Product recommendation strategy generation method and device, computer equipment and storage medium

Abstract

The invention belongs to the technical field of intelligent decision making, is suitable for a financial service scene, and relates to a product recommendation strategy generation method and device, computer equipment and a storage medium, and the method comprises the steps: receiving session consultation data sent by a user terminal; performing consultation feature extraction operation on the session consultation data to obtain consultation feature data; performing keyword extraction operation on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result; calling a product database, and obtaining target product information corresponding to the keyword extraction result from the product database; reading a topic library, and obtaining target topic data corresponding to the session consultation data in the topic library; constructing a target recommendation strategy according to the target product information and the target topic data; and outputting the target product information to the user terminal according to the target recommendation strategy. According to the method and the device, the insurance product can be recommended by effectively utilizing the chat content with the customer, and the recommendation accuracy and the customer satisfaction are effectively improved.

Description

Technical Field

[0001] This application relates to the field of intelligent decision-making technology, and is applicable to the financial field, and particularly relates to a method, device, computer device, and storage medium for generating a product recommendation strategy. Background Art

[0002] With the popularization of Internet technology and the accumulation of user data, more and more enterprises realize that using big data technology for product recommendation can improve user experience, increase sales, and enhance market competitiveness. However, in order to improve the accuracy of product recommendation when recommending products to target users, it is necessary to generate a product recommendation strategy that better conforms to user behavior, so as to recommend products to target users.

[0003] The existing big data product recommendation strategy generation technology recommends based on the correlation between user behavior data and items. Based on the user behavior history and the preferences of similar users for items, by calculating the similarity between users or items, it predicts the items that users like.

[0004] However, in actual applications, the degree of user preference for products will change in real time. Only considering the correlation between users and items makes it impossible to grasp users' preferences in real time, resulting in relatively low accuracy when generating a product recommendation strategy for users. Summary of the Invention

[0005] The purpose of the embodiments of this application is to propose a method, device, computer device, and storage medium for generating a product recommendation strategy to solve the problem of relatively low accuracy of the existing big data product recommendation strategy generation technology.

[0006] To solve the above technical problems, the embodiments of this application provide a method for generating a product recommendation strategy, which adopts the following technical solutions:

[0007] Receive session consultation data sent by a user terminal;

[0008] Perform an operation of extracting consultation features on the session consultation data to obtain consultation feature data;

[0009] Perform a keyword extraction operation on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result;

[0010] Call a product database and obtain target product information corresponding to the keyword extraction result in the product database;

[0011] Read a topic library and obtain target topic data corresponding to the session consultation data in the topic library;

[0012] Construct a target recommendation strategy according to the target product information and the target topic data;

[0013] Output the target product information to the user terminal according to the target recommendation strategy.

[0014] Further, after the step of receiving the session consultation data sent by the user terminal and before the step of performing an operation of extracting consultation features on the session consultation data to obtain consultation feature data, the following steps are further included:

[0015] Perform a first preprocessing operation on the session consultation data to obtain preprocessed consultation data;

[0016] The step of performing an operation of extracting consultation features on the session consultation data to obtain consultation feature data specifically includes the following steps:

[0017] Perform an operation of extracting consultation features on the preprocessed consultation data to obtain the consultation feature data.

[0018] Further, the step of constructing a target recommendation strategy according to the target product information and the target topic data includes the following steps:

[0019] Obtain the user portrait information of the consulting user corresponding to the session consultation data;

[0020] Construct a target recommendation strategy according to the user portrait information, the target product information, and the target topic data.

[0021] Further, after the step of reading the topic library and obtaining the target topic data corresponding to the session consultation data in the topic library and before the step of constructing a target recommendation strategy according to the target product information and the target topic data, the following steps are further included:

[0022] Call the trained sentiment analysis model, and input the session consultation data and the target topic data into the trained sentiment analysis model for sentiment analysis operation to obtain sentiment tendency data;

[0023] After the step of constructing a target recommendation strategy according to the target product information and the target topic data, the following steps are further included:

[0024] Optimize the target recommendation strategy according to the sentiment tendency data to obtain an optimized recommendation strategy;

[0025] The step of outputting the target product information to the user terminal specifically includes the following steps:

[0026] Output the target product information to the user terminal according to the optimized recommendation strategy.

[0027] Further, before the step of calling the trained sentiment analysis model, inputting the session consultation data and the target topic data into the trained sentiment analysis model for sentiment analysis operation, and obtaining sentiment tendency data, the following steps are further included:

[0028] Read the system database, and obtain the training text data labeled with sentiment tags in the system database;

[0029] Perform a second preprocessing operation on the training text data to obtain preprocessed training text data;

[0030] Perform a training feature extraction operation on the training text data to obtain training feature data;

[0031] Construct an initially trained sentiment analysis model, and input the preprocessed training text data and the training feature data into the initially trained sentiment analysis model for model training operation to obtain a trained sentiment analysis model.

[0032] Further, after the step of constructing an initially trained sentiment analysis model, inputting the preprocessed training text data and the training feature data into the initially trained sentiment analysis model for model training operation, and obtaining a trained sentiment analysis model, the following steps are further included:

[0033] Obtain the test text data corresponding to the training text data in the system database;

[0034] Perform a model evaluation operation on the trained sentiment analysis model according to the test text data to obtain a model evaluation result;

[0035] Perform a model optimization operation on the trained sentiment analysis model according to the model evaluation result to obtain an optimized sentiment analysis model.

[0036] Further, after the step of outputting the target product information to the user terminal according to the target recommendation strategy, the following steps are further included:

[0037] Obtain the user feedback information corresponding to the target product information sent by the user terminal;

[0038] Perform an update operation on the product database and the topic library according to the user feedback information.

[0039] To solve the above technical problems, an embodiment of the present application further provides a product recommendation strategy generation device, which adopts the following technical solutions:

[0040] A consultation data acquisition module, configured to receive session consultation data sent by a user terminal;

[0041] A consultation feature extraction module, configured to perform a consultation feature extraction operation on the session consultation data to obtain consultation feature data;

[0042] A keyword extraction module, configured to perform a keyword extraction operation on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result;

[0043] A target product acquisition module, configured to call a product database to obtain target product information corresponding to the keyword extraction result in the product database;

[0044] A target topic acquisition module, configured to read a topic library and obtain target topic data corresponding to the session consultation data in the topic library;

[0045] A recommendation strategy construction module, configured to construct a target recommendation strategy according to the target product information and the target topic data;

[0046] A target product output module, configured to output the target product information to the user terminal according to the target recommendation strategy.

[0047] To solve the above technical problems, an embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0048] It includes a memory and a processor. Computer-readable instructions are stored in the memory. When the processor executes the computer-readable instructions, the steps of the product recommendation strategy generation method described above are implemented.

[0049] To solve the above technical problems, an embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0050] Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by a processor, the steps of the product recommendation strategy generation method described above are implemented.

[0051] The present application provides a method for generating a product recommendation strategy, including: receiving session consultation data sent by a user terminal; performing an operation of extracting consultation features on the session consultation data to obtain consultation feature data; performing an operation of extracting keywords on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result; calling a product database to obtain target product information corresponding to the keyword extraction result in the product database; reading a topic library and obtaining target topic data corresponding to the session consultation data in the topic library; constructing a target recommendation strategy according to the target product information and the target topic data; and outputting the target product information to the user terminal according to the target recommendation strategy. Compared with the prior art, the present application can effectively utilize the chat content with customers to recommend insurance products, effectively improving the accuracy of recommendation and customer satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the solutions in the present application, the following will briefly introduce the drawings required for the description of the embodiments of the present application. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0053] Figure 1 is an exemplary system architecture diagram to which the present application can be applied;

[0054] Figure 2 is a flowchart of the implementation of the method for generating a product recommendation strategy provided by an embodiment of the present application;

[0055] Figure 3 is a schematic structural diagram of a product recommendation strategy generation device provided by an embodiment of the present application;

[0056] Figure 4 is a schematic structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art belonging to the technical field of the present application; the terms used in the description of the present application in the specification are only for the purpose of describing specific embodiments, and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above drawings are intended to cover non-exclusive inclusion. The terms "first", "second", etc. in the specification and claims of the present application or the above drawings are used to distinguish different objects, rather than to describe a specific order.

[0058] Reference to "embodiments" in this document means that the specific features, structures, or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The phrase may not necessarily refer to the same embodiment at various places in the specification, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0059] To enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0060] As Figure 1 shown, the system architecture 100 may include a terminal device 101, a network 102, and a server 103. The terminal device 101 may be a laptop computer 1011, a tablet computer 1012, or a mobile phone 1013. The network 102 is a medium for providing a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0061] A user may use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications may be installed on the terminal device 101, such as a web browser application, a shopping application, a search application, an instant messaging tool, an email client, a social platform software, etc.

[0062] The terminal device 101 may be various electronic devices with a display screen and supporting web browsing. In addition to the laptop computer 1011, the tablet computer 1012, or the mobile phone 1013, the terminal device 101 may also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop portable computer, a desktop computer, etc.

[0063] The server 103 may be a server providing various services, such as a background server supporting the pages displayed on the terminal device 101.

[0064] It should be noted that the product recommendation strategy generation method provided in the embodiments of the present application is generally executed by the server / terminal device. Correspondingly, the product recommendation strategy generation device is generally disposed in the server / terminal device.

[0065] Understand,Figure 1 The numbers of the terminal devices, networks, and servers in

[0066] Continue to refer to Figure 2 , which shows a flowchart of an embodiment of a method for generating a product recommendation strategy according to the present application. The method for generating a product recommendation strategy includes: step S201, step S202, step S203, step S204, step S205, step S206, and step S207.

[0067] In step S201, session consultation data sent by a user terminal is received.

[0068] In the embodiments of the present application, users input a question or consultation through their terminal devices (such as mobile phones, computers, etc.), and this input is received by the system. This session consultation data is the specific content that the user hopes the system to answer. Specifically, the session consultation data may be "transaction data or payment data or business data or purchase data" related to financial institutions (such as banks, etc.). It should be understood that the examples of session consultation data here are only for easy understanding and are not used to limit the present application.

[0069] In the embodiments of the present application, the present application may obtain the session consultation data sent by the user terminal based on the chat channel of WeChat customer service.

[0070] In step S202, a consultation feature extraction operation is performed on the session consultation data to obtain consultation feature data.

[0071] In the embodiments of the present application, the consultation feature extraction operation is mainly used to represent the features of the session consultation text, and these features will be used in the subsequent keyword extraction algorithm. Among them, the features used to represent the session consultation text may include:

[0072] Word frequency: The number of times a word appears in the text.

[0073] Inverse document frequency (IDF): A measure of the importance of a word in distinguishing documents in a document collection.

[0074] TF-IDF: A metric that combines word frequency (TF) and inverse document frequency (IDF) and is used to evaluate the importance of a word in a document.

[0075] Word vector: Represent a word as a vector in a high-dimensional space, such as word vectors generated by models such as Word2Vec, GloVe, or BERT.

[0076] In step S203, a keyword extraction operation is performed on the consultation feature data according to the keyword extraction algorithm to obtain a keyword extraction result.

[0077] In the embodiments of the present application, the keyword extraction algorithm may be:

[0078] Statistical-based method: As an example, such as TF-IDF, which combines term frequency and inverse document frequency to evaluate the importance of words.

[0079] Graph-based method: As an example, such as TextRank, which uses a graph model to calculate the importance of words, where nodes represent words and edges represent co-occurrence relationships between words.

[0080] Machine learning-based method: Use supervised or unsupervised learning models to extract keywords. These models can be trained based on text features (such as TF-IDF, word vectors, etc.).

[0081] Deep learning-based method: Such as using pre-trained language models like BERT for keyword extraction. These models are usually fine-tuned to adapt to specific keyword extraction tasks.

[0082] In some alternative implementation manners of the present application, after the keyword extraction operation, some post-processing operations can also be performed to improve accuracy and readability. Among them, the post-processing operations include:

[0083] Removing duplicate keywords: Ensure that the extracted keyword list does not contain duplicate words;

[0084] Keyword merging: Merge related keywords into one phrase to improve the expressiveness of keywords;

[0085] Keyword sorting: Sort them according to the importance of keywords so that users can more easily find the most important information.

[0086] In step S204, call the product database to obtain the target product information corresponding to the keyword extraction result in the product database.

[0087] In the embodiments of the present application, the product database is a data set storing product information. Among them, the product information may be information related to financial products. As an example, the financial product may be a basic financial product, such as: currency, bond, stock, etc. The financial product may also be a derivative financial product, such as: fund, insurance product, futures, and options, etc. It should be understood that the examples of the product database here are only for easy understanding and are not used to limit the present application.

[0088] In the embodiments of the present application, the target product information refers to the specific product information that matches these keywords, which is of interest to users or directly related to users' queries and needs. Further, the target product information may be a combination of one or more products.

[0089] In the embodiments of the present application, product information matching the keywords can be retrieved through SQL statements.

[0090] In step S205, the topic library is read, and target topic data corresponding to the conversation consultation data is obtained from the topic library.

[0091] In the embodiments of the present application, the topic library is a data set storing multiple topics and their related information. These topics represent various themes of user consultations. As an example, for instance: health, travel, family, etc. Further, the topic can be associated with insurance products. As an example, for instance: the health topic is associated with health insurance products.

[0092] In the embodiments of the present application, the keyword extraction results obtained above can be used to accurately understand the user's consultation intention, and the most suitable topic can be found from the topic library.

[0093] In step S206, a target recommendation strategy is constructed based on the target product information and the target topic data.

[0094] In the embodiments of the present application, the target recommendation strategy refers to a recommendation scheme constructed based on product information and topic data, aiming to provide personalized product recommendations for users and improve user satisfaction and purchase intention.

[0095] In the embodiments of the present application, the target recommendation strategy can be constructed through the following 3 steps. Specifically:

[0096] (1) Data collection and integration:

[0097] First, the system needs to collect and integrate the target product information and the target topic data. This requires obtaining data from multiple data sources and performing cleaning and standardization processing to ensure the accuracy and consistency of the data;

[0098] (2) Data analysis and mining:

[0099] Next, the system needs to analyze and mine the collected data. This includes text analysis, sentiment analysis, association rule mining, etc. to discover potential associations and rules between product information and topic data;

[0100] (3) Strategy construction and optimization:

[0101] Based on the results of data analysis, the system can construct a preliminary recommendation strategy. This includes content-based recommendation, collaborative filtering recommendation, user behavior-based recommendation, etc. Then, the system needs to optimize the recommendation strategy to improve the accuracy and diversity of the recommendation.

[0102] In some alternative implementation manners of the embodiments of the present application, the above step of constructing the target recommendation strategy may further include strategy implementation and evaluation. Specifically:

[0103] The system needs to apply the constructed recommendation strategy to the actual scenario and evaluate it based on the user's feedback and the system's performance. This requires using metrics such as A / B testing, click-through rate, conversion rate, etc. to evaluate the effectiveness of the recommendation strategy and perform iteration and optimization based on the evaluation results.

[0104] In step S207, the target product information is output to the user terminal according to the target recommendation strategy.

[0105] In the embodiments of the present application, a method for generating a product recommendation strategy is provided, including: receiving session consultation data sent by a user terminal; performing a consultation feature extraction operation on the session consultation data to obtain consultation feature data; performing a keyword extraction operation on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result; calling a product database to obtain target product information corresponding to the keyword extraction result in the product database; reading a topic library and obtaining target topic data corresponding to the session consultation data in the topic library; constructing a target recommendation strategy according to the target product information and the target topic data; and outputting the target product information to the user terminal according to the target recommendation strategy. Compared with the prior art, the present application can effectively utilize the chat content with customers to recommend insurance products, effectively improving the accuracy of recommendation and customer satisfaction.

[0106] In some alternative implementation manners of the embodiments of the present application, after the above step of receiving the session consultation data sent by the user terminal and before the above step of performing a consultation feature extraction operation on the session consultation data to obtain consultation feature data, the following steps are further included:

[0107] Performing a first preprocessing operation on the session consultation data to obtain preprocessed consultation data;

[0108] The step of performing a consultation feature extraction operation on the session consultation data to obtain consultation feature data specifically includes the following steps:

[0109] Performing a consultation feature extraction operation on the preprocessed consultation data to obtain consultation feature data.

[0110] In the embodiments of the present application, the preprocessing operation includes:

[0111] (1) Removing irrelevant characters: such as punctuation marks, special symbols, numbers (unless the numbers are important for understanding the text), etc.;

[0112] (2) Word segmentation: splitting the text into words or phrases. For Chinese text, this usually requires using a Chinese word segmentation tool, such as jieba;

[0113] (3) Remove stop words: Stop words are words that frequently appear in a language but contribute little to the meaning of the text, such as "de", "le", "zai", etc.;

[0114] (4) Stemming (for English): Restore a word to its basic form, such as restoring "running" to "run";

[0115] (5) Lemmatization (for English): Unify different forms of a word into the same form, such as unifying "dogs" and "dog" into "dog".

[0116] In some alternative implementation manners of the embodiments of the present application, the step of constructing a target recommendation strategy according to the target product information and the target topic data includes the following steps:

[0117] Obtain the user profile information of the consulting user corresponding to the conversation consultation data;

[0118] Construct a target recommendation strategy according to the user profile information, the target product information, and the target topic data.

[0119] In the embodiments of the present application, the system formulates a personalized recommendation strategy according to the user's profile information and the target topic data. This strategy aims to provide users with recommendations of products or services that match their interests and needs. Among them, the formulation of the recommendation strategy can be:

[0120] (1) Personalized recommendation: The system recommends products or services that best match the user's interests and needs according to the user's profile information (such as hobbies, purchase history, etc.) and the target topic data;

[0121] (2) Contextual recommendation: The system considers the context where the user is currently located (such as time, location, weather, etc.), as well as the target topic data, to provide users with recommendations that are more suitable for the current context;

[0122] (3) Diversified recommendation: The system not only considers the user's direct interests and needs, but also tries to recommend some products or services that are related to the user's interests but may not have been contacted before, in order to increase the user's exploration and diversity.

[0123] In the embodiments of the present application, by obtaining the user's profile information and the target topic data, and formulating a personalized recommendation strategy accordingly, the system can more accurately meet the user's needs and preferences, and improve the accuracy and diversity of the recommendations.

[0124] In some alternative implementation manners of the embodiments of the present application, after the step of reading the topic library and obtaining the target topic data corresponding to the conversation consultation data in the topic library, and before the step of constructing the target recommendation strategy according to the target product information and the target topic data, the following steps are further included:

[0125] Call the trained sentiment analysis model, and input the conversation consultation data and the target topic data into the trained sentiment analysis model for sentiment analysis operation to obtain sentiment tendency data;

[0126] After the step of constructing the target recommendation strategy according to the target product information and the target topic data, the following steps are further included:

[0127] Optimize the target recommendation strategy according to the sentiment tendency data to obtain an optimized recommendation strategy;

[0128] The step of outputting the target product information to the user terminal specifically includes the following steps:

[0129] Output the target product information to the user terminal according to the optimized recommendation strategy.

[0130] In the embodiments of the present application, the trained sentiment analysis model judges the sentiment tendency according to the input chat data and target topic data to obtain sentiment tendency data.

[0131] In the embodiments of the present application, the sentiment tendency data represents the sentiment tendency of the user towards the current topic or chat content, which may be positive, negative or neutral, and sometimes may also include information on the sentiment intensity.

[0132] In the embodiments of the present application, the operation of optimizing the target recommendation strategy refers to the process of adjusting or improving the target recommendation strategy according to the sentiment tendency data. Among them, the optimization operation may involve adjusting the sorting of recommended products, increasing or decreasing the types of recommended products, changing the style of recommended copywriting, etc.

[0133] In the embodiments of the present application, by using sentiment analysis to optimize the recommendation strategy, the accuracy of the recommendation and the user satisfaction can be effectively improved.

[0134] In some alternative implementation manners of the embodiments of the present application, before the step of calling the trained sentiment analysis model, and inputting the conversation consultation data and the target topic data into the trained sentiment analysis model for sentiment analysis operation to obtain sentiment tendency data, the following steps are further included:

[0135] Read the system database and obtain the training text data marked with sentiment labels in the system database;

[0136] Performing a second preprocessing operation on the training text data to obtain preprocessed training text data;

[0137] Performing a training feature extraction operation on the training text data to obtain training feature data;

[0138] An initial trained sentiment analysis model is constructed, and the preprocessed training text data and training feature data are input into the initial trained sentiment analysis model for model training operation to obtain a trained sentiment analysis model.

[0139] In the embodiment of the present application, the system database usually stores a large amount of text data, which may come from various channels, such as social media, user comments, news articles, etc. Further, the text data may also come from financial channels. In order to train the sentiment analysis model, the present application needs to filter out texts that have been marked with sentiment tags from these data. Sentiment tags usually indicate the emotional tendency expressed by the text, such as positive, negative or neutral.

[0140] In the embodiment of the present application, when reading the database, the present application needs to ensure the integrity and accuracy of the data. This includes checking the format, encoding, and consistency and accuracy of the sentiment labels of the data. If there are erroneous or inconsistent labels in the data, it may have a negative impact on the training of the model.

[0141] In the embodiment of the present application, the preprocessing operation is mainly used to reduce noise, improve data quality, and prepare for subsequent feature extraction and model training. The preprocessing operation may include:

[0142] (1) Remove stop words: Stop words are words that appear frequently in a language but do not contribute much to the meaning of the text, such as "的" and "了". Removing these words can reduce the sparsity of the data and improve the efficiency of the model;

[0143] (2) Word segmentation: For Chinese text, word segmentation is the process of dividing the text into words or phrases. The quality of word segmentation has a great impact on subsequent feature extraction and model training;

[0144] (3) Stemming / lemma restoration (for languages ​​like English): restoring words to their basic form, such as restoring "running" to "run", helps reduce vocabulary diversity and improves the generalization ability of the model. Although the concept of lemma restoration is not very applicable in Chinese, word segmentation and removing redundant words are equally important;

[0145] (4) Removing punctuation and numbers: These characters usually do not contribute to sentiment analysis, and removing them can reduce the complexity of the data.

[0146] (5) Text normalization: Convert the text to lowercase or uppercase to ensure that the model is case-insensitive during processing.

[0147] After preprocessing, this application obtains a set of cleaner and uniformly formatted training text data.

[0148] In the embodiments of this application, the training feature extraction operation is mainly used to convert text data into numerical features, which will be used to train the sentiment analysis model. There are many feature extraction methods, including but not limited to:

[0149] (1) Bag of Words (BoW): Count the occurrences of each word in the text to obtain a word frequency vector;

[0150] (2) TF-IDF (Term Frequency - Inverse Document Frequency): On the basis of the bag of words model, adjust the importance of words by considering the frequency of words in the document and the inverse document frequency in the document set;

[0151] (3) Word embeddings: Such as Word2Vec, GloVe, etc., map each word to a high-dimensional vector space to capture the semantic relationships between words;

[0152] (4) N-gram features: Consider N consecutive words in the text as features, which helps to capture phrases and context information.

[0153] In the embodiments of this application, the extracted features can fully reflect the sentiment tendency of the text and provide sufficient information for the model to distinguish different sentiment categories.

[0154] In the embodiments of this application, first select a suitable sentiment analysis model framework, such as logistic regression, support vector machine, naive Bayes, convolutional neural network (CNN), recurrent neural network (RNN) and its variants (LSTM, GRU), etc. Then, this application inputs the preprocessed training text data and the extracted feature data into the model for model training to obtain a trained sentiment analysis model, which can be used for sentiment tendency judgment in actual applications.

[0155] In some optional implementation manners of the embodiments of this application, after the steps of constructing an initially trained sentiment analysis model, inputting the preprocessed training text data and the training feature data into the initially trained sentiment analysis model for model training operation to obtain a trained sentiment analysis model, the following steps are further included:

[0156] Obtain test text data corresponding to the training text data from the system database;

[0157] Perform a model evaluation operation on the trained sentiment analysis model according to the test text data to obtain the model evaluation result;

[0158] Perform a model optimization operation on the trained sentiment analysis model according to the model evaluation result to obtain the optimized sentiment analysis model.

[0159] In the embodiments of the present application, in addition to using the training text data marked with sentiment labels to train the model, the present application also needs to obtain a set of independent test text data to evaluate the performance of the model. This set of test text data is usually stored in the same system database, but is separated from the training text data to ensure the fairness and objectivity of the evaluation.

[0160] In the embodiments of the present application, the test text data also needs to be marked with sentiment labels, and these labels will be used to compare with the prediction results of the model to evaluate performance indicators such as the accuracy, recall rate, and F1 score of the model. To ensure the effectiveness of the evaluation, the distribution, style, and theme of the test text data should be similar to those of the training text data, but should not include exactly the same data that has appeared in the training set.

[0161] In the embodiments of the present application, the preprocessed and feature-extracted test text data is input into the trained sentiment analysis model, and the model is allowed to predict the test data. Then, the present application compares the prediction results of the model with the true sentiment labels of the test text data and calculates various performance indicators. Among them, the performance indicators can include:

[0162] Accuracy: The proportion of correct predictions by the model;

[0163] Precision: Among the samples predicted as positive by the model, the proportion that is truly positive;

[0164] Recall: Among all the samples that are truly positive, the proportion correctly predicted as positive by the model;

[0165] F1 Score: The harmonic mean of precision and recall, used to comprehensively measure the performance of the model;

[0166] Confusion Matrix: A table used to show the correspondence between the model prediction results and the actual labels, which can further analyze the performance of the model.

[0167] In the embodiments of the present application, by calculating these indicators, the present application can comprehensively understand the performance of the model in different aspects, so as to determine the advantages of the model and the areas that need to be improved.

[0168] In the embodiments of the present application, after obtaining the model evaluation results, the present application needs to optimize the model according to these results to improve its accuracy, generalization ability or other performance metrics. Among them, the methods of model optimization include but are not limited to:

[0169] (1) Adjust model parameters: According to the evaluation results, adjust the hyperparameters of the model, such as learning rate, regularization parameter, number of iterations, etc., to improve the performance of the model;

[0170] Feature selection: Through feature selection algorithms, select the feature subset that contributes the most to the model performance from the original feature set to reduce the risk of noise and overfitting;

[0171] (2) Model ensemble: Combine the results of multiple models, such as using voting mechanisms or weighted averages, to improve the stability and accuracy of the model;

[0172] (3) Use a more complex model: If the performance of the current model is poor, consider using a more complex model architecture, such as a deep neural network, to capture more complex features and information in the text;

[0173] (4) Data augmentation: Through data augmentation techniques, such as synonym replacement, sentence restructuring, etc., increase the diversity and quantity of training data to improve the generalization ability of the model.

[0174] In the embodiments of the present application, by performing optimization operations on the model, a sentiment analysis model with better performance and stability can be obtained, which can be applied to actual scenarios to judge and analyze sentiment tendencies.

[0175] In some optional implementation manners of the embodiments of the present application, after the step of outputting the target product information to the user terminal according to the target recommendation strategy, the following steps are further included:

[0176] Obtain the user feedback information corresponding to the target product information sent by the user terminal;

[0177] Update the product database and the topic database according to the user feedback information.

[0178] In the embodiments of the present application, after recommending insurance products to customers, analyze the acceptance and satisfaction of customers with the recommended products, and iteratively optimize the recommendation strategy. Specifically:

[0179] (1) Embed feedback links or forms in the recommendation information to collect customers' evaluations and suggestions on the recommended products. Regularly communicate with customers to understand their satisfaction with the recommended products and changes in needs;

[0180] (2) Continuously optimize the recommendation strategy according to customer feedback and market changes. Update the keyword library and the topic library to maintain the timeliness and accuracy of the recommendation.

[0181] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, Artificial Intelligence (AI) is a 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.

[0182] Artificial intelligence basic technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. Artificial intelligence software technologies mainly include several major directions such as computer vision technology, robotics, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0183] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disc, a Read-Only Memory (ROM), or a Random Access Memory (RAM), etc.

[0184] It should be understood that although the steps in the flowchart of the accompanying drawings are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same moment, but can be executed at different moments, and their execution order is not necessarily sequential, but can be executed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.

[0185] Further reference Figure 3 to Figure 2 As an implementation of the method shown above, an embodiment of a product recommendation strategy generation device is provided in the present application. This device embodiment corresponds to the method embodiment shown in Figure 2 and can be specifically applied to various electronic devices.

[0186] As Figure 3 shown, the product recommendation strategy generation device 200 of the embodiments of the present application includes:

[0187] A consultation data acquisition module 210, configured to receive session consultation data sent by a user terminal;

[0188] A consultation feature extraction module 220, configured to perform a consultation feature extraction operation on the session consultation data to obtain consultation feature data;

[0189] A keyword extraction module 230, configured to perform a keyword extraction operation on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result;

[0190] A target product acquisition module 240, configured to call a product database and acquire target product information corresponding to the keyword extraction result in the product database;

[0191] A target topic acquisition module 250, configured to read a topic library and acquire target topic data corresponding to the session consultation data in the topic library;

[0192] A recommendation strategy construction module 260, configured to construct a target recommendation strategy according to the target product information and the target topic data;

[0193] A target product output module 270, configured to output the target product information to the user terminal according to the target recommendation strategy.

[0194] In an embodiment of the present application, there is provided a product recommendation strategy generation device 200, including: a consultation data acquisition module 210, configured to receive session consultation data sent by a user terminal; a consultation feature extraction module 220, configured to perform a consultation feature extraction operation on the session consultation data to obtain consultation feature data; a keyword extraction module 230, configured to perform a keyword extraction operation on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result; a target product acquisition module 240, configured to call a product database and acquire target product information corresponding to the keyword extraction result in the product database; a target topic acquisition module 250, configured to read a topic library and acquire target topic data corresponding to the session consultation data in the topic library; a recommendation strategy construction module 260, configured to construct a target recommendation strategy according to the target product information and the target topic data; a target product output module 270, configured to output the target product information to the user terminal according to the target recommendation strategy. Compared with the prior art, the present application can effectively utilize the chat content with customers to recommend insurance products, effectively improving the accuracy of recommendation and customer satisfaction.

[0195] In some optional implementation manners of the embodiment of the present application, the above product recommendation strategy generation device 200 further includes: a first preprocessing operation module, and the above consultation feature extraction module 220 includes: a consultation feature extraction sub-module, where:

[0196] The first preprocessing operation module is used to perform the first preprocessing operation on the session consultation data to obtain preprocessed consultation data;

[0197] The consultation feature extraction sub-module is used to perform consultation feature extraction operations on the preprocessed consultation data to obtain consultation feature data.

[0198] In some optional implementation manners of the embodiments of the present application, the above-mentioned recommendation strategy construction module 260 includes:

[0199] The user portrait acquisition sub-module is used to acquire the user portrait information of the consultation user corresponding to the session consultation data;

[0200] The recommendation strategy construction sub-module is used to construct a target recommendation strategy according to the user portrait information, the target product information, and the target topic data.

[0201] In some optional implementation manners of the embodiments of the present application, the above-mentioned product recommendation strategy generation device 200 further includes: a sentiment analysis module and a strategy optimization module. The above-mentioned target product output module 270 includes: a target product output sub-module, where:

[0202] The sentiment analysis module is used to call the trained sentiment analysis model, and input the session consultation data and the target topic data into the trained sentiment analysis model for sentiment analysis operations to obtain sentiment tendency data;

[0203] The strategy optimization module is used to optimize the target recommendation strategy according to the sentiment tendency data to obtain an optimized recommendation strategy;

[0204] The target product output sub-module is used to output the target product information to the user terminal according to the optimized recommendation strategy.

[0205] In some optional implementation manners of the embodiments of the present application, the above-mentioned product recommendation strategy generation device 200 further includes: a training text acquisition module, a second preprocessing module, a training feature extraction module, and a model training module, where:

[0206] The training text acquisition module is used to read the system database and acquire the training text data marked with sentiment labels in the system database;

[0207] The second preprocessing module is used to perform the second preprocessing operation on the training text data to obtain the preprocessed training text data;

[0208] The training feature extraction module is used to perform training feature extraction operations on the training text data to obtain training feature data;

[0209] A model training module, configured to build an initially trained sentiment analysis model, and input the preprocessed training text data and training feature data into the initially trained sentiment analysis model for model training operations, so as to obtain a trained sentiment analysis model.

[0210] In some optional implementation manners of the embodiments of the present application, the above product recommendation strategy generation device 200 further includes: a test text acquisition module, a model evaluation module, and a model optimization module, where:

[0211] The test text acquisition module is configured to acquire test text data corresponding to the training text data in the system database;

[0212] The model evaluation module is configured to perform model evaluation operations on the trained sentiment analysis model according to the test text data to obtain a model evaluation result;

[0213] The model optimization module is configured to perform model optimization operations on the trained sentiment analysis model according to the model evaluation result to obtain an optimized sentiment analysis model.

[0214] In some optional implementation manners of the embodiments of the present application, the above product recommendation strategy generation device 200 further includes: a feedback information acquisition module and a database update module, where:

[0215] The feedback information acquisition module is configured to acquire user feedback information corresponding to the target product information sent by the user terminal;

[0216] The database update module is configured to update the product database and the topic database according to the user feedback information.

[0217] To solve the above technical problems, embodiments of the present application further provide a computer device. For details, please refer to Figure 4 , Figure 4 which is the basic structural block diagram of the computer device in the embodiments of the present application.

[0218] The computer device 300 includes a memory 310, a processor 320, and a network interface 330 that are communicatively connected to each other via a system bus. It should be noted that only the computer device 300 with components 310-330 is shown in the figure, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Among them, those skilled in the art of the present technology can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0219] The computer device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The computer device can perform human-computer interaction with the user through means such as a keyboard, a mouse, a remote control, a touchpad, or a voice control device.

[0220] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, a hard disk, a multimedia card, a card-type memory (such as an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 310 may be an internal storage unit of the computer device 300, such as the hard disk or memory of the computer device 300. In other embodiments, the memory 310 may also be an external storage device of the computer device 300, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 300. Of course, the memory 310 may also include both the internal storage unit of the computer device 300 and its external storage device. In the embodiments of the present application, the memory 310 is generally used to store the operating system and various application software installed on the computer device 300, such as computer-readable instructions for the product recommendation strategy generation method. In addition, the memory 310 may also be used to temporarily store various types of data that have been output or will be output.

[0221] In some embodiments, the processor 320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor 320 is generally used to control the overall operation of the computer device 300. In the embodiments of the present application, the processor 320 is used to run the computer-readable instructions stored in the memory 310 or process data, such as running the computer-readable instructions of the product recommendation strategy generation method.

[0222] The network interface 330 may include a wireless network interface or a wired network interface, and this network interface 330 is generally used to establish a communication connection between the computer device 300 and other electronic devices.

[0223] The computer device provided by the present application can effectively utilize the chat content with customers to recommend insurance products, effectively improving the accuracy of recommendations and customer satisfaction.

[0224] The present application also provides another implementation manner, that is, to provide a computer-readable storage medium storing computer-readable instructions, and the computer-readable instructions can be executed by at least one processor, so that the at least one processor executes the steps of the product recommendation strategy generation method as described above.

[0225] The computer-readable storage medium provided by the present application can effectively utilize the chat content with customers to recommend insurance products, effectively improving the accuracy of recommendations and customer satisfaction.

[0226] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation manner. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disc) and includes several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in the various embodiments of the present application.

[0227] Obviously, the embodiments described above are only a part of the embodiments of this application, rather than all of them. The preferred embodiments of this application are shown in the accompanying drawings, but they do not limit the patent scope of this application. This application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosed content of this application more thorough and comprehensive. Although this application has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions described in the foregoing specific embodiments, or perform equivalent replacements for some of the technical features. Any equivalent structure made by using the content of this application's specification and drawings, directly or indirectly applied in other related technical fields, is equally within the scope of patent protection of this application.

Claims

1. A method for generating a product recommendation strategy, characterized in that: The steps include: Receiving session consultation data sent by a user terminal; Performing a consultation feature extraction operation on the conversation consultation data to obtain consultation feature data; Performing a keyword extraction operation on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result; Calling a product database, and acquiring target product information corresponding to the keyword extraction result in the product database; Reading a topic database, and acquiring target topic data corresponding to the conversation consultation data in the topic database; Constructing a target recommendation strategy based on the target product information and the target topic data; The target product information is output to the user terminal according to the target recommendation strategy.

2. The method for generating a product recommendation strategy according to claim 1, characterized in that: After the step of receiving the conversation consultation data sent by the user terminal and before the step of performing a consultation feature extraction operation on the conversation consultation data to obtain the consultation feature data, the following steps are also included: Performing a first preprocessing operation on the session consultation data to obtain preprocessed consultation data; The step of performing a consultation feature extraction operation on the conversation consultation data to obtain consultation feature data specifically includes the following steps: Perform a consultation feature extraction operation on the pre-processed consultation data to obtain the consultation feature data.

3. The product recommendation strategy generation method according to claim 1, characterized in that: The step of constructing a target recommendation strategy based on the target product information and the target topic data includes the following steps: Obtaining user portrait information of the consulting user corresponding to the conversation consulting data; A target recommendation strategy is constructed based on the user portrait information, the target product information and the target topic data.

4. The method for generating a product recommendation strategy according to claim 1, characterized in that: After the step of reading the topic database and acquiring the target topic data corresponding to the conversation consultation data in the topic database, and before the step of constructing a target recommendation strategy according to the target product information and the target topic data, the following steps are also included: Calling a trained sentiment analysis model, and inputting the conversation consultation data and the target topic data into the trained sentiment analysis model to perform sentiment analysis operations, and obtain sentiment tendency data; After the step of constructing a target recommendation strategy according to the target product information and the target topic data, the following steps are also included: Optimizing the target recommendation strategy according to the sentiment tendency data to obtain an optimized recommendation strategy; The step of outputting the target product information to the user terminal specifically includes the following steps: The target product information is output to the user terminal according to the optimization recommendation strategy.

5. The method for generating a product recommendation strategy according to claim 4, characterized in that: Before the step of calling the trained sentiment analysis model and inputting the conversation consultation data and the target topic data into the trained sentiment analysis model to perform sentiment analysis to obtain sentiment tendency data, the following steps are also included: Reading a system database, and obtaining training text data marked with emotion tags in the system database; Performing a second preprocessing operation on the training text data to obtain preprocessed training text data; Performing a training feature extraction operation on the training text data to obtain training feature data; An initially trained sentiment analysis model is constructed, and the preprocessed training text data and the training feature data are input into the initially trained sentiment analysis model for model training operation to obtain a trained sentiment analysis model.

6. The method for generating a product recommendation strategy according to claim 5, characterized in that: After the step of constructing the initially trained sentiment analysis model, and inputting the preprocessed training text data and the training feature data into the initially trained sentiment analysis model to perform a model training operation to obtain the trained sentiment analysis model, the following step is also included: Acquire test text data corresponding to the training text data in the system database; Performing a model evaluation operation on the trained sentiment analysis model according to the test text data to obtain a model evaluation result; According to the model evaluation result, a model optimization operation is performed on the trained sentiment analysis model to obtain an optimized sentiment analysis model.

7. The method for generating a product recommendation strategy according to claim 1, characterized in that: After the step of outputting the target product information to the user terminal according to the target recommendation strategy, the following steps are also included: Acquiring user feedback information corresponding to the target product information sent by the user terminal; The product database and the topic database are updated according to the user feedback information.

8. A product recommendation strategy generation device, characterized in that: include: A consultation data acquisition module, used to receive session consultation data sent by a user terminal; A consultation feature extraction module, used for performing a consultation feature extraction operation on the session consultation data to obtain consultation feature data; A keyword extraction module, used to perform a keyword extraction operation on the consultation feature data according to a keyword extraction algorithm to obtain a keyword extraction result; A target product acquisition module, used to call a product database and acquire target product information corresponding to the keyword extraction result in the product database; A target topic acquisition module, used for reading a topic database and acquiring target topic data corresponding to the conversation consultation data in the topic database; A recommendation strategy building module, used to build a target recommendation strategy based on the target product information and the target topic data; A target product output module is used to output the target product information to the user terminal according to the target recommendation strategy.

9. A computer device comprising a memory and a processor, characterized in that: The memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the product recommendation strategy generation method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the product recommendation strategy generation method according to any one of claims 1 to 7 are implemented.

Images