Video script generation method and device, electronic equipment and storage medium

Abstract

The invention provides a video script generation method and device, electronic equipment and a storage medium, and belongs to the technical field of financial science and technology. The method comprises the following steps: in response to a video script generation request for a target insurance product, acquiring insurance product associated information of the target insurance product; performing feature extraction on the insurance product associated information to obtain key description features of a target insurance product; based on the key description features, script generation is carried out through a script generation model and insurance domain knowledge, and a preliminary video script of the target insurance product is obtained; determining a target optimization strategy in the plurality of candidate optimization strategies based on the insurance product association information; and based on the target optimization strategy, script optimization is performed on the preliminary video script to obtain a target video script, and the target video script is used for generating insurance video content corresponding to the target insurance product. According to the invention, the video script of each insurance video can be quickly generated, so that the video generation efficiency of various insurance products is improved.

Description

Technical Field

[0001] The present application relates to the field of financial technology, and in particular to a method and device for generating a video script, an electronic device and a storage medium. Background Art

[0002] At present, in various business fields, it is often necessary to explain business to various objects so that the objects can fully understand different businesses. For example, in the insurance field, insurance agents often need to explain different insurance products to policyholders, and the insurance agents often rely on the insurance agents' verbal descriptions or the graphic content of the slide show when explaining. However, the verbal description may have the problem of missing or inaccurate information, and the slide show can only present static content, which has the problem of poor explanation richness.

[0003] Based on this, the related technology considers that for each insurance product, the insurance agent produces a corresponding video to assist in explaining each insurance product. However, there are often many types and numbers of insurance products, and this method has the problem of difficulty and time-consuming in video production. Summary of the invention

[0004] The main purpose of the embodiments of the present application is to propose a video script generation method and device, an electronic device and a storage medium, aiming to quickly generate video scripts for various insurance videos, thereby improving the video generation efficiency of various insurance products.

[0005] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes a method for generating a video script, the method comprising:

[0006] In response to a video script generation request for a target insurance product, obtaining insurance product associated information of the target insurance product;

[0007] Extracting features from the insurance product-related information to obtain key descriptive features of the target insurance product;

[0008] Based on the key description features, a script is generated through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product;

[0009] Based on the insurance product association information, determining a target optimization strategy from a plurality of candidate optimization strategies;

[0010] Based on the target optimization strategy, the preliminary video script is optimized to obtain a target video script, which is used to generate insurance video content corresponding to the target insurance product.

[0011] In some embodiments, the script generation based on the key description features is performed through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product, including:

[0012] Based on the script generation model, the key description feature is split into features to obtain multiple descriptor features;

[0013] Based on each of the descriptor features, the insurance domain knowledge is screened to obtain target domain knowledge;

[0014] Extracting features of the target domain knowledge through the script generation model to obtain domain knowledge features;

[0015] Combining the plurality of descriptor features and the domain knowledge features to obtain a plurality of insurance knowledge combination features;

[0016] Based on the multiple insurance knowledge combination features, script generation is performed through the script generation model to obtain the preliminary video script.

[0017] In some embodiments, the insurance product association information includes insurance product description information of the target insurance product and target claims process description information; the feature extraction of the insurance product association information to obtain key description features of the target insurance product includes:

[0018] Extracting features from the insurance product description information to obtain insurance product description features;

[0019] Extracting features of the target claims process description information to obtain claims process description features;

[0020] The insurance product description features and the claims process description features are concatenated to obtain the key description features.

[0021] In some embodiments, the script generation model is trained in the following manner:

[0022] Obtain sample insurance description information of sample insurance products;

[0023] Based on the sample insurance description information, a script is generated for the sample insurance product through a preset large language model to obtain a sample insurance video script;

[0024] Performing script evaluation on the sample insurance video script based on a preset reward function to obtain a reward value of the sample insurance video script;

[0025] Based on the reward values ​​of the plurality of sample insurance video scripts, the large language model is trained to obtain the script generation model.

[0026] In some embodiments, the script optimization of the preliminary video script based on the target optimization strategy to obtain the target video script includes:

[0027] Extracting target style information and target content structure in the target optimization strategy;

[0028] The language style of the preliminary video script is optimized based on the target style information, and the content structure of the preliminary video script is optimized based on the target content structure to obtain the target video script.

[0029] In some embodiments, in response to a video script generation request for a target insurance product, obtaining insurance product associated information of the target insurance product includes:

[0030] Parsing the video script generation request to obtain the insurance product identifier of the target insurance product and the associated claim settlement process identifier;

[0031] Based on the insurance product identifier, screening out the insurance product description information of the target insurance product from a plurality of candidate insurance description information;

[0032] Based on the associated claim process identifier, select target claim process description information associated with the target insurance product from multiple candidate claim process description information;

[0033] The insurance product description information and the target claim settlement process description information are combined into the insurance product association information.

[0034] In some embodiments, the video script generation request is generated in the following manner:

[0035] Get the target login information entered by the target object;

[0036] If it is determined that the target login information has passed the verification, a target interactive interface for generating a video script is displayed;

[0037] The video script generation request is generated based on the insurance product identifier and the claim process identifier input by the target object on the target interaction interface.

[0038] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides a video script generation device, the device comprising:

[0039] an acquisition unit, configured to acquire insurance product associated information of a target insurance product in response to a video script generation request for the target insurance product;

[0040] An extraction unit, used to extract features from the insurance product related information to obtain key descriptive features of the target insurance product;

[0041] A script generation unit, configured to generate a script based on the key description features through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product;

[0042] A determination unit, configured to determine a target optimization strategy from a plurality of candidate optimization strategies based on the insurance product association information;

[0043] A script optimization unit is used to perform script optimization on the preliminary video script based on the target optimization strategy to obtain a target video script, and the target video script is used to generate insurance video content corresponding to the target insurance product.

[0044] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the video script generation method described in the first aspect is implemented.

[0045] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the video script generation method described in the first aspect is implemented.

[0046] The video script generation method and device, electronic device and storage medium proposed in the present application obtain the insurance product related information of the target insurance product by responding to the video script generation request for the target insurance product, and extracting the features of the insurance product related information to obtain the key description features of the target insurance product, which can eliminate the interference of irrelevant features on script generation and improve the accuracy of video script generation using key description features. Then, based on the key description features, the script is generated through the preset script generation model and the pre-acquired insurance field knowledge to obtain the preliminary video script of the target insurance product, which can improve the accuracy of script generation and make the footsteps content of the preliminary video script have better coherence and logic. Further, based on the insurance product related information, the target optimization strategy is determined from multiple candidate optimization strategies, and based on the target optimization strategy, the preliminary video script is optimized to obtain the target video script, which can improve the footsteps quality of the target video script finally generated. This method can quickly generate the target video script of the target insurance video, thereby improving the video generation efficiency of the target insurance product. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1is a flow chart of a method for generating a video script provided in an embodiment of the present application;

[0048] Figure 2 yes Figure 1 Flow chart of step S101 in FIG.

[0049] Figure 3 is a flowchart of generating a video script generation request according to an embodiment of the present application;

[0050] Figure 4 yes Figure 1 Flow chart of step S102 in FIG.

[0051] Figure 5 yes Figure 1 Flow chart of step S103 in FIG.

[0052] Figure 6 yes Figure 1 Flow chart of step S105 in FIG.

[0053] Figure 7 is a flowchart of training a script generation model according to an embodiment of the present application;

[0054] Figure 8 It is a structural schematic diagram of a video script generating device provided in an embodiment of the present application;

[0055] Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0057] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0059] The video script generation method and device, electronic device and storage medium provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the video script generation method in the embodiments of the present application is described.

[0060] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0061] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

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

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

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

[0065] Figure 1 is an optional flowchart of the video script generation method provided in the embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S105.

[0066] Step S101, in response to a video script generation request for a target insurance product, obtaining insurance product associated information of the target insurance product;

[0067] Step S102, extracting features from insurance product related information to obtain key descriptive features of the target insurance product;

[0068] Step S103, based on the key description features, a script is generated through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product;

[0069] Step S104, determining a target optimization strategy from a plurality of candidate optimization strategies based on the insurance product association information;

[0070] Step S105, based on the target optimization strategy, the preliminary video script is optimized to obtain the target video script.

[0071] Steps S101 to S106 shown in the embodiment of the present application, by responding to a video script generation request for a target insurance product, obtain insurance product related information of the target insurance product, and extract features of the insurance product related information to obtain key description features of the target insurance product, which can eliminate the interference of irrelevant features on script generation and improve the accuracy of video script generation using key description features. Then, based on the key description features, script generation is performed through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product, which can improve the accuracy of script generation and make the footsteps content of the preliminary video script have better coherence and logic. Further, based on the insurance product related information, a target optimization strategy is determined from multiple candidate optimization strategies, and based on the target optimization strategy, the preliminary video script is script optimized to obtain the target video script, which can improve the footsteps quality of the target video script finally generated. This method can quickly generate the target video script of the target insurance video, thereby improving the video generation efficiency of the target insurance product.

[0072] In step S101 of some embodiments, in response to a video script generation request for a target insurance product, insurance product associated information of the target insurance product is obtained.

[0073] The target insurance product refers to the specific insurance product that the insurance agent wants to introduce to the insured. The target insurance product can be any insurance product under any type of insurance product category.

[0074] The video script generation request refers to a request initiated by an insurance agent when using a video script to generate an insurance video for a target insurance product.

[0075] The insurance product association information is used to indicate the specific product information of the target insurance product (insurance coverage, insurance terms, and insurance time limit, etc.), as well as the claims settlement process applicable to the target insurance product, etc.

[0076] Among them, the video script of the embodiment of the present application is a text description of the video content of the insurance video corresponding to the insurance product, which includes dialogues, scene descriptions, action instructions, etc.

[0077] See also Figure 2 In some embodiments, step S101 may include but is not limited to steps S201 to S204:

[0078] Step S201, parsing the video script generation request to obtain the insurance product identifier of the target insurance product and the associated claim process identifier;

[0079] Step S202, based on the insurance product identifier, screening out the insurance product description information of the target insurance product from the plurality of candidate insurance description information;

[0080] Step S203, based on the associated claim process identifier, select the target claim process description information associated with the target insurance product from the plurality of candidate claim process description information;

[0081] Step S204, combining the insurance product description information and the target claim settlement process description information into insurance product related information.

[0082] Steps S201 to S204 are described in detail below.

[0083] In step S201 of some embodiments, the insurance product identifier is used to identify the target insurance product; the associated claim process identifier is used to identify the claim process applicable to the target insurance product. The insurance product identifier and the associated claim process identifier are both unique, and different insurance products have different insurance product identifiers, and different claim processes have different claim process identifiers.

[0084] In the specific implementation of this embodiment, first, the video script generation request is parsed, and the code segment in the video script generation request in code form is detected to obtain the insurance product identifier of the target insurance product and the associated claims process identifier, and the specific character string loaded in the code segment is extracted into the insurance product identifier of the target insurance product and the associated claims process identifier.

[0085] In step S202 of some embodiments, the candidate insurance description information is used to indicate the specific product content of each candidate insurance product on the insurance platform. Each candidate insurance product corresponds to one candidate insurance description information.

[0086] In the specific implementation of this embodiment, first, the candidate insurance product identifier of the candidate insurance product corresponding to the candidate insurance description information is extracted from the candidate insurance description information. Then, the insurance product identifier is compared with the candidate insurance product identifiers of each candidate insurance product. Further, the candidate insurance description information corresponding to the candidate insurance product identifier that is consistent with the insurance product identifier is used as the insurance product description information of the target insurance product.

[0087] In step S203 of some embodiments, the candidate claim settlement process description information is used to indicate the specific process content of each candidate claim settlement process on the insurance platform. Each candidate claim settlement process corresponds to a candidate claim settlement process description information.

[0088] In the specific implementation of this embodiment, first, the candidate claim process identifier of the candidate claim process corresponding to the candidate claim process description information is extracted from the candidate claim process description information. Then, the associated claim process identifier is compared with each candidate claim process identifier. Further, the candidate claim process description information corresponding to the candidate claim process identifier that is consistent with the associated claim process identifier is used as the target claim process description information associated with the target insurance product.

[0089] In step S204 of some embodiments, the insurance product description information and the target claims process description information are integrated into the same document to achieve a combination of the insurance product description information and the target claims process description information, and the specific content of the integrated document is determined as insurance product related information.

[0090] The benefit of this embodiment is that, according to the insurance product identifier and the associated claim process identifier extracted in the video script generation request, the insurance product description information corresponding to the target insurance product is found in the multiple candidate insurance description information, and the target claim process description information associated with the target insurance product is found in the multiple candidate claim process description information, which can improve the accuracy and efficiency of the description information screening. Finally, the insurance product description information and the target claim process description information are assembled into insurance product associated information, which can improve the comprehensiveness of the insurance product associated information.

[0091] See also Figure 3 In some embodiments, the specific process of generating a video script request may include but is not limited to steps S301 to S303:

[0092] Step S301, obtaining target login information input by the target object;

[0093] Step S302, if it is determined that the target login information verification has passed, then displaying a target interactive interface for generating a video script;

[0094] Step S303, generating a video script generation request based on the insurance product identifier and the claim process identifier input by the target object on the target interactive interface.

[0095] Steps S301 to S303 are described in detail below.

[0096] In step S301 of some embodiments, the target object refers to a user who needs to make a video script, for example, the target object may be an insurance agent. The target login information is used to indicate the login name and login password entered by the target object on the login interface of the target platform when the target object logs in to the target platform for generating the video script on the object terminal.

[0097] In the specific implementation of this embodiment, when the target object wants to make a video script, the target object will open the target platform for generating the video script on the object terminal. At this time, the login interface of the target platform for generating the video script will be displayed on the object terminal, wherein the login interface has an input area for inputting a login name and an input area for inputting a login password. Based on this, the target object will enter a specific login user name in the input area for inputting a login name, enter a specific login key in the input area for inputting a login password, and confirm access to the target platform based on the input login user name and login key. At this time, the server will detect the input of the specific login user name and the specific login key, and assemble the login user name and the login key into the target login information.

[0098] Furthermore, the server will search the registration information of the registered users according to the login user name, find the registration information pointed to by the login user name, and extract the login password set by the user when registering from the registration information. Then, the login key entered by the target object is compared with the set login password. If the login key entered by the target object is consistent with the set login password, it is determined that the target login information verification has passed. If the login key entered by the target object is inconsistent with the set login password, it is determined that the target login information verification has failed, and a login failure prompt is displayed to the target object.

[0099] In step S302 of some embodiments, if it is determined that the target login information verification has passed, a target interaction interface for generating a video script is displayed on the target object's object terminal. The target interaction interface has an input area for inputting an insurance product identifier of an insurance product for which an insurance video script is to be generated, and an input area for inputting a claim process identifier of a claim process applicable to the insurance product for which an insurance video is to be generated.

[0100] In step S303 of some embodiments, the target object will enter the insurance product identifier corresponding to the target insurance product in the input area for entering the insurance product identifier, enter the specific claim process identifier associated with the target insurance product in the input area for entering the claim process identifier, and confirm that for the insurance product with the insurance product identifier, an insurance video script is produced according to the claim process corresponding to the claim process identifier. Based on this, the server will receive the insurance product identifier and claim process identifier entered by the target object on the target interactive interface, and assemble the insurance product identifier and claim process identifier entered by the target object on the target interactive interface into a video script generation request according to the format required by the preset transmission protocol.

[0101] The benefit of this embodiment is that it provides a target interactive interface for generating a video script for the target object. After the target object's login information is verified, the target object can set specific parameters corresponding to the insurance product for which the video script is to be generated on the target interactive interface, thereby enabling the customized setting of the insurance video script to meet the personalized needs of different objects in different scenarios and improve the flexibility of generating the insurance video script. At the same time, the customized setting of the embodiment of the present application is implemented in a visual form, which can improve the transparency of the generation of the insurance video script.

[0102] In step S102 of some embodiments, feature extraction is performed on insurance product related information to obtain key descriptive features of the target insurance product.

[0103] Key descriptive features are used to indicate important descriptive information related to the target insurance product.

[0104] See also Figure 4 In some embodiments, step S102 may include but is not limited to steps S401 to S403:

[0105] Step S401, extracting features from the insurance product description information to obtain insurance product description features;

[0106] Step S402, extracting features from the target claim process description information to obtain claim process description features;

[0107] Step S403, concatenating the insurance product description features and the claims process description features to obtain key description features.

[0108] Steps S401 to S403 are described in detail below.

[0109] In step S401 of some embodiments, the insurance product description feature is used to indicate a feature representation of the insurance product description information in a latent vector space.

[0110] In the specific implementation of this embodiment, first, the insurance product description information is mapped from the data space to the latent vector space to obtain the insurance product description embedding vector. Then, the insurance product description embedding vector is convoluted to extract the key description feature information in the insurance product description information to obtain the insurance product description feature.

[0111] In step S402 of some embodiments, the claim process description feature is used to indicate the feature representation of the target claim process description information in the latent vector space.

[0112] In the specific implementation of this embodiment, first, the target claim process description information is mapped from the data space to the latent vector space to obtain the claim process description embedding vector. Then, the claim process description embedding vector is convoluted to extract the key description feature information in the target claim process description information to obtain the claim process description feature.

[0113] In step S403 of some embodiments, when the insurance product description features and the claims process description features are concatenated, the insurance product description features and the claims process description features are first mapped to a feature space of the same vector dimension, so as to obtain the insurance product description features and the claims process description features with the same vector dimension. Then, the insurance product description features and the claims process description features with the same vector dimension are concatenated into a vector with a larger feature length to obtain the key description features.

[0114] The benefit of this embodiment is that feature extraction is performed on the feature representations of the insurance product description information and the target claims process description information in the latent vector space respectively, to obtain the insurance product description features and the claims process description features, and the two description features representing different feature information are spliced ​​at the vector level, so as to generate key description features with richer feature information, improve the comprehensiveness of the key description features, and thus improve the accuracy of video script generation using the key description features.

[0115] In step S103 of some embodiments, based on the key description features, script generation is performed through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product.

[0116] The script generation model refers to a neural network model that can generate the video script of the insurance video corresponding to each insurance product based on the relevant text description of each insurance product. The input of the script generation model is generally the vector representation of the relevant text description of each insurance product, and the output is the video script for generating the insurance video of the insurance product.

[0117] Insurance domain knowledge is used to refer to general knowledge in the insurance field, such as basic insurance knowledge (such as the principles and types of insurance, etc.), insurance products, insurance clauses, and insurance-related laws and regulations, etc.

[0118] The preliminary video script refers to the script data of the video content of the insurance video that can describe the insurance product, which is predicted by the script generation model based on the vector representation of the relevant text description of the insurance video.

[0119] See also Figure 5 In some embodiments, step S103 may include but is not limited to steps S501 to S505:

[0120] Step S501, performing feature splitting on the key description feature based on the script generation model to obtain multiple descriptor features;

[0121] Step S502, screening insurance domain knowledge based on each descriptor feature to obtain target domain knowledge;

[0122] Step S503, extracting features of the target domain knowledge through the script generation model to obtain domain knowledge features;

[0123] Step S504, combining multiple descriptor features and domain knowledge features to obtain multiple insurance knowledge combination features;

[0124] Step S505, based on multiple insurance knowledge combination features, a script is generated through a script generation model to obtain a preliminary video script.

[0125] Steps S501 to S505 are described in detail below.

[0126] In step S501 of some embodiments, since the key description feature is a feature sequence composed of multiple feature elements. Based on this, when the key description feature is split based on the script generation model, first, the feature similarity is calculated for the feature elements adjacent to each other in the key description feature to obtain the cosine similarity of the feature elements adjacent to each other; according to the cosine similarity of the feature elements adjacent to each other, a segmentation mark is set between the adjacent feature elements whose cosine similarity is less than a preset threshold. Then, according to the segmentation mark, the key description feature is segmented at the position where the segmentation mark is set, and the key description feature is divided into multiple descriptor features. Among them, each descriptor feature is used to represent a specific part of the key description information.

[0127] In step S502 of some embodiments, for each insurance domain knowledge, the insurance domain knowledge is first mapped from the data space to the latent vector space to obtain an insurance domain knowledge embedding vector. Next, feature similarity is calculated for a single descriptor feature and each insurance domain knowledge embedding vector to obtain feature similarity, and the insurance domain knowledge defect corresponding to the insurance domain knowledge embedding vector with the largest feature similarity is taken as the target domain knowledge corresponding to the single descriptor feature.

[0128] In step S503 of some embodiments, the target domain knowledge is convolved by using a script generation model to extract associated knowledge information in the target domain knowledge and obtain domain knowledge features.

[0129] In step S504 of some embodiments, for each descriptor feature, the descriptor feature and the domain knowledge feature corresponding to the descriptor feature are concatenated, and the concatenated feature vector is used as an insurance knowledge combination feature.

[0130] In step S505 of some embodiments, first, multiple insurance knowledge combination features are spliced ​​to obtain a target insurance knowledge feature. Then, the script generation model is used to perform attention calculation on the target insurance knowledge feature, and the target insurance knowledge feature is linearly projected on the query channel, the key channel, and the value channel to obtain a query vector output by the query channel, a key vector output by the key channel, and a value vector output by the value channel, respectively. Then, a dot product operation is performed on the key vector and the query vector to obtain an attention score, and the square root of the feature dimension of the key vector is used as a scaling factor, and the attention score is divided by the scaling factor to achieve a scaling operation on the attention score to obtain a scaled attention score. Further, the scaled attention score is normalized using a softmax function to obtain an attention weight, and the value vector is weighted using the attention weight to obtain an attention calculation result. Further, the attention calculation result is converted into a sequence form based on the script generation model to obtain a preliminary video script.

[0131] The benefit of this embodiment is that, according to the correlation between each feature element in the key descriptive feature, highly correlated feature elements are split into a group, and unrelated feature elements are split into different groups, which can effectively reduce the redundancy between feature elements. Furthermore, for each descriptor feature, the domain knowledge feature corresponding to the descriptor feature is screened out in the target domain knowledge, and the domain knowledge feature is combined with the descriptor feature, so that the domain knowledge feature with a mapping relationship can be associated with the descriptor feature, thereby improving the correlation of the feature information of the insurance knowledge combination feature. Finally, the script generation model is used to generate scripts based on multiple insurance knowledge combination features, which can improve the accuracy of script generation and make the footsteps content of the preliminary video script have better coherence and logic.

[0132] In step S104 of some embodiments, a target optimization strategy is determined from a plurality of candidate optimization strategies based on insurance product association information.

[0133] The candidate optimization strategy is used to indicate the specific optimization operation to be performed on the video script data of the insurance product. The candidate optimization strategy includes the optimization type and specific optimization method of the video script data of the insurance product.

[0134] When this embodiment is implemented, first, the product type information of the target insurance product is extracted from the insurance product association information. Then, multiple candidate optimization strategies are screened according to the product type information, and the candidate optimization strategy that matches the product type information is determined as the target optimization strategy. Among them, each candidate optimization strategy is pre-set with the insurance product type to which the candidate optimization strategy is applicable. When screening the candidate optimization strategies, the candidate optimization strategy whose set insurance product type is consistent with the product type information is determined as the target optimization strategy.

[0135] In step S105 of some embodiments, based on a target optimization strategy, the preliminary video script is optimized to obtain a target video script.

[0136] The target video script is used to generate insurance video content corresponding to the target insurance product.

[0137] See also Figure 6 In some embodiments, step S105 may include but is not limited to steps S601 to S602:

[0138] Step S601, extracting target style information and target content structure in the target optimization strategy;

[0139] Step S602, optimizing the language style of the preliminary video script based on the target style information, and optimizing the content structure of the preliminary video script based on the target content structure, to obtain the target video script.

[0140] Steps S601 to S602 are described in detail below.

[0141] In step S601 of some embodiments, the target style information is used to indicate the language style requirements that the video script of the target insurance product should meet, and the target content structure is used to indicate the content structure requirements that the video script of the target insurance product should meet.

[0142] When this embodiment is implemented, the target optimization strategy can be extracted by keyword search. The language style and content structure are used as keywords to traverse and search the policy content of the target optimization strategy, extract the policy content related to the language style as the target style information, and extract the policy content related to the content structure as the target content structure.

[0143] In step S602 of some embodiments, when optimizing the language style of the preliminary video script based on the target style information, the tone and vocabulary in the preliminary video script are adjusted according to the target style information, so that the tone and vocabulary in the adjusted preliminary video script meet the language style requirements in the target style information. At the same time, when optimizing the content structure of the preliminary video script based on the target content structure, the footsteps content of the preliminary video script is reorganized according to the target content structure, so that the information of the preliminary video script has better logic and fluency, and the key information points of the preliminary video script are highlighted, and the rhythm and duration of each part of the preliminary video script are adjusted according to the total duration of the insurance video of the target insurance product to be generated. Finally, the preliminary video script that has been optimized in style and content structure is determined as the target video script.

[0144] The benefit of this embodiment is that when optimizing the script, the language style and content structure of the preliminary video script are optimized, so that the language style and expression of the preliminary video script can be modified, and the content of the preliminary video script can be reorganized and key information can be strengthened, and the rhythm and duration of the entire script can be adjusted, thereby improving the pace quality of the target video script and making the target video script more in line with the real needs of the target object.

[0145] See also Figure 7 In some embodiments, the specific process of training the script generation model may include but is not limited to steps S701 to S704:

[0146] Step S701, obtaining sample insurance description information of a sample insurance product;

[0147] Step S702, based on the sample insurance description information, a script is generated for the sample insurance product through a preset large language model to obtain a sample insurance video script;

[0148] Step S703, performing script evaluation on the sample insurance video script based on a preset reward function to obtain a reward value of the sample insurance video script;

[0149] Step S704: Based on the reward values ​​of multiple sample insurance video scripts, a large language model is trained to obtain a script generation model.

[0150] Steps S701 to S704 are described in detail below.

[0151] In step S701 of some embodiments, the sample insurance description information is used to indicate the specific product information of the sample insurance product (insurance coverage, insurance terms, and insurance period, etc.), as well as the claims process applicable to the sample insurance product, etc.

[0152] When this embodiment is implemented, the specific process of step S701 is similar to the above step S101, and will not be described in detail to save space.

[0153] In step S702 of some embodiments, the preset large language model refers to a ChatGPT model used to automatically generate insurance video scripts; the sample insurance video script is used to indicate a text description generated for a sample insurance product and used to describe the video content of the insurance video corresponding to the sample insurance product.

[0154] When this embodiment is implemented, the specific process of step S702 is similar to the above steps S401 to S403 and S501 to S505, which will not be described in detail to save space.

[0155] In step S703 of some embodiments, a preset reward function is used to generate a reward for a sample insurance video script generated by a large language model based on sample insurance description information in the current environment, and the reward value is used to indicate the accuracy of the sample insurance video script generated in the current environment.

[0156] When this embodiment is implemented, first, the current environment parameters are obtained, where the current environment parameters are used to indicate the speed at which the large language model outputs the sample insurance video script in the current iteration round. Then, the current environment parameters and the sample insurance video script are input into the reward function together, and the reward function performs script evaluation on the sample insurance video script and outputs the reward value corresponding to the sample insurance video script.

[0157] In step S704 of some embodiments, first, the reward values ​​of multiple sample insurance video scripts are summed to obtain a total reward value. Then, the large language model is trained according to the total reward value, so that the total reward value is maximized as the training target, the model parameters of the large language model are adjusted, and the above steps S701 to S704 are repeated to implement iterative training of the large language model, and the model parameters that maximize the total reward value are used as the final model parameters, and the large language model with the final model parameters is used as the script generation model obtained by training.

[0158] The benefit of this embodiment is that, based on the reinforcement learning method, for sample insurance products, a preset reward function is used to evaluate the sample insurance video scripts predicted by the large language model, and a corresponding reward value is generated, and a total reward value is constructed based on the reward values ​​of multiple sample insurance video scripts. This method can train a script generation model by maximizing the total reward value, which is beneficial to improving the model training effect, and further improving the model's prediction accuracy for generating insurance video scripts corresponding to each insurance product.

[0159] See also Figure 8The embodiment of the present application also provides a video script generation device, which can implement the above-mentioned video script generation method, and the device includes:

[0160] An acquisition unit 801 is configured to acquire insurance product associated information of a target insurance product in response to a video script generation request for a target insurance product;

[0161] An extraction unit 802 is used to extract features from insurance product related information to obtain key descriptive features of a target insurance product;

[0162] The script generation unit 803 is used to generate a script based on the key description features, through a preset script generation model, and pre-acquired insurance field knowledge, to obtain a preliminary video script of the target insurance product;

[0163] A determination unit 804, configured to determine a target optimization strategy from a plurality of candidate optimization strategies based on the insurance product association information;

[0164] The script optimization unit 805 is used to perform script optimization on the preliminary video script based on the target optimization strategy to obtain a target video script, and the target video script is used to generate insurance video content corresponding to the target insurance product.

[0165] In some embodiments, the acquisition unit 801 is used to:

[0166] Parse the video script generation request to obtain the insurance product identifier of the target insurance product and the associated claim process identifier;

[0167] Based on the insurance product identifier, screening out the insurance product description information of the target insurance product from the plurality of candidate insurance description information;

[0168] Based on the associated claim process identifier, select target claim process description information associated with the target insurance product from multiple candidate claim process description information;

[0169] The insurance product description information and the target claims process description information are combined into insurance product related information.

[0170] In some embodiments, the extraction unit 802 is used to:

[0171] Extract features from the insurance product description information to obtain insurance product description features;

[0172] Extract features of the target claims process description information to obtain the claims process description features;

[0173] The insurance product description features and the claims process description features are concatenated to obtain the key description features.

[0174] In some embodiments, the script generation unit 803 is used to:

[0175] Based on the script generation model, the key description features are split to obtain multiple descriptor features;

[0176] Based on the features of each descriptor, the insurance domain knowledge is screened to obtain the target domain knowledge;

[0177] The target domain knowledge is extracted through the script generation model to obtain domain knowledge features;

[0178] Combining multiple descriptor features and domain knowledge features to obtain multiple insurance knowledge combination features;

[0179] Based on the combined features of multiple insurance knowledge, the script is generated through the script generation model to obtain a preliminary video script.

[0180] In some embodiments, the script optimization unit 805 is used to:

[0181] Extract target style information and target content structure in the target optimization strategy;

[0182] The language style of the preliminary video script is optimized based on the target style information, and the content structure of the preliminary video script is optimized based on the target content structure to obtain the target video script.

[0183] In some embodiments, the script generation model is trained in the following manner:

[0184] Obtain sample insurance description information of sample insurance products;

[0185] Based on the sample insurance description information, a script is generated for the sample insurance product through a preset large language model to obtain a sample insurance video script;

[0186] Perform script evaluation on the sample insurance video script based on a preset reward function to obtain a reward value of the sample insurance video script;

[0187] Based on the reward values ​​of multiple sample insurance video scripts, a large language model is trained to obtain a script generation model.

[0188] In some embodiments, the video script generation request is generated by:

[0189] Get the target login information entered by the target object;

[0190] If it is determined that the target login information verification has passed, then the target interactive interface for generating the video script is displayed;

[0191] A video script generation request is generated based on the insurance product identifier and the claim process identifier input by the target object on the target interactive interface.

[0192] The specific implementation of the video script generation device is basically the same as the specific implementation of the above-mentioned video script generation method, and will not be repeated here.

[0193] The embodiment of the present application also provides an electronic device, the electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program is executed by the processor to realize the above-mentioned video script generation method. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0194] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0195] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0196] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the video script generation method of the embodiment of this application;

[0197] Input / output interface 903, used to implement information input and output;

[0198] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WI FI, Bluetooth, etc.);

[0199] A bus 905 that transmits information between the various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);

[0200] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0201] An embodiment of the present application also provides a computer-readable storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the above-mentioned video script generation method.

[0202] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0203] The video script generation method, video script generation device, electronic device and computer-readable storage medium provided in the embodiment of the present application obtain insurance product-related information of the target insurance product in response to a video script generation request for the target insurance product, and extract features of the insurance product-related information to obtain the key description features of the target insurance product, which can eliminate the interference of irrelevant features on script generation and improve the accuracy of video script generation using key description features. Then, based on the key description features, script generation is performed through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product, which can improve the accuracy of script generation and make the footsteps content of the preliminary video script have better coherence and logic. Further, based on the insurance product-related information, a target optimization strategy is determined from multiple candidate optimization strategies, and based on the target optimization strategy, the preliminary video script is script optimized to obtain the target video script, which can improve the footsteps quality of the target video script finally generated. This method can quickly generate the target video script of the target insurance video, thereby improving the video generation efficiency of the target insurance product.

[0204] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0205] It can be understood by those skilled in the art that Figure 1-7 The technical solutions shown in the figure do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than those shown in the figure, or a combination of certain steps, or different steps.

[0206] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0207] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0208] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0209] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0210] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0211] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0212] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0213] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0214] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. A video script generation method, characterized in that: The video script generation method comprises: In response to a video script generation request for a target insurance product, obtaining insurance product associated information of the target insurance product; Extracting features from the insurance product-related information to obtain key descriptive features of the target insurance product; Based on the key description features, a script is generated through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product; Based on the insurance product association information, determining a target optimization strategy from a plurality of candidate optimization strategies; Based on the target optimization strategy, the preliminary video script is optimized to obtain a target video script, and the target video script is used to generate insurance video content corresponding to the target insurance product.

2. The method according to claim 1, characterized in that The script generation based on the key description features is performed through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product, including: Based on the script generation model, the key description feature is split into features to obtain multiple descriptor features; Based on each of the descriptor features, the insurance domain knowledge is screened to obtain target domain knowledge; Extracting features of the target domain knowledge through the script generation model to obtain domain knowledge features; Combining the plurality of descriptor features and the domain knowledge features to obtain a plurality of insurance knowledge combination features; Based on the multiple insurance knowledge combination features, script generation is performed through the script generation model to obtain the preliminary video script.

3. The method according to claim 1, characterized in that The insurance product association information includes insurance product description information of the target insurance product and target claims process description information; the feature extraction of the insurance product association information to obtain key description features of the target insurance product includes: Extracting features from the insurance product description information to obtain insurance product description features; Extracting features of the target claims process description information to obtain claims process description features; The insurance product description features and the claims process description features are concatenated to obtain the key description features.

4. The method according to claim 1, characterized in that: The script generation model is trained in the following way: Obtain sample insurance description information of sample insurance products; Based on the sample insurance description information, a script is generated for the sample insurance product through a preset large language model to obtain a sample insurance video script; Performing script evaluation on the sample insurance video script based on a preset reward function to obtain a reward value of the sample insurance video script; Based on the reward values ​​of the plurality of sample insurance video scripts, the large language model is trained to obtain the script generation model.

5. The method according to claim 1, characterized in that The step of optimizing the preliminary video script based on the target optimization strategy to obtain the target video script includes: Extracting target style information and target content structure in the target optimization strategy; The language style of the preliminary video script is optimized based on the target style information, and the content structure of the preliminary video script is optimized based on the target content structure to obtain the target video script.

6. The method according to claim 1, characterized in that The step of obtaining insurance product-related information of the target insurance product in response to a video script generation request for the target insurance product includes: Parsing the video script generation request to obtain the insurance product identifier of the target insurance product and the associated claim settlement process identifier; Based on the insurance product identifier, screening out the insurance product description information of the target insurance product from a plurality of candidate insurance description information; Based on the associated claim process identifier, select target claim process description information associated with the target insurance product from multiple candidate claim process description information; The insurance product description information and the target claim settlement process description information are combined into the insurance product association information.

7. The method according to any one of claims 1 to 6, characterized in that: The video script generation request is generated in the following way: Get the target login information entered by the target object; If it is determined that the target login information has passed the verification, a target interactive interface for generating a video script is displayed; The video script generation request is generated based on the insurance product identifier and the claim process identifier input by the target object on the target interaction interface.

8. A video script generating device, characterized in that: The video script generating device comprises: an acquisition unit, configured to acquire insurance product associated information of a target insurance product in response to a video script generation request for the target insurance product; An extraction unit, used to extract features from the insurance product related information to obtain key descriptive features of the target insurance product; A script generation unit, configured to generate a script based on the key description features through a preset script generation model and pre-acquired insurance field knowledge to obtain a preliminary video script of the target insurance product; A determination unit, configured to determine a target optimization strategy from a plurality of candidate optimization strategies based on the insurance product association information; A script optimization unit is used to perform script optimization on the preliminary video script based on the target optimization strategy to obtain a target video script, and the target video script is used to generate insurance video content corresponding to the target insurance product.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the video script generation method described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the video script generation method described in any one of claims 1 to 7 is implemented.

Images