Intelligent customer mining method and device, computer equipment and storage medium

Abstract

The invention belongs to the field of artificial intelligence, and relates to an intelligent customer mining method, which comprises the following steps: acquiring standard customer data including position information, attribute information and behavior information of customers; integrating the position information with a geographic information system to obtain geographic position information of the position information on a preset map; according to the attribute information and the behavior information, utilizing a product recommendation model to extract attribute features and behavior features, and determining a target recommended product according to the attribute features and the behavior features; according to the target recommended product, generating a contact verbal skill corresponding to the target recommended product by utilizing a contact verbal skill generation model; and when the target customer is selected on the preset map, displaying the product recommendation information and the customer contact verbal skill of the target customer. In addition, the invention also relates to a block chain technology, and client data and the like can be stored in a block chain. The invention further provides an intelligent customer mining device, computer equipment and a storage medium. The client mining efficiency and success rate can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and in particular to an intelligent customer mining method. Background Art

[0002] In today's ever-changing business environment, customer mining has become an indispensable core element in marketing and sales strategies, and its strategic value is becoming increasingly significant. Traditional customer mining methods mainly rely on salesmen to find potential customers through visits and telemarketing, which lacks customer targeting, resulting in low efficiency and high time cost of customer mining;

[0003] With the vigorous development of artificial intelligence technology, keyword search and type search have emerged, providing strong support for enterprises to screen potential customers and optimizing the customer mining process to a certain extent;

[0004] However, the limitations of customer mining methods based on keyword search and type search are becoming increasingly obvious, especially in the geographic space dimension. It is unable to accurately meet the needs of acquiring customers nearby, causing customers to often spend a lot of time and energy in finding potential customers. Even if they contact potential customers, there are no recommendations for customer contact speech, resulting in low efficiency and success rate of customer mining. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose an intelligent customer mining method, device, computer equipment and storage medium, the main purpose of which is to improve the accuracy and success rate of customer mining.

[0006] In order to solve the above technical problems, the present application embodiment provides an intelligent customer mining method, which adopts the following technical solutions:

[0007] Acquire standard customer data, wherein the standard customer data includes the customer's location information, attribute information, and behavior information;

[0008] Integrate the location information with a geographic information system to obtain geographic location information of the location information on a preset map;

[0009] Extracting attribute features and behavior features using a pre-trained product recommendation model based on the attribute information and the behavior information, and determining a target recommended product based on the attribute features and the behavior features;

[0010] According to the target recommended product, a touch customer speech corresponding to the target recommended product is generated using a pre-trained touch customer speech generation model;

[0011] When a target customer is selected on the preset map, the product recommendation information and customer contact words of the target customer are displayed visually.

[0012] Furthermore, before integrating the location information with the geographic information system, the method further includes:

[0013] Verifying the accuracy of the location information according to preset accuracy verification rules;

[0014] When the accuracy verification fails, the reason for the verification failure is obtained, and the location information is corrected according to the reason for the verification failure to obtain the location information that passes the accuracy verification.

[0015] Furthermore, the integrating the location information with a geographic information system to obtain the geographic location of the location information on a preset map includes:

[0016] Identifying an initial data format of the location information, and determining whether the initial data format is a standard data format;

[0017] When it is identified that the initial data format is not the standard data format, converting the initial data format of the location information into the standard data format according to a predefined data format;

[0018] The location information in the standard data format is imported into the geographic information system using a preset import function, and the geographic location of the location information on the preset map is calculated according to a preset matching algorithm and the preset map corresponding to the geographic information system.

[0019] Furthermore, extracting attribute features and behavior features using a pre-trained product recommendation model based on the attribute information and the behavior information includes:

[0020] Converting the attribute information into an attribute vector using the deep neural network of the pre-trained product recommendation model;

[0021] Using neurons of the deep neural network to extract features from the attribute vector to obtain the attribute features;

[0022] Converting the behavior information into a behavior vector using the convolutional neural network of the pre-trained product recommendation model;

[0023] A convolution operation is performed on the behavior vector to obtain the behavior feature.

[0024] Furthermore, determining the target recommended product according to the attribute characteristics and behavior characteristics includes:

[0025] According to a preset weighted weight, a weighted fusion operation is performed on the attribute feature and the behavior feature to obtain a fusion feature;

[0026] Calculate the distance between the fusion feature and each product feature in the preset product list to obtain the product recommendation priority;

[0027] According to the product recommendation priority, the preset product list is sorted from high to low, and the product with the highest product recommendation priority is selected from the sorted preset product list as the target recommended product.

[0028] Furthermore, the step of generating a touch customer speech corresponding to the target recommended product by using a pre-trained touch customer speech generation model according to the target recommended product includes:

[0029] Obtain a key information set of the target recommended product, and use the pre-trained customer contact speech generation model to convert each key information of the key information set into a key vector;

[0030] Generating at least one candidate customer contact speech according to the key vector;

[0031] Calculate the speech score of each candidate customer contact speech, and select the candidate customer contact speech with the highest speech score as the customer contact speech corresponding to the target recommended product.

[0032] Furthermore, the step of calculating the speech score of each candidate customer contact speech and selecting the candidate customer contact speech with the highest speech score as the customer contact speech corresponding to the target recommended product includes:

[0033] Extracting candidate keywords of each of the candidate customer contact words, and converting the candidate keywords into candidate keyword vectors;

[0034] Calculate the similarity between the key vector and the candidate keyword vector to obtain the skill score corresponding to each candidate customer contact skill;

[0035] According to the speech score corresponding to each of the candidate customer contact speech, all candidate customer contact speech are sorted from high to low according to the speech score;

[0036] The candidate customer contact speech corresponding to the highest speech score is used as the customer contact speech corresponding to the target recommended product.

[0037] In order to solve the above technical problems, the embodiment of the present application also provides an intelligent customer mining device, which adopts the following technical solution:

[0038] An acquisition module, used to acquire standard customer data, wherein the standard customer data includes the customer's location information, attribute information, and behavior information;

[0039] An integration module, used to integrate the location information with a geographic information system to obtain geographic location information of the location information on a preset map;

[0040] A recommendation module, configured to extract attribute features and behavior features based on the attribute information and the behavior information using a pre-trained product recommendation model, and determine a target recommended product based on the attribute features and the behavior features;

[0041] A generation module, configured to generate a customer touch speech corresponding to the target recommended product based on the target recommended product using a pre-trained customer touch speech generation model;

[0042] The visualization module is used to visually display the product recommendation information and customer contact words of the target customer when the target customer is selected on the preset map.

[0043] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0044] Acquire standard customer data, wherein the standard customer data includes the customer's location information, attribute information, and behavior information;

[0045] Integrate the location information with a geographic information system to obtain geographic location information of the location information on a preset map;

[0046] Extracting attribute features and behavior features using a pre-trained product recommendation model based on the attribute information and the behavior information, and determining a target recommended product based on the attribute features and the behavior features;

[0047] According to the target recommended product, a touch customer speech corresponding to the target recommended product is generated using a pre-trained touch customer speech generation model;

[0048] When a target customer is selected on the preset map, the product recommendation information and customer contact words of the target customer are displayed visually.

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

[0050] Acquire standard customer data, wherein the standard customer data includes the customer's location information, attribute information, and behavior information;

[0051] Integrate the location information with a geographic information system to obtain geographic location information of the location information on a preset map;

[0052] Extracting attribute features and behavior features using a pre-trained product recommendation model based on the attribute information and the behavior information, and determining a target recommended product based on the attribute features and the behavior features;

[0053] According to the target recommended product, a touch customer speech corresponding to the target recommended product is generated using a pre-trained touch customer speech generation model;

[0054] When a target customer is selected on the preset map, the product recommendation information and customer contact words of the target customer are displayed visually.

[0055] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0056] In this embodiment, by acquiring standard customer data including attribute information, behavior information, and geographic information, the geographical area where the customer is located can be accurately located, and the efficiency of customer mining can be improved based on the geographical location distribution; based on the behavior information and attribute information, the accuracy of customer mining can be improved, thereby improving the efficiency of customer mining.

[0057] The location information is integrated with a geographic information system. Through the geographic information system, the location information can be intuitively displayed on a map, allowing users to know the location distribution information and to perform spatial queries to quickly find the location distribution of a specific area, providing a basis for subsequent customer mining.

[0058] According to the attribute information and the behavior information, a pre-trained product recommendation model is used to extract attribute features and behavior features, and target recommended products are determined according to the attribute features and behavior features. Based on multi-dimensional analysis, the accuracy of product recommendations can be improved, thereby improving the efficiency and success rate of customer mining.

[0059] According to the target recommended product, the customer contact words corresponding to the target recommended product are generated using a pre-trained customer contact words generation model. This can automatically generate customer contact words corresponding to the target recommended product, reducing the time and cost of manual word writing, thereby improving the efficiency of characterization mining; and the customer contact words generated based on the target recommended product can increase users' interest in the product, thereby improving the success rate of customer mining.

[0060] By visually displaying the product recommendation information and customer contact words of the target customers, it is possible to display a large number of potential customers on the map and display the target recommended products and the corresponding customer contact words to improve the efficiency and success rate of customer contact; through customer type screening and map zoom functions, it is possible to quickly lock in the target customer group and display the customer distribution area, thereby improving the efficiency and success rate of customer mining. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. 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.

[0062] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0063] Figure 2 A flowchart of an embodiment of an intelligent customer mining method according to the present application;

[0064] Figure 3 It is a structural schematic diagram of an embodiment of an intelligent customer mining device according to the present application;

[0065] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein 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-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0067] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0068] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0069] like Figure 1As 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 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for 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 optical fiber cables.

[0070] The user can 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 can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0071] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to a laptop computer 1011, a tablet computer 1012 or a mobile phone 1013, the terminal device 101 can 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), a laptop computer, a desktop computer, etc.

[0072] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .

[0073] It should be noted that the intelligent customer mining method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the intelligent customer mining device is generally set in the server / terminal device.

[0074] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0075] Continue to refer Figure 2 , showing a flow chart of an embodiment of an intelligent customer mining method according to the present application.

[0076] In this embodiment, the purpose is that when a user opens a preset map, he clicks on the customer logo on the preset map to automatically recommend the target recommended products corresponding to the customer and the customer contact words corresponding to the target recommended products; the customer distribution range can also be displayed on the preset map through a search engine to improve the efficiency and success rate of customer mining.

[0077] The method for constructing the art font library comprises the following steps:

[0078] S201. Acquire standard customer data, where the standard customer data includes the customer's location information, attribute information, and behavior information.

[0079] In this embodiment, customer data is obtained from a preset customer database, and the customer data includes the customer's location information, attribute information and behavior information; the location information refers to the fixed address filled in by the customer, and the customer's address information can also be obtained in real time after obtaining the customer's consent; the attribute information refers to the customer's basic information, such as age, gender, etc.; the behavior information refers to the customer's operation information, such as browsing products, adding products to the shopping cart, and placing orders in the product APP, and can also be browsing financial products in the financial APP, purchasing financial products, etc., and can also be browsing short videos, articles, etc. in the social media APP, and liking short videos, articles, etc.; the customer data stored in the preset customer database can be obtained from within the company, for example, the information filled in by the customer when registering in the company's APP and some behavior information of the customer in the company's APP (such as browsing, adding to the shopping cart, placing an order, etc.), and can also be obtained from outside the company, for example, cooperating with a third-party data provider to purchase or subscribe to customer data; through AP The I interface obtains customer information on social media from social media platforms; obtains customer public data from public databases; shares customer data with partners; conducts market research and questionnaires, etc.; and can also extract customer data from public information on the Internet through web crawler technology.

[0080] In one embodiment, after obtaining the customer data, the method further includes:

[0081] Cleaning the customer data to delete duplicate customer data;

[0082] The deduplicated customer data is standardized to obtain standard customer data.

[0083] In this embodiment, since the preset customer database stores customer data from different sources and in different formats, some customer data may be duplicated. The customer data from different sources are integrated according to the unique identifier corresponding to the customer data to obtain integrated customer data, and the integrated customer data is cleansed and the duplicate customer data is deleted. Finally, the deduplicated customer data is standardized according to a preset data format to obtain standard customer data including location information, attribute information and behavior information.

[0084] In this embodiment, the acquired customer data is integrated to avoid incomplete customer data, which may lead to errors in the subsequent customer data analysis. The efficiency and accuracy of the subsequent customer data analysis are improved through data cleaning and standardization.

[0085] In an implementable example A, customer data is obtained from different sources, such as Xiao Ming's customer data obtained from source 1: attribute information, behavior information, and Xiao Ming's customer data obtained from source 2: location information, behavior information, and socio-economic information; customer data with the same unique identifier are integrated to obtain Xiao Ming's customer data: attribute information, behavior information, location information, behavior information, and socio-economic information; data cleaning operations are performed on the integrated data to obtain attribute information, behavior information, location information, and socio-economic information; the cleaned customer data is converted into the format of customer unique identifier: attribute information, behavior information, location information to obtain standard customer information: Xiao Ming's unique identifier, attribute information, behavior information, and location information.

[0086] In this embodiment, by acquiring standard customer data including attribute information, behavior information, and geographic information, the geographical area where the customer is located can be accurately located, and the efficiency of customer mining can be improved based on the geographical location distribution; based on the behavior information and attribute information, the accuracy of customer mining can be improved, thereby improving the efficiency of customer mining.

[0087] S202, integrating the location information with a geographic information system to obtain geographic location information of the location information on a preset map;

[0088] In this embodiment, location information is extracted from the standard data, and the location information may be the detailed address of the customer or the longitude and latitude coordinates of the customer, etc.; the geographic information system refers to the GIS geographic information system, which is a computer system for capturing, storing, checking, analyzing and displaying geographic data, and the GIS geographic information system can combine the user's geographic location information with other types of data (such as customer information); the preset map refers to a map that has been configured in the GIS system, or it may be a map customized during design, which is used to display the geographic location in the customer data, and when the customer is selected, display recommended product information and customer contact words.

[0089] In this embodiment, since the GIS geographic information system usually imports location information in the form of a file, the location information is first stored in a file. The file includes but is not limited to an Excel table, a CSV file or a Shapefile containing the location information. The location information is imported into the GIS geographic information system in the form of a file, and the imported location information is matched with the map data in the GIS geographic information system or the pre-connected map data, and the location information is spliced ​​onto the corresponding map. The customer's location information is visualized on the preset map through a visualization tool to obtain the geographic location information of the customer data on the preset map, and is optimized and analyzed on the map so that the customer's distribution, distance, etc. can be viewed on the map.

[0090] In one embodiment, before integrating the location information with a geographic information system, the method further comprises:

[0091] Verifying the accuracy of the location information according to preset accuracy verification rules;

[0092] When the accuracy verification fails, the reason for the verification failure is obtained, and the location information is corrected according to the reason for the verification failure to obtain the location information that passes the accuracy verification.

[0093] In this embodiment, since the location information input by some customers may contain typos or be incorrectly input, the location information is first verified for accuracy according to preset accuracy verification rules. When the accuracy verification passes, the location information is imported into the geographic information system. When the location information fails the accuracy verification, the reason for the failure of the accuracy verification is obtained, and the location information is corrected according to the reason for the failure of the accuracy verification.

[0094] Specifically, the preset accuracy verification rules include but are not limited to: comparison with known landmarks, comparison with a database, and verification using a verification tool, etc.; wherein, the selection of the comparison method can be selected according to the format of the location information. If the location information includes landmark data, the accuracy verification is performed on the application of the "comparison with known landmarks" method, by selecting known and accurate landmarks and performing consistency verification with the landmark data of the location information. If there is no deviation, the accuracy verification passes; if there is a deviation or error, the accuracy verification is determined to have failed, the cause of the deviation or the error is obtained, and corrections are made to the cause to obtain location information that can pass the accuracy verification.

[0095] In this embodiment, by verifying the accuracy of the location information, erroneous information can be effectively identified and corrected in a timely manner, thereby improving the reliability of the location information, providing a scientific decision-making basis for subsequent customer mining, and avoiding visualization of erroneous location information on subsequent maps.

[0096] In one embodiment, the integrating the location information with a geographic information system to obtain the geographic location of the location information on a preset map includes:

[0097] Identifying an initial data format of the location information, and determining whether the initial data format is a standard data format;

[0098] When it is identified that the initial data format is not the standard data format, converting the initial data format of the location information into the standard data format according to a predefined data format;

[0099] The location information in the standard data format is imported into the geographic information system using a preset import function, and the geographic location of the location information on the preset map is calculated according to a preset matching algorithm and the preset map corresponding to the geographic information system.

[0100] In this embodiment, the preset import function includes but is not limited to the import function of the geographic information system itself or an API interface, etc.; the preset matching algorithm refers to a map matching algorithm, and the map matching is the process of matching the input location information to a specific point on a preset map; firstly, the initial data format of the location information is identified, and it is determined whether the initial data format is a standard data format. When it is identified that the initial data format is not the standard data format, the location information in the initial data format is converted into a standard data format that can be integrated with the geographic information system. The standard data format includes but is not limited to a vector data format or a raster data format, such as Shapefile. In vector data format, location information is usually represented by a collection of shapes such as points, lines, and surfaces. These geometric shapes are composed of a series of coordinate points, each of which contains longitude (or X coordinate) and latitude (or Y coordinate) information. The generated location information can also be added with additional attribute information, such as customer attribute information and behavior information, to facilitate subsequent analysis through attribute information and behavior information. The location information is imported into the geographic information system through a preset import function (the import function of the geographic information itself or the API interface), and the geographic location of the location information on the preset map is calculated from the geographic information system according to the preset map corresponding to the geographic information system and the map matching algorithm.

[0101] Integrate location information in standard data format with the geographic information system, and based on the integration results and visualization tools (including visualization tools that come with geographic information), intuitively display the customer's location information on a preset map.

[0102] In this embodiment, by determining whether the initial data format of the location information is a standard data format, when the initial data format is not a standard data format, converting the initial data format into a standard data format according to a predefined data format, data compatibility can be improved to avoid the failure of integrating the location information with the geographic information system due to data format errors; simplifying the data processing process, thereby reducing data errors and anomalies, and thus improving integration efficiency; by importing the location information into the geographic information system, the location information can be displayed intuitively, and other data can be displayed subsequently, thereby improving the extensibility of the location information in the preset map.

[0103] Continuing with the above-mentioned implementable example A, after obtaining Xiao Ming's standard customer data, Xiao Ming's location information (xx floor, xx building, xx district, xx city, xx province) is extracted from the standard customer data, the data format of the location information is identified, and the corresponding accuracy verification rule is selected for verification according to the data format. When Xiao Ming's location information passes the accuracy verification, the location information is imported into the geographic information system, Xiao Ming's data format is identified, and the location information is converted into standard location information according to the data format to obtain standard location information (longitude, latitude), and the GIS geographic information system is integrated according to the standard location information, and Xiao Ming's location information is visualized on the preset map using visualization tools and the preset map associated with the GIS geographic information system.

[0104] In order to further improve the accuracy of location information on the preset map and increase the efficiency and success rate of customer mining;

[0105] In another embodiment, after calculating the geographical location of the location information on the preset map, the method further includes:

[0106] When the real-time location information of the customer data is obtained, the geographical location on the preset map is updated in real time using the real-time location information.

[0107] In this embodiment, after calculating the geographical location of the location information of the customer data on the preset map, a real-time location acquisition request is sent to the user. When the customer clicks to agree to obtain the real-time location acquisition request, the customer's real-time location information is obtained, and the real-time location information is integrated with the geographic information system, so that the geographical location on the preset map is updated in real time.

[0108] In this embodiment, by obtaining the user's real-time location information, the situation in which the user is not in the geographical location and thus causes an erroneous display is avoided, thereby further improving the accuracy of the geographical location and thus improving the efficiency and success rate of customer mining.

[0109] It is worth mentioning that in this embodiment, according to the characteristics of the geographic information system, the corresponding behavior information and attribute information are also imported into the geographic information system while the location information is imported into the geographic information system. When the user clicks on the customer's mark in the preset map, the attribute information and behavior information of the customer can be analyzed, and the recommended products and corresponding customer contact words can be displayed.

[0110] In this embodiment, the location information is integrated with a geographic information system. Through the geographic information system, the location information can be intuitively displayed on a map, so that users can know the location distribution information and can also perform spatial queries to quickly find the location distribution of a specific area, providing a basis for subsequent customer mining.

[0111] S203: extract attribute features and behavior features using a pre-trained product recommendation model based on the attribute information and the behavior information, and determine a target recommended product based on the attribute features and the behavior features.

[0112] In this embodiment, the product recommendation model is composed of a deep neural network and a convolutional neural network; the deep neural network is a fully connected network, and the deep neural network includes an input layer, an output layer and a hidden layer; the convolutional neural network includes an input layer, a convolution layer, a maximum pooling layer and an output layer; the attribute information includes but is not limited to name, gender, age, height, weight, etc.; the behavior information includes but is not limited to historical purchase data, historical browsing data, etc.

[0113] In this embodiment, the attribute features of the attribute information are extracted by a deep neural network of a pre-trained product recommendation model, the behavioral features of the behavioral information are extracted by a convolutional neural network of a pre-trained product recommendation model, the product recommendation priority is calculated based on the attribute features and the behavioral features, and the target recommended product is determined based on the product recommendation priority.

[0114] In one embodiment, extracting attribute features and behavior features using a pre-trained product recommendation model based on the attribute information and the behavior information includes:

[0115] Converting the attribute information into an attribute vector using the deep neural network of the pre-trained product recommendation model;

[0116] Using neurons of the deep neural network to extract features from the attribute vector to obtain the attribute features;

[0117] Converting the behavior information into a behavior vector using the convolutional neural network of the pre-trained product recommendation model;

[0118] A convolution operation is performed on the behavior vector to obtain the behavior feature.

[0119] In this embodiment, the attribute information is mapped from the semantic space to the vector space by using the embedding technology through the input layer of the deep neural network to obtain the attribute vector, the attribute vector is feature extracted by the neurons of the hidden layer of the deep neural network, and the attribute features are output through the output layer;

[0120] Through the input layer of the convolutional neural network, the embedding technology is used to map the behavior information from the semantic space to the vector space to obtain the behavior vector, and the convolution layer of the convolutional neural network is used to perform a convolution operation on the behavior vector to obtain a convolution vector, and the convolution vector is pooled using the maximum pooling layer of the convolutional neural network. The convolution vector is divided into several sub-regions in the dimension of the horizontal axis, the several sub-regions are downsampled, and then the maximum value is taken, and the behavior features are output through the output layer.

[0121] In this embodiment, the advantages of deep neural networks and convolutional neural networks are combined, and different neural networks are used for feature extraction based on the characteristics of attribute information and behavior information. This can extract richer features of attribute information and behavior information, thereby improving the accuracy of subsequent product recommendations.

[0122] In one embodiment, determining the target recommended product according to the attribute characteristics and the behavior characteristics includes:

[0123] According to a preset weighted weight, a weighted fusion operation is performed on the attribute feature and the behavior feature to obtain a fusion feature;

[0124] Calculate the distance between the fusion feature and each product feature in the preset product list to obtain the product recommendation priority;

[0125] According to the product recommendation priority, the preset product list is sorted from high to low, and the product with the highest product recommendation priority is selected from the sorted preset product list as the target recommended product.

[0126] In this embodiment, the preset product list refers to all products to be recommended; attribute features and behavior features are normalized to ensure that the values ​​of the attribute features and behavior features are within a comparable range; the normalized attribute features and behavior features are weighted and fused according to preset weighted weights to obtain fused features, and the preset weighted weights can be analyzed according to historical data; the feature vector of each product in the preset product list is extracted, and the distance value between the fused feature and the feature vector of each product is calculated by a preset calculation method (such as a Euclidean distance calculation method, a cosine similarity calculation method, etc.), the smaller the distance value, the higher the recommendation degree of the product, the priority of each product is determined according to the distance value, and according to the priority of each product, each product in the preset product list is sorted in order of priority from high to low to obtain the sorted preset product list, and the product with the highest priority is selected from the sorted preset product list as the target recommended product.

[0127] In this embodiment, the attribute features and behavioral features are weighted and fused through preset weighted weights to obtain fused features, the product recommendation priority is calculated based on the fused features, and the target recommended products are determined according to the product recommendation priority. Personalized recommendations can be made based on customer information, thereby improving the accuracy of product recommendations.

[0128] Continuing with the above-mentioned feasible example A, obtain attribute information including Xiao Ming's name, gender, age, job, education level, etc., as well as obtain Xiao Ming's behavior information that has purchased medium-risk financial products and browsed high-risk financial products, perform feature extraction on the attribute information through the deep neural network of a pre-trained product recommendation model to obtain attribute features, perform feature extraction on the behavior information through the convolutional neural network of a pre-trained product recommendation model to obtain behavior features, perform weighted fusion of the attribute features and the behavior features through pre-defined weighted weights to obtain fused features; calculate the distance value between the fused features and each financial product in the financial product recommendation table, determine the corresponding product recommendation priority according to the distance value, sort the financial product recommendation table from high to low according to the product recommendation priority, and select the medium-risk financial product with the highest product recommendation priority as the target recommended product.

[0129] In one embodiment, before extracting attribute features and behavior features using the pre-trained product recommendation model, the model needs to be trained. The specific training steps are as follows:

[0130] Construct an initial product recommendation model composed of a convolutional neural network and a deep neural network, obtain a large amount of labeled user attribute training data (such as name, gender, age, etc.) and behavior training data (historical purchase data, historical browsing data, etc.), perform preprocessing operations on the obtained training data, divide the preprocessed training data into a training set and a validation set according to a preset ratio, input the preprocessed training set into the initial product recommendation model, use the attribute information of the training set to train the deep neural network, use the behavior information of the training set to train the convolutional neural network, define a loss function (such as a cross entropy loss function, etc.), and complete the training when the loss function is less than a preset loss threshold. Use the validation set to verify the initial product recommendation model. When the prediction accuracy reaches the preset accuracy threshold, obtain the product recommendation model. When the preset accuracy does not reach the preset accuracy threshold, re-use the training set to train the initial product recommendation model until the prediction accuracy reaches the preset accuracy threshold to obtain the product recommendation model.

[0131] In this embodiment, based on the attribute information and the behavior information, a pre-trained product recommendation model is used to extract attribute features and behavior features, and target recommended products are determined based on the attribute features and behavior features. Multi-dimensional analysis is performed to improve the accuracy of product recommendations, thereby improving the efficiency and success rate of customer mining.

[0132] S204: Based on the target recommended product, a pre-trained customer contact speech generation model is used to generate customer contact speech corresponding to the target recommended product.

[0133] In this embodiment, the pre-trained touch customer speech generation model refers to the BERT model, which is a pre-trained language representation model based on the Transformer s architecture, and processes text through a self-attention mechanism and a bidirectional encoder.

[0134] In this embodiment, key information of the target recommended product is extracted, the key information is normalized, and the normalized key information is processed using a pre-trained customer contact script generation model to generate a customer contact script corresponding to the target recommended product. Users can conduct customer mining based on the customer contact script.

[0135] Continuing with the above feasible example A, when it is determined that the target recommended product for the customer is a medium-risk financial product, the key information of the medium-risk financial product is obtained: medium risk, 12-month investment period, expected rate of return of 4.5%, and a minimum investment of at least RMB 10,000. The key information is normalized to obtain a fixed format:

[0136] Product name: Steady growth financial plan

[0137] Risk level: Medium risk

[0138] Investment period: 12 months

[0139] Expected rate of return: 4.5%

[0140] Investment threshold: RMB 10,000

[0141] Input key information into the pre-trained customer contact script generation model to generate at least one candidate script, calculate the script score of each candidate script, and select the candidate script with the highest script score "Our steady growth financial plan is a medium-risk financial product with an investment period of 12 months and an expected rate of return of up to 4.5%. The minimum investment is only 10,000 yuan" as the customer contact script corresponding to the medium-risk financial product.

[0142] In one embodiment, the step of generating a touch customer speech corresponding to the target recommended product by using a pre-trained touch customer speech generation model according to the target recommended product includes:

[0143] Obtain a key information set of the target recommended product, and use the pre-trained customer contact speech generation model to convert each key information of the key information set into a key vector;

[0144] Generating at least one candidate customer contact speech according to the key vector;

[0145] Calculate the speech score of each candidate customer contact speech, and select the candidate customer contact speech with the highest speech score as the customer contact speech corresponding to the target recommended product.

[0146] In this embodiment, multiple key information of the target recommended product is obtained and collected to obtain a key information set, and each key information in the key information set is converted into a vector representation through the BERT embedding technology to obtain a key vector; the key vector is encoded through the encoding layer of a pre-trained touch customer speech generation model to obtain a key encoding vector, and the language features and semantic information of the key encoding vector are extracted. Based on the language features and semantic information, at least one candidate touch customer speech is generated through a decoder of a pre-trained touch customer speech generation model. For each candidate touch customer speech, the candidate keywords of each candidate touch customer speech are extracted, and the candidate keywords are converted into candidate keyword vectors. By calculating the similarity between the candidate keyword vector and the key vector, the speech score of each candidate touch customer speech is calculated respectively, and according to the speech score, each candidate touch customer speech is ranked, and the candidate touch customer speech with the highest candidate score is selected as the touch customer speech.

[0147] In this embodiment, customer contact words are generated based on the key information of the target recommended product to ensure that the words are relevant to the product features, thereby improving the relevance and attractiveness of the words. In addition, through the customer contact words generation model, the words generation time can be reduced, the accuracy and efficiency of the words generation can be improved, and the efficiency and success rate of customer mining can be improved.

[0148] In another embodiment, calculating the speech score of each candidate customer contact speech, and selecting the candidate customer contact speech with the highest speech score as the customer contact speech corresponding to the target recommended product includes:

[0149] Extracting candidate keywords of each of the candidate customer contact words, and converting the candidate keywords into candidate keyword vectors;

[0150] Calculate the similarity between the key vector and the candidate keyword vector to obtain the skill score corresponding to each candidate customer contact skill;

[0151] According to the speech score corresponding to each of the candidate customer contact speech, all candidate customer contact speech are sorted from high to low according to the speech score;

[0152] The candidate customer contact speech corresponding to the highest speech score is used as the customer contact speech corresponding to the target recommended product.

[0153] In this embodiment, the keywords of each candidate touch customer speech are extracted through the keyword extraction algorithm of natural language processing technology (NLP) (such as TF-IDF, TextRank, etc.), and all the keywords are converted into vector representations to obtain the candidate keyword vector of each candidate touch customer speech. The similarity between the candidate keyword vector and the original key vector is calculated by similarity calculation methods such as distance algorithms (such as Euclidean distance, pre-similarity) to obtain the speech score of each candidate touch customer speech. Based on the speech score of each candidate touch customer speech, all candidate touch customer speech are sorted from high to low according to the speech score, and the candidate touch customer speech corresponding to the highest speech score is selected as the touch customer speech corresponding to the target recommended product.

[0154] In this embodiment, by extracting keywords from candidate customer contact words and converting them into vectors, the core information in the words can be captured more accurately. Subsequently, the similarity between these keyword vectors and the key information of the target product is calculated to ensure that there is a high correlation between the selected customer contact words and the target product, thereby improving the accuracy of the match; using customer contact words that are closely related to the target product can more effectively convey the characteristics and advantages of the product, enhance users' cognition and interest in the product, and thus improve the efficiency and success rate of customer mining.

[0155] In this embodiment, based on the target recommended product, a pre-trained customer talk generation model is used to generate customer talk corresponding to the target recommended product. The customer talk corresponding to the target recommended product can be automatically generated, reducing the time and cost of manual writing of talk, thereby improving the efficiency of characterization mining; and the customer talk generated based on the target recommended product can increase user interest in the product, thereby improving the success rate of customer mining.

[0156] S205. When a target customer is selected on the preset map, product recommendation information and customer contact words for the target customer are visually displayed.

[0157] In this embodiment, geographic location information, target recommended products, and customer contact words are associated. When the location information is imported into the geographic information system, the attribute information and behavior information can also be imported at the same time. When the association is successful, when the user selects a target customer on a preset map, the product recommendation information and customer contact words of the target customer are visually displayed.

[0158] In this embodiment, a search engine and a screening tool may also be embedded in the preset map, and the customer distribution situation may be displayed based on the recommended information or attribute information of each target product through the screening tool.

[0159] Continuing with the above-mentioned feasible example A, when the user opens the preset map and clicks on Xiao Ming's avatar, the target recommended product corresponding to Xiao Ming is automatically displayed as: a medium-risk financial product; the customer contact wording is "Our steady growth financial plan is a medium-risk financial product with an investment period of 12 months and an expected rate of return of up to 4.5%, with a minimum investment of only 10,000 yuan", and the user can promote the target recommended product to Xiao Ming based on the customer contact wording.

[0160] In this embodiment, by visually displaying the product recommendation information and customer contact words of the target customers, a large number of potential customers can be displayed on the map and the target recommended products and corresponding customer contact words can be displayed to improve the efficiency and success rate of customer contact; through customer type screening and map zoom functions, the target customer group can be quickly locked in and the customer distribution area can be displayed, thereby improving the efficiency and success rate of customer mining.

[0161] It is worth mentioning that any privacy data mentioned above (such as customer information, etc.) must be used under the premise of legality and compliance.

[0162] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0163] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0164] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the 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-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0165] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, 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, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0166] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of an intelligent customer mining device, and the device embodiment is Figure 2 Corresponding to the method embodiment shown, the apparatus can be specifically applied to various computer devices.

[0167] like Figure 3 As shown, the intelligent customer mining device 300 described in this embodiment includes: an acquisition module 301, an integration module 302, a recommendation module 303, a generation module 304, and a visualization module 305. Among them:

[0168] An acquisition module 301 is used to acquire standard customer data, wherein the standard customer data includes the customer's location information, attribute information, and behavior information;

[0169] An integration module 302 is used to integrate the location information with a geographic information system to obtain geographic location information of the location information on a preset map;

[0170] A recommendation module 303 is used to extract attribute features and behavior features based on the attribute information and the behavior information using a pre-trained product recommendation model, and determine a target recommended product based on the attribute features and the behavior features;

[0171] A generation module 304 is used to generate a customer contact speech corresponding to the target recommended product by using a pre-trained customer contact speech generation model according to the target recommended product;

[0172] The visualization module 305 is used to visually display the product recommendation information and customer contact words of the target customer when the target customer is selected on the preset map.

[0173] In this embodiment, by acquiring standard customer data including attribute information, behavior information, and geographic information, the geographical area where the customer is located can be accurately located, and the efficiency of customer mining can be improved based on the geographical location distribution; based on the behavior information and attribute information, the accuracy of customer mining can be improved, thereby improving the efficiency of customer mining.

[0174] The location information is integrated with a geographic information system. Through the geographic information system, the location information can be intuitively displayed on a map, allowing users to know the location distribution information and to perform spatial queries to quickly find the location distribution of a specific area, providing a basis for subsequent customer mining.

[0175] According to the attribute information and the behavior information, a pre-trained product recommendation model is used to extract attribute features and behavior features, and target recommended products are determined according to the attribute features and behavior features. Based on multi-dimensional analysis, the accuracy of product recommendations can be improved, thereby improving the efficiency and success rate of customer mining.

[0176] According to the target recommended product, the customer contact words corresponding to the target recommended product are generated using a pre-trained customer contact words generation model. This can automatically generate customer contact words corresponding to the target recommended product, reducing the time and cost of manual word writing, thereby improving the efficiency of characterization mining; and the customer contact words generated based on the target recommended product can increase users' interest in the product, thereby improving the success rate of customer mining.

[0177] By visually displaying the product recommendation information and customer contact words of the target customers, it is possible to display a large number of potential customers on the map and display the target recommended products and the corresponding customer contact words to improve the efficiency and success rate of customer contact; through customer type screening and map zoom functions, it is possible to quickly lock in the target customer group and display the customer distribution area, thereby improving the efficiency and success rate of customer mining.

[0178] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0179] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 4 with a memory 41, a processor 42, and a network interface 43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art 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 (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0180] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0181] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the construction method of the art font library, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0182] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer-readable instructions stored in the memory 41 or process data, such as computer-readable instructions for running the intelligent customer mining method.

[0183] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0184] During the implementation of the electronic device of the present application, by acquiring standard customer data including attribute information, behavior information, and geographic information, it can accurately locate the geographical area where the customer is located, and improve the efficiency of customer mining based on the geographical location distribution; based on the behavior information and attribute information, it can improve the accuracy of customer mining, thereby improving the efficiency of customer mining. The location information is integrated with the geographic information system, and through the geographic information system, the location information can be intuitively displayed on the map, so that the user can know the information of the location distribution, and can also query in space to quickly find the location distribution of a specific area, providing a basis for subsequent customer mining. Based on the attribute information and the behavior information, the pre-trained product recommendation model is used to extract attribute features and behavior features, and the target recommended products are determined based on the attribute features and behavior features. Based on multi-dimensional analysis, the accuracy of product recommendations can be improved, thereby improving the efficiency and success rate of customer mining. According to the target recommended product, the customer contact words corresponding to the target recommended product are generated using a pre-trained customer contact words generation model. The customer contact words corresponding to the target recommended product can be automatically generated, reducing the time and cost of manually writing words, thereby improving the efficiency of characterization mining; and the customer contact words generated based on the target recommended product can increase user interest in the product, thereby improving the success rate of customer mining. By visually displaying the product recommendation information and customer contact words of the target customer, a large number of potential customers can be displayed on the map, and the target recommended products and corresponding customer contact words can be displayed to improve the efficiency and success rate of customer contact; through customer type screening and map zooming functions, the target customer group can be quickly locked in and the customer distribution area can be displayed, thereby improving the efficiency and success rate of customer mining.

[0185] The present application also provides another implementation, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the intelligent customer mining method as described above.

[0186] During the implementation process, the computer-readable storage medium of the present application can accurately locate the geographical area where the customer is located by obtaining standard customer data including attribute information, behavior information, and geographic information, and improve the efficiency of customer mining based on the geographical location distribution; based on the behavior information and attribute information, it can improve the accuracy of customer mining, thereby improving the efficiency of customer mining. The location information is integrated with the geographic information system, and through the geographic information system, the location information can be intuitively displayed on the map, so that the user can know the information of the location distribution, and can also query in space to quickly find the location distribution of a specific area, providing a basis for subsequent customer mining. Based on the attribute information and the behavior information, the pre-trained product recommendation model is used to extract attribute features and behavior features, and the target recommended products are determined based on the attribute features and behavior features. Based on multi-dimensional analysis, the accuracy of product recommendations can be improved, thereby improving the efficiency and success rate of customer mining. According to the target recommended product, the customer contact words corresponding to the target recommended product are generated using a pre-trained customer contact words generation model. The customer contact words corresponding to the target recommended product can be automatically generated, reducing the time and cost of manually writing words, thereby improving the efficiency of characterization mining; and the customer contact words generated based on the target recommended product can increase user interest in the product, thereby improving the success rate of customer mining. By visually displaying the product recommendation information and customer contact words of the target customer, a large number of potential customers can be displayed on the map, and the target recommended products and corresponding customer contact words can be displayed to improve the efficiency and success rate of customer contact; through customer type screening and map zooming functions, the target customer group can be quickly locked in and the customer distribution area can be displayed, thereby improving the efficiency and success rate of customer mining.

[0187] The non-Company software tools or components appearing in the embodiments of this application are merely examples and do not represent actual use.

[0188] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0189] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. An intelligent customer mining method, characterized in that: The steps include: Acquire standard customer data, wherein the standard customer data includes the customer's location information, attribute information, and behavior information; Integrate the location information with a geographic information system to obtain geographic location information of the location information on a preset map; Extracting attribute features and behavior features using a pre-trained product recommendation model based on the attribute information and the behavior information, and determining a target recommended product based on the attribute features and the behavior features; According to the target recommended product, a touch customer speech corresponding to the target recommended product is generated using a pre-trained touch customer speech generation model; When a target customer is selected on the preset map, the product recommendation information and customer contact words of the target customer are displayed visually.

2. The intelligent customer mining method according to claim 1, characterized in that: Before integrating the location information with the geographic information system, the method further comprises: Verifying the accuracy of the location information according to preset accuracy verification rules; When the accuracy verification fails, the reason for the verification failure is obtained, and the location information is corrected according to the reason for the verification failure to obtain the location information that passes the accuracy verification.

3. The intelligent customer mining method according to claim 1, characterized in that: The step of integrating the location information with a geographic information system to obtain the geographic location of the location information on a preset map includes: Identifying an initial data format of the location information, and determining whether the initial data format is a standard data format; When it is identified that the initial data format is not the standard data format, converting the initial data format of the location information into the standard data format according to a predefined data format; The location information in the standard data format is imported into the geographic information system using a preset import function, and the geographic location of the location information on the preset map is calculated according to a preset matching algorithm and the preset map corresponding to the geographic information system.

4. The intelligent customer mining method according to claim 1, characterized in that: The extracting attribute features and behavior features using a pre-trained product recommendation model according to the attribute information and the behavior information includes: Converting the attribute information into an attribute vector using the deep neural network of the pre-trained product recommendation model; Using neurons of the deep neural network to extract features from the attribute vector to obtain the attribute features; Converting the behavior information into a behavior vector using the convolutional neural network of the pre-trained product recommendation model; A convolution operation is performed on the behavior vector to obtain the behavior feature.

5. The intelligent customer mining method according to claim 1, characterized in that: The step of determining the target recommended product according to the attribute characteristics and the behavior characteristics includes: According to a preset weighted weight, a weighted fusion operation is performed on the attribute feature and the behavior feature to obtain a fusion feature; Calculate the distance between the fusion feature and each product feature in the preset product list to obtain the product recommendation priority; According to the product recommendation priority, the preset product list is sorted from high to low, and the product with the highest product recommendation priority is selected from the sorted preset product list as the target recommended product.

6. The intelligent customer mining method according to claim 1, characterized in that: The step of generating a touch customer speech corresponding to the target recommended product by using a pre-trained touch customer speech generation model according to the target recommended product includes: Obtain a key information set of the target recommended product, and use the pre-trained customer contact speech generation model to convert each key information of the key information set into a key vector; Generating at least one candidate customer contact speech according to the key vector; Calculate the speech score of each candidate customer contact speech, and select the candidate customer contact speech with the highest speech score as the customer contact speech corresponding to the target recommended product.

7. The intelligent customer mining method according to claim 6, characterized in that: The step of calculating the speech score of each candidate customer contact speech and selecting the candidate customer contact speech with the highest speech score as the customer contact speech corresponding to the target recommended product includes: Extracting candidate keywords of each of the candidate customer contact words, and converting the candidate keywords into candidate keyword vectors; Calculate the similarity between the key vector and the candidate keyword vector to obtain the skill score corresponding to each candidate customer contact skill; According to the speech score corresponding to each of the candidate customer contact speech, all candidate customer contact speech are sorted from high to low according to the speech score; The candidate customer contact speech corresponding to the highest speech score is used as the customer contact speech corresponding to the target recommended product.

8. An intelligent customer mining device, characterized in that: The device comprises: An acquisition module, used to acquire standard customer data, wherein the standard customer data includes the customer's location information, attribute information, and behavior information; An integration module, used to integrate the location information with a geographic information system to obtain geographic location information of the location information on a preset map; A recommendation module, configured to extract attribute features and behavior features based on the attribute information and the behavior information using a pre-trained product recommendation model, and determine a target recommended product based on the attribute features and the behavior features; A generation module, configured to generate a customer touch speech corresponding to the target recommended product based on the target recommended product using a pre-trained customer touch speech generation model; The visualization module is used to visually display the product recommendation information and customer contact words of the target customer when the target customer is selected on the preset map.

9. A computer device, characterized in that: The computer device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the intelligent customer mining method as described in any one of claims 1 to 7.

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

Images