Insurance product personalized recommendation method and device, equipment and storage medium

Abstract

The invention relates to the technical field of natural language processing and financial science and technology, and particularly discloses a personalized recommendation method and device for insurance products, equipment and a storage medium. Obtaining a first dialogue voice between the current customer and the business personnel and first product purchase data of the current customer, and performing text conversion to obtain a first dialogue text; analyzing the first dialogue text and the first product purchase data to obtain a full life cycle label of the customer; and based on the full life cycle label, determining an optimal communication verbal skill and a target recommended product, and obtaining a recommended verbal skill corresponding to the target recommended product. According to the method and the device, the full-life-cycle label of the customer can be generated according to the conversation voice between the customer and the business personnel and the product purchase data of the customer, the optimal communication verbal skill matched with the current customer is determined according to the label, and the recommendation verbal skill for recommending the target recommendation product is generated in combination with the target recommendation product; and the accuracy of the target recommended product and the individuation degree of product recommendation are improved.

Description

Technical Field

[0001] The present application relates to the fields of natural language processing and financial technology, and in particular to a method, apparatus, computer equipment and storage medium for personalized recommendation of insurance products. Background Art

[0002] With the development of computer technology, insurance recommendation systems are gradually applied to insurance companies to accurately recommend various insurance products to customers. However, current insurance recommendations are only based on the customer's historical purchase data, but the historical purchase data cannot infer the type of insurance that the customer wants to buy at the current stage, resulting in the recommended insurance products failing to meet user expectations. In addition, due to the uneven skills of service personnel and the rapid development of the insurance industry, the update and iteration cycle of insurance products is getting shorter and shorter, and new products are emerging in an endless stream. It is impossible to accurately understand the most suitable insurance products for users at the current stage based only on customer preferences. At the same time, without understanding the customer's preferences, business personnel blindly recommend products, which can easily cause customers to resist, resulting in poor customer experience and low product purchase rate. Therefore, how to achieve accurate and personalized product recommendations, thereby improving customer experience and product purchase rate has become an urgent problem to be solved. Summary of the invention

[0003] The present application provides a personalized recommendation method, apparatus, computer device and storage medium for insurance products to achieve accurate and personalized product recommendations, thereby improving customer experience and product purchase rate.

[0004] In a first aspect, the present application provides a method for personalized recommendation of insurance products, the method comprising:

[0005] Acquire a first conversation voice between a current customer and a business person and first product purchase data of the current customer, and convert the first conversation voice into text to obtain a first conversation text;

[0006] Analyze the first conversation text and the first product purchase data to obtain a full life cycle label of the current customer;

[0007] Based on the full life cycle label of the current customer, determine the best communication words and target recommended products corresponding to the current customer;

[0008] Based on the optimal communication words and the target recommended product, the recommended words corresponding to the target recommended product are obtained for reference by business personnel.

[0009] In a second aspect, the present application further provides a personalized recommendation device for insurance products, the device comprising:

[0010] A data acquisition module, used to acquire a first conversation voice between a current customer and a business person and first product purchase data of the current customer, and to convert the first conversation voice into text to obtain a first conversation text;

[0011] A label acquisition module, used for analyzing the first conversation text and the first product purchase data to obtain a full life cycle label of the current customer;

[0012] A speech and product determination module, used to determine the best communication speech and target recommended product corresponding to the current customer based on the full life cycle label of the current customer;

[0013] The recommended speech acquisition module is used to obtain the recommended speech corresponding to the target recommended product based on the optimal communication speech and the target recommended product for reference by business personnel.

[0014] In a third aspect, the present application also provides a computer device, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the personalized recommendation method for insurance products as described above when executing the computer program.

[0015] In a fourth aspect, the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to implement the personalized recommendation method for insurance products as described above.

[0016] The present application discloses a personalized recommendation method, device, computer equipment and storage medium for insurance products, which obtains the first conversation voice between the current customer and the business personnel and the first product purchase data of the current customer, and converts the first conversation voice into text to obtain the first conversation text; analyzes the first conversation text and the first product purchase data to obtain the full life cycle label of the current customer; based on the full life cycle label of the current customer, determines the best communication speech and target recommended product corresponding to the current customer; based on the best communication speech and the target recommended product, obtains the recommended speech corresponding to the target recommended product for reference by the business personnel. The present application can generate the customer's full life cycle label based on the conversation voice between the customer and the business personnel and the customer's product purchase data, determine the best communication speech adapted to the current customer based on the label, and generate the recommended speech for recommending the target recommended product in combination with the target recommended product, thereby improving the accuracy of the target recommended product and the personalization of the product recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 This is a schematic flow chart of a first embodiment of a personalized recommendation method for insurance products provided in an embodiment of the present application;

[0019] Figure 2 is a schematic flow chart of a second embodiment of a personalized recommendation method for insurance products provided in an embodiment of the present application;

[0020] Figure 3 is a schematic flow chart of a third embodiment of a personalized recommendation method for insurance products provided in an embodiment of the present application;

[0021] Figure 4 A schematic block diagram of a personalized recommendation device for insurance products provided in an embodiment of the present application;

[0022] Figure 5 A schematic block diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0024] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0025] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0026] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0027] The embodiments of the present application provide a personalized recommendation method, device, computer equipment and storage medium for insurance products. The personalized recommendation method for insurance products can be applied to a server, generating a customer's full life cycle label based on the conversation voice between the customer and the business personnel and the customer's product purchase data, determining the best communication words adapted to the current customer based on the label, and generating a recommendation word for the target recommended product in combination with the target recommended product, thereby improving the accuracy of the target recommended product and the personalization of the product recommendation. The server can be an independent server or a server cluster.

[0028] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0029] See also Figure 1 , Figure 1 This is a schematic flow chart of a personalized recommendation method for insurance products provided by an embodiment of the present application. The personalized recommendation method for insurance products can be applied to a server, and a customer's full life cycle label is generated based on the conversation voice between the customer and the business personnel and the customer's product purchase data. The best communication words adapted to the current customer are determined based on the label, and a recommendation word is generated to recommend the target recommended product in combination with the target recommended product, thereby improving the accuracy of the target recommended product and the personalization of the product recommendation.

[0030] like Figure 1 As shown, the personalized recommendation method for insurance products specifically includes steps S101 to S104.

[0031] S101, obtaining a first conversation voice between a current customer and a business person and first product purchase data of the current customer, and performing text conversion on the first conversation voice to obtain a first conversation text;

[0032] In one embodiment, the first conversation voice may be the voice of a salesperson conducting telephone sales or handling business with a current customer, or may be the voice of a conversation between a current customer and a salesperson on site.

[0033] In one embodiment, the first product purchase data is the products purchased by the current customer and the product information.

[0034] In one embodiment, the first conversation speech may be converted into the first conversation text by using speech-to-text technology such as ASR (Automatic Speech Recognition).

[0035] S102: Analyze the first conversation text and the first product purchase data to obtain a full life cycle label of the current customer;

[0036] In one embodiment, natural language processing technologies such as word segmentation, part-of-speech tagging, and named entity recognition are applied to the first conversation text to extract keywords and entities. Features such as purchase preferences and consumption habits are extracted from the first purchase data. Based on the keywords and entities, combined with the purchase habits and consumption habits obtained from the first product purchase data, the current customer's interests, preferences, purchase history, behavior patterns, and other labels are analyzed and annotated.

[0037] In one embodiment, the analysis results are integrated and summarized to obtain the full life cycle label of the current customer. For example, assuming that the customer Mr. Li is a 40-year-old middle-level manager of a company with a stable income and family, has previously purchased health insurance, and has mentioned topics such as health and medical care many times in conversations with business personnel, then the full life cycle label of the customer can be "health-conscious", "middle-aged", "high income", etc.

[0038] In one embodiment, the full life cycle tags can be further subdivided and classified according to actual needs, such as by age group, gender, region, etc., so as to make more accurate personalized recommendations.

[0039] S103, determining the best communication words and target recommended products corresponding to the current customer based on the full life cycle label of the current customer;

[0040] In one embodiment, based on the current customer's full life cycle label, the communication scripts that match the customer's communication preferences are filtered out from a preset communication script library. At the same time, based on the customer's purchase history and preferences, insurance products that meet the customer's needs are filtered out from the insurance product library. For example, if the customer has purchased health insurance multiple times or mentioned health issues during communication, some new health insurance products can be recommended.

[0041] In one embodiment, a comprehensive matching evaluation is performed on the screened communication words and products, and the communication words and products with the highest comprehensive matching degree are selected as the target communication words and target recommended products.

[0042] In one embodiment, if there are multiple communication scripts and products with the same comprehensive matching degree, further screening can be performed based on other factors (such as customer preference, business personnel experience, etc.) to ultimately determine a target communication script and a target recommended product.

[0043] S104. Based on the optimal communication words and the target recommended product, obtain the recommended words corresponding to the target recommended product for reference by business personnel.

[0044] In one embodiment, based on the product information of the target recommended product, recommendation words related to the product are screened out from a preset insurance product recommendation words library.

[0045] In one embodiment, the selected recommendation words are modified and adjusted in combination with the characteristics of the best communication words and the target recommended product to ensure its coordination and consistency with the best communication words. The final recommendation words are generated and provided to the business personnel for reference. For example, if the target recommended product is a health insurance, some relevant recommendation words can be generated based on the customer's interests and needs, such as "This health insurance can help you cope with unexpected health risks and protect your physical health and life safety."

[0046] The above embodiment provides a personalized recommendation method, device, computer equipment and storage medium for insurance products, which obtains the first conversation voice between the current customer and the business personnel and the first product purchase data of the current customer, and converts the first conversation voice into text to obtain the first conversation text; analyzes the first conversation text and the first product purchase data to obtain the full life cycle label of the current customer; based on the full life cycle label of the current customer, determines the best communication words and target recommended products corresponding to the current customer; based on the best communication words and the target recommended products, obtains the recommended words corresponding to the target recommended products for reference by business personnel. This application can generate the customer's full life cycle label based on the conversation voice between the customer and the business personnel and the customer's product purchase data, and determine the best communication words adapted to the current customer based on the label, and generate the recommended words recommending the target recommended product in combination with the target recommended product, thereby improving the accuracy of the target recommended product and the personalization of the product recommendation.

[0047] See also Figure 2 , Figure 2This is a schematic flow chart of a personalized recommendation method for insurance products provided by an embodiment of the present application. The personalized recommendation method for insurance products can be applied to a server, and is used to determine a target communication speech in each recommended communication speech according to the matching degree of each communication speech with all recommended products, so that the obtained target communication speech is adapted to each target recommended product, thereby improving the personalization of product recommendations and further improving user experience.

[0048] like Figure 2 As shown, the personalized recommendation method for insurance products specifically includes steps S201 to S203.

[0049] S201, analyzing the full life cycle label of the current customer based on a preset communication script recommendation model to obtain at least one recommended communication script;

[0050] Furthermore, the preset communication words recommendation model is based on which the full life cycle tags of the current customer are analyzed to obtain at least one recommended communication words, including: based on the communication words recommendation model, the full life cycle tags of the current customer are analyzed to obtain analysis results; based on the analysis results, a preset keyword library is searched to obtain a recommended keyword set for the current customer; based on the recommended keyword set, at least one recommended communication words is generated.

[0051] In one embodiment, the full life cycle tags include tags such as the customer's interests, preferences, purchase history, and behavior patterns.

[0052] In one embodiment, the communication speech recommendation model is used to identify speech-related tags in the full life cycle tags, and to conduct a comprehensive analysis of the speech-related tags. Specifically, feature extraction is performed on the speech-related tags and converted into a format that can be used for model analysis. For example, text tags are converted into numerical features, or the sentiment tendency in the tags is extracted to generate analysis results. The model is used to search in the keyword library based on the analysis results to obtain a set of recommended keywords.

[0053] In one embodiment, the keyword library may include different categories of words, such as product features, customer needs, market trends, etc. The keyword library may be maintained by keywords based on opinions provided by business personnel.

[0054] In one embodiment, based on the recommended keyword set, at least one recommended communication speech is automatically generated using a communication speech recommendation model.

[0055] Furthermore, the preset communication language recommendation model analyzes the full life cycle label of the current customer, and before obtaining at least one recommended communication language, it also includes: obtaining second product purchase data of at least one historical customer and at least one second conversation voice between a historical business person and the historical customer; based on the second product purchase data and the second conversation voice, obtaining the full life cycle label of the historical customer and the training communication language for model training, and generating a training data set; based on the training data set, training the pre-trained model to obtain the communication language recommendation model.

[0056] In one embodiment, the second product purchase data of the historical customer and the second conversation voice of the historical business personnel and the historical customer are collected, such as telephone recordings, face-to-face communication recordings, etc. The second conversation voice is converted into text to obtain a second conversation text, and the second conversation text is processed to obtain at least one set of question-and-answer dialogues between the historical customer and the historical business personnel, wherein the set of question-and-answer dialogues includes a question text of the business personnel and at least one answer text of the customer.

[0057] In one embodiment, the question text and answer text in each group of question-and-answer dialogues are analyzed to obtain historical customer purchase tendency information and emotional information. For example, if a customer mentions health issues many times, then it can be marked as "health-conscious"; if the user repeatedly says "no need" and the user's tone is impatient, then the emotion is marked as negative.

[0058] In one embodiment, the question text is modified based on the purchase tendency information and the emotional information to obtain the training communication words for model training.

[0059] In one embodiment, historical customer life cycle labels, training communication scripts, and product purchase data are integrated together to generate a training data set.

[0060] In one embodiment, based on the training data set, the pre-trained model is trained to obtain a communication speech recommendation model, which can be used to predict and recommend communication speech and products related to the current customer.

[0061] Furthermore, the obtaining of training communication language for model training based on the second product purchase data and the second conversation voice includes: converting the second conversation voice into text to obtain a second conversation text, and processing the second conversation text to obtain at least one group of question-and-answer conversations between the historical customer and the historical business personnel, wherein a group of the question-and-answer conversations includes a question text of the business personnel and at least one answer text of the customer; analyzing the question text and the answer text in each group of the question-and-answer conversations to obtain the purchase tendency information and emotional information of the historical customer; and modifying the question text based on the purchase tendency information and emotional information to obtain training communication language for model training.

[0062] In one embodiment, natural language processing techniques such as word segmentation, part-of-speech tagging, and named entity recognition are performed on the second conversation text to extract keywords and entities.

[0063] In one embodiment, the second dialogue text is divided into a plurality of question-answer dialogues according to keywords and entities, wherein each question-answer dialogue includes a question text and at least one answer text. For example, "What type of insurance do you need to buy?" can be used as a question text, and "I want to buy an accident insurance" can be used as an answer text.

[0064] In one embodiment, the question text and answer text in each group of question-and-answer dialogues are analyzed to obtain historical customer purchase tendency information and emotional information. For example, if a customer mentions health issues many times, then it can be marked as "health-conscious"; if the user repeatedly says "no need" and the user's tone is impatient, then the emotion is marked as negative.

[0065] In one embodiment, the question text is modified based on the purchase tendency information and the emotion information to obtain the training communication words for model training. For example, if the customer is in a negative mood, the question text can be modified to a more daily promotional wording rather than a wording related to product introduction.

[0066] S202: Analyze the full life cycle label of the current customer based on a preset insurance product recommendation model to obtain at least one target recommended product;

[0067] Furthermore, based on the preset insurance product recommendation model, the full life cycle label of the current customer is analyzed to obtain at least one target recommended product, including: based on the insurance product recommendation model, the full life cycle label is identified to obtain a label related to the insurance product; feature extraction is performed on the label related to the insurance product to obtain product features, and matching is performed in a preset product database based on the product features to obtain at least one of the target recommended products.

[0068] In one embodiment, the full life cycle tags are analyzed according to the insurance product recommendation model to obtain key information of insurance products that the current customer may be interested in, such as premiums, insurance categories (health insurance, property insurance, etc.), insurance claim conditions, etc. Based on the key information of insurance products, matching products are searched in the insurance product library as recommended products.

[0069] Specifically, the entire life cycle labels can be integrated, and feature engineering such as normalization, missing value processing, discrete feature encoding, and continuous feature standardization can be performed on them. The insurance product promotion model can be used to match the features obtained after feature engineering in the product database to obtain at least one target recommended product.

[0070] S203: Determine the target communication words based on the comprehensive matching degree between each of the recommended communication words and all the target recommended products.

[0071] In one embodiment, a matching score is calculated for each recommended communication script and each recommended product, and the score can be based on multiple factors, such as customer interests, product features, market trends, etc.

[0072] In one embodiment, a weighted average of the matching scores of each recommended communication script and all recommended products is performed to obtain a comprehensive matching score of the recommended communication script.

[0073] In one embodiment, the comprehensive matching scores of all recommended communication words are sorted, and the recommended communication words with the highest comprehensive matching score are selected as the target communication words.

[0074] In one embodiment, if there are multiple recommended communication scripts with the same comprehensive matching scores, further screening can be performed based on other factors (such as customer preference, business personnel experience, etc.) to ultimately determine a target communication script.

[0075] In the above embodiment, the customer's full life cycle labels are analyzed according to the communication script recommendation model and the insurance product recommendation model respectively to obtain recommended communication scripts and recommended products, and the target communication scripts are determined in each recommended communication script according to the matching degree between each communication script and all recommended products, so that the obtained target communication scripts are adapted to each target recommended product, thereby improving the personalization of product recommendations and further improving the user experience.

[0076] See also Figure 3 , Figure 3This is a schematic flow chart of a personalized recommendation method for insurance products provided in an embodiment of the present application. The personalized recommendation method for insurance products can be applied to a server, and by analyzing the obtained product preference information and speech preference information, a full life cycle label of the current customer is generated, which can accurately reflect the user's preference information at all stages. The use of full life cycle labels for insurance product recommendations and speech recommendations can improve the accuracy and personalization of the results.

[0077] like Figure 3 As shown, the personalized recommendation method for insurance products specifically includes steps S301 to S304.

[0078] S301, performing semantic analysis on the first dialogue text to obtain first product preference information and speech preference information;

[0079] S302: Analyze the first product purchase data to obtain second product preference information;

[0080] S303: Generate product preference information of the current customer based on the first product preference information and the second product preference information;

[0081] S304: Based on the product preference information and the speech preference information, obtain the full life cycle label of the current customer.

[0082] In one embodiment, natural language processing technology is used to conduct in-depth analysis of the conversation text to identify the customer's interests, emotional tendencies, and needs. Key phrases and words in the conversation are identified as the basis for speech preferences.

[0083] In one embodiment, the purchase data is analyzed and combined with the insurance product related information in the conversation voice to obtain the current customer's product preference information.

[0084] Specifically, the conversation text can be segmented and POS tagged to convert the text into a form that can be processed by a computer. Based on the POS tagging results, feature words related to the product are extracted. Cluster analysis is performed on the feature words to group similar feature words into one category, thereby obtaining product categories. For each product category, the number and frequency of times the customer mentions it in the conversation are counted to obtain the first product preference information. For each product category, the style and preference of the customer's speech in the conversation are counted to obtain speech preference information, such as whether the customer prefers concise and clear expressions or detailed descriptions.

[0085] In one embodiment, the purchase records are cleaned and preprocessed, including removing duplicate records, processing missing values, etc. The purchase records are converted into a binary matrix, where each row represents a customer and each column represents a product. If the customer purchases the product, the corresponding element is 1, otherwise it is 0. The binary matrix is ​​analyzed using an association rule mining algorithm to obtain frequent item sets and association rules. According to the support, confidence, and lift of the association rules, important association rules are screened out to obtain the second product preference information. For each product category, the number and frequency of purchases by the customer in the historical purchase records are counted to obtain the second product preference information.

[0086] In one embodiment, the first product preference information and the second product preference information are combined to obtain the product preference information of the current customer.

[0087] In one embodiment, product preference information is used to label customers with relevant tags, such as "prefer high-risk investment products", "prefer family insurance", etc.

[0088] In one embodiment, the customer is labeled with a communication style, such as "prefers direct communication", "prefers detailed explanation", etc., in combination with the speech preference information.

[0089] In the above embodiment, by analyzing the obtained product preference information and speech preference information, a full life cycle label for the current customer is generated, which can accurately reflect the user's preference information at all stages. Using the full life cycle label for insurance product recommendations and speech recommendations can improve the accuracy and personalization of the results.

[0090] See also Figure 4 , Figure 4 The embodiment of the present application provides a schematic block diagram of a personalized recommendation device for insurance products, wherein the personalized recommendation device for insurance products is used to execute the aforementioned personalized recommendation method for insurance products. The personalized recommendation device for insurance products can be configured on a server.

[0091] like Figure 4 As shown, the personalized recommendation device 400 for insurance products includes:

[0092] The data acquisition module 401 is used to acquire a first conversation voice between a current customer and a business person and first product purchase data of the current customer, and to convert the first conversation voice into text to obtain a first conversation text;

[0093] A label obtaining module 402 is used to analyze the first conversation text and the first product purchase data to obtain a full life cycle label of the current customer;

[0094] The speech and product determination module 403 is used to determine the best communication speech and target recommended product corresponding to the current customer based on the full life cycle tag of the current customer;

[0095] The recommended speech acquisition module 404 is used to obtain the recommended speech corresponding to the target recommended product based on the optimal communication speech and the target recommended product for reference by business personnel.

[0096] Furthermore, the speech and product determination module 403 includes:

[0097] A recommended communication script obtaining unit, configured to analyze the full life cycle tag of the current customer based on a preset communication script recommendation model to obtain at least one recommended communication script;

[0098] A recommended product obtaining unit, configured to analyze the full life cycle label of the current customer based on a preset insurance product recommendation model to obtain at least one target recommended product;

[0099] The target communication words determining unit is used to determine the target communication words based on the comprehensive matching degree between each of the recommended communication words and all the target recommended products.

[0100] Furthermore, the recommended communication words obtaining unit includes:

[0101] An analysis result obtaining subunit is used to analyze the full life cycle label of the current customer based on the communication speech recommendation model to obtain an analysis result;

[0102] A keyword set obtaining subunit, configured to search in a preset keyword library based on the analysis result to obtain a recommended keyword set for the current customer;

[0103] The recommended communication words generation subunit is used to generate at least one of the recommended communication words based on the recommended keyword set.

[0104] Furthermore, the target recommended product obtaining unit includes:

[0105] A label acquisition subunit, used to identify the full life cycle label based on the insurance product recommendation model to obtain a label related to the insurance product;

[0106] The recommended product acquisition subunit is used to extract features from the labels related to the insurance products to obtain product features, and to match the product features in a preset product database to obtain at least one of the target recommended products.

[0107] Furthermore, the speech and product determination module 403 further includes:

[0108] A second conversation voice acquisition unit, used to acquire second product purchase data of at least one historical customer and at least one second conversation voice of a historical business person and the historical customer;

[0109] A training data set generating unit, configured to obtain the full life cycle label of the historical customer and the training communication speech for model training based on the second product purchase data and the second conversation voice, and generate a training data set;

[0110] The model training unit is used to train the pre-trained model based on the training data set to obtain the communication speech recommendation model.

[0111] Furthermore, the training data set generating unit comprises:

[0112] a question-and-answer dialogue obtaining subunit, configured to convert the second dialogue speech into text to obtain a second dialogue text, and process the second dialogue text to obtain at least one set of question-and-answer dialogues between the historical customer and the historical business personnel, wherein the set of question-and-answer dialogues includes a question text of the business personnel and at least one answer text of the customer;

[0113] An information acquisition subunit, used to analyze the question text and the answer text in each group of the question-answer dialogue to obtain the purchase tendency information and emotion information of the historical customers;

[0114] The training communication script acquisition subunit is used to modify the question text based on the purchase tendency information and the emotional information to obtain the training communication script for model training.

[0115] Furthermore, the label obtaining module 402 includes:

[0116] A preference information obtaining unit, configured to perform semantic analysis on the first dialogue text to obtain first product preference information and speech preference information;

[0117] The preference information obtaining unit is further used to analyze the first product purchase data to obtain second product preference information;

[0118] The preference information obtaining unit is further used to generate the product preference information of the current customer based on the first product preference information and the second product preference information;

[0119] The label obtaining unit is used to obtain the full life cycle label of the current customer based on the product preference information and the speech preference information.

[0120] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0121] The above-mentioned device can be implemented in the form of a computer program. Figure 5 Runs on the computer device shown.

[0122] See also Figure 5 , Figure 5 1 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device may be a server.

[0123] See also Figure 5 The computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.

[0124] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute a personalized recommendation method for any insurance product.

[0125] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0126] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute a personalized recommendation method for any insurance product.

[0127] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0128] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0129] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0130] Acquire a first conversation voice between a current customer and a business person and first product purchase data of the current customer, and convert the first conversation voice into text to obtain a first conversation text;

[0131] Analyze the first conversation text and the first product purchase data to obtain a full life cycle label of the current customer;

[0132] Based on the full life cycle label of the current customer, determine the best communication words and target recommended products corresponding to the current customer;

[0133] Based on the optimal communication words and the target recommended product, the recommended words corresponding to the target recommended product are obtained for reference by business personnel.

[0134] In one embodiment, when the processor determines the best communication words and target recommended products corresponding to the current customer based on the full life cycle tag of the current customer, it is used to implement:

[0135] Based on a preset communication script recommendation model, the full life cycle tag of the current customer is analyzed to obtain at least one recommended communication script;

[0136] Based on a preset insurance product recommendation model, analyzing the full life cycle label of the current customer to obtain at least one target recommended product;

[0137] Based on the comprehensive matching degree between each of the recommended communication words and all the target recommended products, the target communication words are determined.

[0138] In one embodiment, when the processor implements the preset communication speech recommendation model, analyzes the full life cycle tag of the current customer, and obtains at least one recommended communication speech, it is used to implement:

[0139] Based on the communication speech recommendation model, the full life cycle label of the current customer is analyzed to obtain analysis results;

[0140] Searching in a preset keyword library based on the analysis result to obtain a recommended keyword set for the current customer;

[0141] Based on the recommended keyword set, at least one of the recommended communication words is generated.

[0142] In one embodiment, when the processor analyzes the full life cycle tag of the current customer based on a preset insurance product recommendation model to obtain at least one target recommended product, it is configured to implement:

[0143] Based on the insurance product recommendation model, the full life cycle label is identified to obtain a label related to the insurance product;

[0144] Feature extraction is performed on the labels related to the insurance products to obtain product features, and matching is performed in a preset product database based on the product features to obtain at least one of the target recommended products.

[0145] In one embodiment, before the processor implements the preset communication speech recommendation model to analyze the full life cycle tag of the current customer and obtain at least one recommended communication speech, it is further used to implement:

[0146] Acquire second product purchase data of at least one historical customer and at least one second conversation voice of a historical business person and the historical customer;

[0147] Based on the second product purchase data and the second conversation voice, obtain the full life cycle label of the historical customer and the training communication words for model training, and generate a training data set;

[0148] Based on the training data set, the pre-training model is trained to obtain the communication speech recommendation model.

[0149] In one embodiment, when the processor obtains the training communication speech for model training based on the second product purchase data and the second conversation voice, it is configured to implement:

[0150] Converting the second conversation speech into text to obtain a second conversation text, and processing the second conversation text to obtain at least one set of question-and-answer conversations between the historical customer and the historical business personnel, wherein the set of question-and-answer conversations includes a question text of the business personnel and at least one answer text of the customer;

[0151] Analyze the question text and the answer text in each group of the question-and-answer dialogue to obtain the purchase tendency information and emotion information of the historical customers;

[0152] Based on the purchase tendency information and the emotional information, the question text is modified to obtain training communication words for model training.

[0153] In one embodiment, when the processor analyzes the first conversation text and the first product purchase data to obtain the full life cycle label of the current customer, it is configured to implement:

[0154] Performing semantic analysis on the first dialogue text to obtain first product preference information and speech preference information;

[0155] Analyze the first product purchase data to obtain second product preference information;

[0156] generating product preference information of the current customer based on the first product preference information and the second product preference information;

[0157] Based on the product preference information and the speech preference information, a full life cycle label of the current customer is obtained.

[0158] A computer-readable storage medium is also provided in an embodiment of the present application, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and the processor executes the program instructions to implement any personalized recommendation method for insurance products provided in the embodiments of the present application.

[0159] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (Smart Medi a Card, SMC), a secure digital (Secure Digital, SD) card, a flash memory card (Flash Card), etc., equipped on the computer device.

[0160] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A personalized recommendation method for insurance products, characterized in that: include: Acquire a first conversation voice between a current customer and a business person and first product purchase data of the current customer, and convert the first conversation voice into text to obtain a first conversation text; Analyze the first conversation text and the first product purchase data to obtain a full life cycle label of the current customer; Based on the full life cycle label of the current customer, determine the best communication words and target recommended products corresponding to the current customer; Based on the optimal communication words and the target recommended product, the recommended words corresponding to the target recommended product are obtained for reference by business personnel.

2. The personalized recommendation method for insurance products according to claim 1, characterized in that: Determining the best communication words and target recommended products corresponding to the current customer based on the full life cycle tag of the current customer includes: Based on a preset communication script recommendation model, the full life cycle tag of the current customer is analyzed to obtain at least one recommended communication script; Based on a preset insurance product recommendation model, analyzing the full life cycle label of the current customer to obtain at least one target recommended product; Based on the comprehensive matching degree between each of the recommended communication words and all the target recommended products, the target communication words are determined.

3. The personalized recommendation method for insurance products according to claim 2, characterized in that: The preset communication words recommendation model is used to analyze the full life cycle tag of the current customer to obtain at least one recommended communication words, including: Based on the communication speech recommendation model, the full life cycle label of the current customer is analyzed to obtain analysis results; Searching in a preset keyword library based on the analysis result to obtain a recommended keyword set for the current customer; Based on the recommended keyword set, at least one of the recommended communication words is generated.

4. The personalized recommendation method for insurance products according to claim 2, characterized in that: The method of analyzing the full life cycle label of the current customer based on the preset insurance product recommendation model to obtain at least one target recommended product includes: Based on the insurance product recommendation model, the full life cycle label is identified to obtain a label related to the insurance product; Feature extraction is performed on the labels related to the insurance products to obtain product features, and matching is performed in a preset product database based on the product features to obtain at least one of the target recommended products.

5. The personalized recommendation method for insurance products according to claim 2, characterized in that: Before analyzing the full life cycle tag of the current customer based on the preset communication speech recommendation model to obtain at least one recommended communication speech, the method further includes: Acquire second product purchase data of at least one historical customer and at least one second conversation voice of a historical business person and the historical customer; Based on the second product purchase data and the second conversation voice, obtain the full life cycle label of the historical customer and the training communication words for model training, and generate a training data set; Based on the training data set, the pre-training model is trained to obtain the communication speech recommendation model.

6. The personalized recommendation method for insurance products according to claim 5, characterized in that: The obtaining, based on the second product purchase data and the second conversation voice, a training communication speech for model training includes: Converting the second conversation speech into text to obtain a second conversation text, and processing the second conversation text to obtain at least one set of question-and-answer conversations between the historical customer and the historical business personnel, wherein the set of question-and-answer conversations includes a question text of the business personnel and at least one answer text of the customer; Analyze the question text and the answer text in each group of the question-and-answer dialogue to obtain the purchase tendency information and emotion information of the historical customers; Based on the purchase tendency information and the emotional information, the question text is modified to obtain training communication words for model training.

7. The personalized recommendation method for insurance products according to any one of claims 1 to 6, characterized in that: The analyzing the first conversation text and the first product purchase data to obtain the full life cycle label of the current customer includes: Performing semantic analysis on the first dialogue text to obtain first product preference information and speech preference information; Analyze the first product purchase data to obtain second product preference information; generating product preference information of the current customer based on the first product preference information and the second product preference information; Based on the product preference information and the speech preference information, a full life cycle label of the current customer is obtained.

8. A personalized recommendation device for insurance products, characterized in that: include: A data acquisition module, used to acquire a first conversation voice between a current customer and a business person and first product purchase data of the current customer, and to convert the first conversation voice into text to obtain a first conversation text; A label acquisition module, used for analyzing the first conversation text and the first product purchase data to obtain a full life cycle label of the current customer; A speech and product determination module, used to determine the best communication speech and target recommended product corresponding to the current customer based on the full life cycle label of the current customer; The recommended speech acquisition module is used to obtain the recommended speech corresponding to the target recommended product based on the optimal communication speech and the target recommended product for reference by business personnel.

9. A computer device, characterized in that: The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is used to execute the computer program and implement the personalized recommendation method for insurance products according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, enables the processor to implement the personalized recommendation method for insurance products according to any one of claims 1 to 7.

Images