may be implemented by the following logic.

The computer program includes logic 610 for determining whether to initiate an insurance-related procedure based on the user's chat record received through the chatbot interface of the user terminal, and receiving the insurance subscription document inputted through the chatbot interface from the user terminal logic 620 for inputting the insurance subscription document into the document analysis model, and logic 630 for analyzing the insurance subscription document using the document analysis model, and for determining insurance subscription based on the analysis result for the insurance subscription document logic 640 .

Alternatively, the insurance subscription document may include an image of a document related to insurance subscription.

Alternatively, the logic for receiving the insurance subscription document input through the chatbot interface from the user terminal may include: logic for transmitting insurance guide information to the user terminal through the chatbot interface provided to the user terminal; logic for receiving a user confirmation signal for the user's insurance guide information through a chatbot interface provided to the user terminal; and logic for receiving the insurance subscription document based on the user's upload input for the insurance subscription document through a chatbot interface provided to the user terminal.

Alternatively, the logic for inputting the insurance subscription document into the document analysis model to analyze the insurance subscription document using the document analysis model is logic for identifying one or more characters and a signature included in the insurance subscription document using the document analysis model ; and logic for reading the structure of the insurance subscription document by performing natural language processing on one or more characters using the document analysis model.

Alternatively, the logic for reading the structure of the insurance subscription document by performing natural language processing on one or more characters using the document analysis model may include: logic to identify more than one item; and logic for reading the structure of the insurance subscription document based on the context analysis result for one or more items.

Alternatively, the logic for reading the structure of the insurance subscription document based on the context analysis result for one or more items may include: logic for reading the structure of the insurance subscription document by comparing the context analysis result with an existing insurance subscription document structure; or logic for reading the structure of the insurance subscription document by inputting the context analysis result into the structural analysis submodel; It may include at least one logic of

Alternatively, the structural analysis submodel is trained based on the analysis result of the imaged insurance subscription document, and compares the structural information included in the selected essential document with the structural information of the imaged insurance subscription document using a policy network. process to determine whether to select other required documents, calculate a reward based on the structural information contained in the other essential documents and the structural information of the imaged insurance policy document, and select other essential documents that can maximize the reward can be learned

Alternatively, the logic for determining insurance subscription based on the analysis result of the insurance subscription document may include: logic for recognizing at least one of the type or content of the insurance subscription document based on the analysis result of the insurance subscription document; and logic for determining insurance subscription based on at least one of a type or content of an insurance subscription document.

Alternatively, the logic for determining insurance subscription based on at least one of the type or content of the insurance subscription document includes logic for generating an examination result for insurance subscription based on at least one of the type or content of the insurance subscription document can do.

Alternatively, the logic for generating the examination result for insurance subscription based on at least one of the type or content of the insurance subscription document includes logic for generating underwriting supplementary information based on the recognition result for each item included in the insurance subscription document can do.

Alternatively, the logic for generating the examination result for insurance subscription based on at least one of the type or content of the insurance subscription document includes logic for generating conditional approval information based on the recognition result for each item included in the insurance subscription document can do.

Alternatively, the logic for determining insurance subscription based on the analysis result of the insurance subscription document may include: logic for generating document supplemental information for insurance subscription based on the analysis result of the insurance subscription document; and logic for transmitting supplementary information to the user terminal.

Alternatively, the logic for generating document supplemental information for insurance subscription based on the analysis result of the insurance subscription document may include: logic for determining whether a missing item exists in the insurance subscription document; and logic for determining whether an anomaly exists in an item included in the insurance subscription document.

According to another embodiment of the present disclosure, logic for implementing a computer program may be implemented by a model, circuit or means for implementing the computer program.

Those skilled in the art will further appreciate that the various illustrative logical blocks, configurations, models, circuits, means, logics, and algorithm steps described in connection with the embodiments disclosed herein are compatible with electronic hardware, computer software, or combinations of both. It should be recognized that it can be implemented as To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, configurations, means, logics, models, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

7 depicts a simplified, general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure may be implemented.

Although the present disclosure has been described above generally in the context of computer-executable instructions that may be executed on one or more computers, it will be appreciated by those skilled in the art that the present disclosure may be implemented in combination with other program models and/or as a combination of hardware and software. will be.

Generally, a program model includes routines, programs, components, data structures, etc. that perform particular tasks or implement particular abstract data types. In addition, those skilled in the art will appreciate that the methods of the present disclosure are suitable for single-processor or multiprocessor computer systems, minicomputers, mainframe computers as well as personal computers, handheld computing devices, microprocessor-based or programmable consumer electronics, etc. (each of which is It will be appreciated that other computer system configurations may be implemented, including those that may operate in connection with one or more associated devices.

The described embodiments of the present disclosure may also be practiced in distributed computing environments where certain tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program models may be located in both local and remote memory storage devices.

Computers typically include a variety of computer-readable media. Any medium accessible by a computer can be a computer readable medium, and such computer readable media includes volatile and nonvolatile media, transitory and non-transitory media, removable and non-transitory media. including removable media. By way of example, and not limitation, computer-readable media may include computer-readable storage media and computer-readable transmission media. Computer-readable storage media includes volatile and non-volatile media, transitory and non-transitory media, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program models or other data. includes media. Computer storage media may include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital video disk (DVD) or other optical disk storage device, magnetic cassette, magnetic tape, magnetic disk storage device or other magnetic storage device; or any other medium that can be accessed by a computer and used to store the desired information.

Computer-readable transmission media typically embodies computer-readable instructions, data structures, program models, or other data in a modulated data signal, such as a carrier wave or other transport mechanism, and the like. Includes any information delivery medium. The term modulated data signal means a signal in which one or more of the characteristics of the signal is set or changed so as to encode information in the signal. By way of example, and not limitation, computer-readable transmission media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared, and other wireless media. Combinations of any of the above are also intended to be included within the scope of computer-readable transmission media.

An example environment 1100 implementing various aspects of the disclosure is shown including a computer 1102 , the computer 1102 including a processing unit 1104 , a system memory 1106 , and a system bus 1108 . do. The system bus 1108 couples system components, including but not limited to system memory 1106 , to the processing device 1104 . The processing device 1104 may be any of a variety of commercially available processors. Dual processor and other multiprocessor architectures may also be used as processing unit 1104 .

The system bus 1108 may be any of several types of bus structures that may further interconnect a memory bus, a peripheral bus, and a local bus using any of a variety of commercial bus architectures. System memory 1106 includes read only memory (ROM) 1110 and random access memory (RAM) 1112 . A basic input/output system (BIOS) is stored in non-volatile memory 1110, such as ROM, EPROM, EEPROM, etc., which is the basic input/output system (BIOS) that helps transfer information between components within computer 1102, such as during startup. contains routines. RAM 1112 may also include high-speed RAM, such as static RAM, for caching data.

The computer 1102 also has an internal hard disk drive (HDD) 1114 (eg, EIDE, SATA). The internal hard disk drive 2114 may also be configured for external use within a suitable chassis (not shown). ), a magnetic floppy disk drive (FDD) 1116 (eg, for reading from or writing to removable diskette 1118), and an optical disk drive 1120 (eg, a CD-ROM disk). (for reading 1122 or for reading from or writing to other high capacity optical media such as DVD). The hard disk drive 1114 , the magnetic disk drive 1116 , and the optical disk drive 1120 are connected to the system bus 1108 by the hard disk drive interface 1124 , the magnetic disk drive interface 1126 , and the optical drive interface 1128 , respectively. ) can be connected to The interface 1124 for external drive implementation includes at least one or both of Universal Serial Bus (USB) and IEEE 1394 interface technologies.

These drives and their associated computer-readable media provide non-volatile storage of data, data structures, computer-executable instructions, and the like. In the case of computer 1102, drives and media correspond to storing any data in a suitable digital format. Although the description of computer readable media above refers to HDDs, removable magnetic disks, and removable optical media such as CDs or DVDs, those skilled in the art will use zip drives, magnetic cassettes, flash memory cards, cartridges, etc. It will be appreciated that other tangible computer-readable media such as etc. may also be used in the exemplary operating environment and any such media may include computer-executable instructions for performing the methods of the present disclosure.

A number of program models may be stored in drives and RAM 1112 , including operating system 1130 , one or more application programs 1132 , other program models 1134 , and program data 1136 . All or portions of the operating system, applications, models, and/or data may also be cached in RAM 1112 . It will be appreciated that the present disclosure may be implemented in various commercially available operating systems or combinations of operating systems.

A user may enter commands and information into the computer 1102 via one or more wired/wireless input devices, for example, a pointing device such as a keyboard 1138 and a mouse 1140 . Other input devices (not shown) may include a microphone, IR remote control, joystick, game pad, stylus pen, touch screen, and the like. Although these and other input devices are often connected to the processing unit 1104 through an input device interface 1142 that is connected to the system bus 1108, parallel ports, IEEE 1394 serial ports, game ports, USB ports, IR interfaces, It may be connected by other interfaces, etc.

A monitor 1144 or other type of display device is also coupled to the system bus 1108 via an interface, such as a video adapter 1146 . In addition to the monitor 1144, the computer typically includes other peripheral output devices (not shown), such as speakers, printers, and the like.

Computer 1102 may operate in a networked environment using logical connections to one or more remote computers, such as remote computer(s) 1148 via wired and/or wireless communications. Remote computer(s) 1148 may be workstations, computing device computers, routers, personal computers, portable computers, microprocessor-based entertainment devices, peer devices, or other common network nodes, and are typically connected to computer 1102 . Although it includes many or all of the components described for it, only memory storage device 1150 is shown for simplicity. The logical connections shown include wired/wireless connections to a local area network (LAN) 1152 and/or a larger network, eg, a wide area network (WAN) 1154 . Such LAN and WAN networking environments are common in offices and companies, and facilitate enterprise-wide computer networks, such as intranets, all of which can be connected to a worldwide computer network, for example, the Internet.

When used in a LAN networking environment, the computer 1102 is connected to the local network 1152 through a wired and/or wireless communication network interface or adapter 1156 . Adapter 1156 may facilitate wired or wireless communication to LAN 1152 , which also includes a wireless access point installed therein for communicating with wireless adapter 1156 . When used in a WAN networking environment, the computer 1102 may include a modem 1158, be connected to a communication computing device on the WAN 1154, or establish communications over the WAN 1154, such as over the Internet. have other means. A modem 1158 , which may be internal or external and a wired or wireless device, is coupled to the system bus 1108 via a serial port interface 1142 . In a networked environment, program models described for computer 1102 , or portions thereof, may be stored in remote memory/storage device 1150 . It will be appreciated that the network connections shown are exemplary and other means of establishing a communication link between the computers may be used.

Computer 1102 may be associated with any wireless device or object that is deployed and operates in wireless communication, for example, printers, scanners, desktop and/or portable computers, portable data assistants (PDAs), communications satellites, wireless detectable tags. It operates to communicate with any device or place, and phone. This includes at least Wi-Fi and Bluetooth wireless technologies. Accordingly, the communication may be a predefined structure as in a conventional network or may simply be an ad hoc communication between at least two devices.

Wi-Fi (Wireless Fidelity) makes it possible to connect to the Internet, etc. without a wired connection. Wi-Fi is a wireless technology such as cell phones that allows these devices, eg, computers, to transmit and receive data indoors and outdoors, ie anywhere within range of a base station. Wi-Fi networks use a radio technology called IEEE 802.11 (a, b, g, etc.) to provide secure, reliable, and high-speed wireless connections. Wi-Fi can be used to connect computers to each other, to the Internet, and to wired networks (using IEEE 802.3 or Ethernet). Wi-Fi networks may operate in unlicensed 2.4 and 5 GHz radio bands, for example, at 11 Mbps (802.11a) or 54 Mbps (802.11b) data rates, or in products that include both bands (dual band). have.

One of ordinary skill in the art of this disclosure will understand that information and signals may be represented using any of a variety of different technologies and techniques. For example, data, instructions, instructions, information, signals, bits, symbols and chips that may be referenced in the above description are voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields particles or particles, or any combination thereof.

A person of ordinary skill in the art of the present disclosure will recognize that the various illustrative logical blocks, models, processors, means, circuits and algorithm steps described in connection with the embodiments disclosed herein include electronic hardware, (convenience For this purpose, it will be understood that it may be implemented by various forms of program or design code (referred to herein as "software") or a combination of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, models, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. A person skilled in the art of the present disclosure may implement the described functionality in various ways for each specific application, but such implementation decisions should not be interpreted as a departure from the scope of the present disclosure.

The various embodiments presented herein may be implemented as methods, apparatus, or articles of manufacture using standard programming and/or engineering techniques. The term “article of manufacture” includes a computer program, carrier, or media accessible from any computer-readable device. For example, computer-readable media include magnetic storage devices (eg, hard disks, floppy disks, magnetic strips, etc.), optical disks (eg, CDs, DVDs, etc.), smart cards, and flash memory. devices (eg, EEPROMs, cards, sticks, key drives, etc.). Also, various storage media presented herein include one or more devices and/or other machine-readable media for storing information.

It is to be understood that the specific order or hierarchy of steps in the presented processes is an example of exemplary approaches. Based on design priorities, it is understood that the specific order or hierarchy of steps in the processes may be rearranged within the scope of the present disclosure. The appended method claims present elements of the various steps in a sample order, but are not meant to be limited to the specific order or hierarchy presented.

The description of the presented embodiments is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the scope of the present disclosure. Thus, the present disclosure is not intended to be limited to the embodiments presented herein, but is to be construed in the widest scope consistent with the principles and novel features presented herein.

Claims (15)

A computer program stored in a computer-readable storage medium, wherein the computer program, when executed by one or more processors, performs the following operations for analyzing a document for insurance subscription, the operations comprising:
determining whether to initiate an insurance-related procedure based on the user's chatting record received through the chatbot interface of the user terminal;
receiving an insurance subscription document input through the chatbot interface from the user terminal;
generating structural information of the insurance subscription document using a document analysis model;
determining a required document corresponding to the insurance subscription document by inputting a context analysis result included in the structure information into a reinforcement-learned structure analysis submodel; and
determining insurance subscription based on a result of analyzing the insurance subscription document;
including,
The structural analysis submodel is,
Reinforced learning to receive information about the retrieved essential document as at least part of a state and to determine whether to search for another essential document as at least part of an action,
A computer program stored on a computer-readable storage medium.
The method of claim 1,
The insurance subscription documents are:
Including images of documents related to insurance,
A computer program stored on a computer-readable storage medium.
The method of claim 1,
The operation of receiving the insurance subscription document from the user terminal through the chatbot interface includes:
transmitting insurance guide information to the user terminal through a chatbot interface provided to the user terminal;
receiving a user confirmation signal for the insurance guide information of the user through a chatbot interface provided to the user terminal; and
receiving the insurance subscription document based on a user's upload input for the insurance subscription document through a chatbot interface provided to the user terminal;
comprising,
A computer program stored on a computer-readable storage medium.
The method of claim 1,
The operation of generating the structure information includes:
identifying one or more characters and a signature included in the insurance subscription document using the document analysis model; and
reading the structure of the insurance subscription document by performing natural language processing on the one or more characters using the document analysis model;
comprising,
A computer program stored on a computer-readable storage medium.
5. The method of claim 4,
The operation of reading the structure of the insurance subscription document by performing natural language processing on the one or more characters using the document analysis model includes:
identifying one or more items included in the insurance subscription document based on a result of natural language processing for the insurance subscription document; and
reading the structure of the insurance subscription document based on a context analysis result for the one or more items;
comprising,
A computer program stored on a computer-readable storage medium.
6. The method of claim 5,
Reading the structure of the insurance subscription document based on the context analysis result for the one or more items includes:
reading the structure of the insurance subscription document by comparing the context analysis result with the existing insurance subscription document structure;
comprising,
A computer program stored on a computer-readable storage medium.
The method of claim 1,
The structural analysis submodel is,
It is learned based on the analysis result of the imaged insurance subscription document,
By processing the result of comparing the structural information contained in the selected essential document with the structural information of the imaged insurance subscription document using the policy network, it is determined whether to select another required document, and the structural information contained in the other essential document and the imaged Learn to calculate a reward based on the structural information of the insurance subscription document, and select another essential document that can maximize the reward,
A computer program stored on a computer-readable storage medium.
The method of claim 1,
The operation of determining insurance subscription based on the analysis result of the insurance subscription document includes:
recognizing at least one of a type or content of the insurance subscription document based on an analysis result of the insurance subscription document; and
determining insurance subscription based on at least one of a type or content of the insurance subscription document;
comprising,
A computer program stored on a computer-readable storage medium.
9. The method of claim 8,
The operation of determining insurance subscription based on at least one of a type or content of the insurance subscription document includes:
generating a screening result for insurance subscription based on at least one of a type or content of the insurance subscription document;
comprising,
A computer program stored on a computer-readable storage medium.
10. The method of claim 9,
The operation of generating a review result for insurance subscription based on at least one of a type or content of the insurance subscription document includes:
generating underwriting supplementary information based on a recognition result for each item included in the insurance subscription document;
comprising,
A computer program stored on a computer-readable storage medium.
10. The method of claim 9,
The operation of generating a review result for insurance subscription based on at least one of a type or content of the insurance subscription document includes:
generating conditional approval information based on a recognition result for each item included in the insurance subscription document;
comprising,
A computer program stored on a computer-readable storage medium.
The method of claim 1,
The operation of determining insurance subscription based on the analysis result of the insurance subscription document includes:
generating document supplementary information for insurance subscription based on a result of analyzing the insurance subscription document; and
transmitting the supplementary information to the user terminal;
comprising,
A computer program stored on a computer-readable storage medium.
13. The method of claim 12,
The operation of generating document supplemental information for insurance subscription based on the analysis result of the insurance subscription document includes:
determining whether a missing item exists in the insurance subscription document; and
determining whether an anomaly exists in an item included in the insurance subscription document;
comprising,
A computer program stored on a computer-readable storage medium.
A method for analyzing a document for an insurance subscription performed on one or more processors of a computing device, the method comprising:
determining whether to initiate an insurance-related procedure based on the user's chat record received through the chatbot interface of the user terminal;
receiving an insurance subscription document input through the chatbot interface from the user terminal;
generating structural information of the insurance subscription document using a document analysis model;
determining a required document corresponding to the insurance subscription document by inputting a context analysis result included in the structure information into a reinforcement-learned structure analysis submodel; and
determining insurance subscription based on a result of analyzing the insurance subscription document;
including,
The structural analysis submodel is,
Reinforced learning to receive information about the retrieved essential document as at least part of a state and to determine whether to search for another essential document as at least part of an action,
A method of analyzing a document performed on one or more processors of a computing device.
As a server for document analysis method related to insurance subscription,
a processor including one or more cores;
Memory; and
a network unit for transmitting and receiving data to and from the user terminal;
including,
The processor is
Determine whether to start insurance-related procedures based on the user's chat record received through the chatbot interface of the user terminal,
receiving an insurance subscription document input through the chatbot interface from the user terminal;
generating structural information of the insurance subscription document using a document analysis model;
Input the context analysis result included in the structure information into the reinforcement-learned structure analysis submodel to determine the required document corresponding to the insurance subscription document, and
Decision to purchase insurance based on the analysis result of the insurance subscription document,
The structural analysis submodel is,
Reinforced learning to receive information about the retrieved essential document as at least part of a state and to determine whether to search for another essential document as at least part of an action,
Server for document analysis method.

Images