CN118606559A

CN118606559A - Product recommendation method, device, equipment and storage medium

Info

Publication number: CN118606559A
Application number: CN202410794081.0A
Authority: CN
Inventors: 吴俣; 曹志豪; 叶俊伟; 郑贤茹; 赵煜祺
Original assignee: China Merchants Bank Co Ltd
Current assignee: China Merchants Bank Co Ltd
Priority date: 2024-06-19
Filing date: 2024-06-19
Publication date: 2024-09-06

Abstract

The present application relates to the field of data processing technologies, and in particular, to a product recommendation method, device, equipment, and storage medium. The method comprises the steps of classifying preset product data and client data to obtain multi-type data; carrying out standardization processing on the multi-type data to obtain characteristic standard data; then generating an initial customer portrait based on the feature standard data and the preset historical feature data; acquiring customer interaction data, and updating an initial customer portrait based on the customer interaction data and the potential demand data of the customer to obtain a target customer portrait; and finally, determining a target recommended product according to the target customer portrait. According to the method, the initial customer portrait is generated through data classification and standardization processing, the customer portrait is updated to reflect customer demands in real time, and finally, products are recommended based on target customer portraits, so that data processing consistency and recommendation accuracy are improved, high matching between recommended products and the customer demands is ensured, and the effect of product recommendation is improved.

Description

Product recommendation method, device, equipment and storage medium

Technical Field

The present application relates to the field of data processing technologies, and in particular, to a product recommendation method, device, equipment, and storage medium.

Background

The product recommendation process should be spread around consumer behavior and thus it is necessary to want to understand consumers in order to facilitate the execution of transactions. At present, analysis of product recommendation mainly depends on inputting relevant data into Excel, service personnel manually analyze and evaluate the service efficiency of product recommendation by looking up marketing data in the Excel, but a large amount of manpower and material resources are consumed in the mode, a large amount of labor cost is increased for enterprises, and proper product recommendation strategies cannot be predicted manually according to the data in the Excel, so that the effect of product recommendation is poor.

Therefore, how to improve the accuracy of product recommendation and the product recommendation effect is a problem to be solved at present.

The foregoing is provided merely for the purpose of facilitating understanding of the technical solutions of the present application and is not intended to represent an admission that the foregoing is prior art.

Disclosure of Invention

The application mainly aims to provide a product recommending method, device, equipment and storage medium, and aims to solve the technical problem of how to improve the accuracy of product recommendation and the effect of product recommendation.

In order to achieve the above object, the present application provides a product recommendation method, which includes the steps of:

classifying preset product data and customer data to obtain multi-type data;

Carrying out standardization processing on the multi-type data to obtain characteristic standard data;

generating an initial customer portrait based on the feature standard data and preset historical feature data;

acquiring customer interaction data, and updating the initial customer portrait based on the customer interaction data and the customer potential demand data to obtain a target customer portrait;

and determining a target recommended product according to the target customer portrait.

In one embodiment, the step of normalizing the multi-type data to obtain feature standard data includes:

performing data cleaning on the multi-type data to obtain cleaning data, wherein the data cleaning comprises one or more of missing value processing, repeated value processing and abnormal value processing;

Normalizing the cleaning data to obtain normalized data;

And integrating the normalized data into a preset formatting table to obtain the characteristic standard data.

In one embodiment, the step of generating the initial customer representation based on the feature standard data and the preset historical feature data includes:

Extracting data, of which the similarity with the characteristic standard data exceeds a preset similarity threshold value, in the preset historical characteristic data to obtain data to be fused;

Carrying out weighted fusion on the data to be fused and the characteristic standard data to obtain fusion data;

And carrying out feature extraction on the fusion data, and generating the initial customer portrait based on a feature extraction result and a preset portrait generation model.

In one embodiment, the step of obtaining customer interaction data and updating the initial customer representation based on the customer interaction data and the customer potential demand data to obtain a target customer representation comprises:

identifying the initial customer portrait and determining potential demand data of customers;

acquiring the customer interaction data, and carrying out the standardized processing on the customer interaction data and the customer potential demand data to obtain supplementary feature data;

based on the supplementary feature data, adjusting parameters and weights of the preset portrait generation model;

And obtaining the target customer portrait according to the adjusted preset portrait generation model, the supplementary feature data and the feature extraction result.

In one embodiment, the step of identifying the initial customer representation and determining customer potential demand data comprises:

Based on a preset feature recognition algorithm, performing feature recognition on the initial customer portrait to obtain behavior features, relationship features and preference features;

predicting the client behavior and the client demand according to the behavior characteristics, the relation characteristics and the preference characteristics to obtain behavior prediction data and demand prediction data;

And carrying out demand probability evaluation on the behavior prediction data and the demand prediction data, and taking the behavior prediction data or the demand prediction data corresponding to a probability evaluation result exceeding a preset demand probability threshold value as the potential demand data of the client.

In one embodiment, the step of determining a target recommended product based on the target customer representation includes:

Extracting key features of the target customer portrait to obtain key feature data;

Normalizing the key characteristic data and the preset product data to obtain normalized key characteristic data and normalized product data;

matching the normalized key feature data with each product feature data in the normalized product data;

And determining the target recommended product according to the product characteristic data corresponding to the matching degree higher than the preset matching degree threshold.

In one embodiment, after the step of determining a target recommended product based on the target customer representation, the method further comprises:

acquiring purchase detail information, and judging whether the purchase detail information comprises the target recommended product or not;

If not, acquiring customer feedback information, and determining a purchase decision influencing factor according to the customer feedback information;

And generating a product recommendation optimization strategy based on the purchase decision influencing factors and the key feature data.

In addition, in order to achieve the above object, the present application also provides a product recommendation device, including:

the classification module is used for classifying preset product data and client data to obtain multi-type data;

The standardized module is used for carrying out standardized processing on the multi-type data to obtain characteristic standard data;

the initial portrait generation module is used for generating an initial customer portrait based on the feature standard data and the preset historical feature data;

The portrait updating module is used for acquiring customer interaction data, updating the initial customer portrait based on the customer interaction data and the potential demand data of the customer, and obtaining a target customer portrait;

And the product recommending module is used for determining a target recommended product according to the target customer portrait.

In addition, in order to achieve the above object, the present application also proposes a product recommendation apparatus, the apparatus comprising: a memory, a processor, and a product recommendation program stored on the memory and executable on the processor, the product recommendation program configured to implement the steps of the product recommendation method as described above.

In addition, in order to achieve the above object, the present application also proposes a storage medium having stored thereon a product recommendation program which, when executed by a processor, implements the steps of product recommendation as described above.

The method comprises the steps of firstly classifying preset product data and client data to obtain multi-type data; then, carrying out standardization processing on the multi-type data to obtain characteristic standard data; then generating an initial customer portrait based on the feature standard data and the preset historical feature data; acquiring customer interaction data, and updating an initial customer portrait based on the customer interaction data and the potential demand data of the customer to obtain a target customer portrait; and finally, determining a target recommended product according to the target customer portrait. According to the application, the initial customer portrait is generated through data classification and standardization processing, and the customer portrait is dynamically updated to reflect the customer demands in real time, and finally the product is accurately recommended based on the target customer portrait, so that the data processing consistency and recommendation accuracy are improved, the highly-matched recommended product and the customer demands are ensured, and the product recommendation effect is improved.

Drawings

FIG. 1 is a schematic flow chart of a first embodiment of a product recommendation method according to the present application;

FIG. 2 is a schematic view of a sub-process in a second embodiment of the product recommendation method of the present application;

FIG. 3 is a schematic view of a second embodiment of a product recommendation method according to the present application;

FIG. 4 is a schematic view of a sub-process in a third embodiment of the product recommendation method of the present application;

FIG. 5 is a schematic view of a third embodiment of a product recommendation method according to the present application;

FIG. 6 is a diagram illustrating an exemplary system architecture of a product recommendation method according to an embodiment of the present application;

FIG. 7 is a diagram showing an example of data processing in an embodiment of a product recommendation method according to the present application;

FIG. 8 is a schematic block diagram of a product recommendation device according to an embodiment of the present application;

fig. 9 is a schematic device structure diagram of a hardware operating environment related to a product recommendation method in an embodiment of the present application.

The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.

For a better understanding of the technical solution of the present application, the following detailed description will be given with reference to the drawings and the specific embodiments.

It should be noted that the product recommendation process should be spread around consumer behavior, so it is necessary to want to understand consumers to facilitate the execution of transactions. At present, analysis of product recommendation mainly depends on inputting relevant data into Excel, service personnel manually analyze and evaluate the service efficiency of product recommendation by looking up marketing data in the Excel, but a large amount of manpower and material resources are consumed in the mode, a large amount of labor cost is increased for enterprises, and proper product recommendation strategies cannot be predicted manually according to the data in the Excel, so that the effect of product recommendation is poor. Therefore, how to improve the accuracy of product recommendation and the product recommendation effect is a problem to be solved at present.

The main solution of the application is as follows: classifying preset product data and client data to obtain multi-type data; then, carrying out standardization processing on the multi-type data to obtain characteristic standard data; then generating an initial customer portrait based on the feature standard data and the preset historical feature data; acquiring customer interaction data, and updating an initial customer portrait based on the customer interaction data and the potential demand data of the customer to obtain a target customer portrait; and finally, determining a target recommended product according to the target customer portrait.

According to the method, the initial customer portrait is generated through data classification and standardization processing, the customer portrait is dynamically updated to reflect customer demands in real time, and finally the product is accurately recommended based on the target customer portrait, so that the consistency of data processing and the accuracy of product recommendation are improved, the highly matched recommended product and the customer demands are ensured, the product recommendation effect is improved, and the marketing efficiency is improved.

It should be noted that, the execution body of the method of this embodiment may be a computing service device having functions of data processing, network communication and program running, or may be the product recommendation device having the same or similar functions. The present embodiment and the following embodiments will be described by taking a product recommendation apparatus as an example.

Based on this, a first embodiment of the product recommendation method of the present application is provided, and referring to fig. 1, fig. 1 is a schematic flow chart of the first embodiment of the product recommendation method of the present application.

In this embodiment, the product recommendation method includes the following steps:

s1: classifying preset product data and customer data to obtain multi-type data;

Specifically, collecting preset product data refers to collecting information about the bank product from a bank internal system or other data source. Such data includes product name, category, characteristics, price, applicable customer group, historical sales records, and the like. Collecting customer data refers to collecting basic information of enterprises such as industry information of the enterprises, regional information of the enterprises, enterprise registration capital information and the like.

It should be noted that, in this embodiment, the customer who performs product recommendation refers to a public customer.

Further, the multi-type data includes tag data, relationship data, and unstructured data. Tag data is structured data, such as values, categories, etc., that facilitates direct analysis and processing. Including basic information of the business, account balance, transaction frequency, etc. Relationship data is a data describing relationships between an enterprise and a product or other enterprise, such as products purchased by the enterprise, relationships between the enterprise and a customer manager, and the like. Unstructured data is text, images, etc. unstructured data such as feedback comments, chat records, etc. After classification, the classified multi-type data are stored in a unified database, so that the data organization structure is clear, and the subsequent processing and analysis are convenient.

The classified data structure is clear, so that standardized processing and subsequent analysis are facilitated, and the complexity of processing heterogeneous data is reduced. The data of different types can be stored in a classified mode, various data can be utilized more effectively, richer public customer information and product characteristics are provided, and a foundation is laid for accurate recommendation. By classifying and storing the structured, relational and unstructured data, multidimensional data analysis and mining can be performed, and the accuracy of portraits and the accuracy of a recommendation system are improved.

S2: carrying out standardization processing on the multi-type data to obtain characteristic standard data;

Specifically, the standardized processing includes data cleaning, data format unification, data normalization, feature extraction and generation, and feature merging and storage. Data cleansing refers to the process of processing missing values, repeated values, and outliers in data to ensure data quality. Data normalization refers to the process of converting data to a uniform scale for subsequent analysis and modeling. Feature extraction refers to the process of extracting useful features from raw data. Feature generation refers to the process of generating new features based on business requirements.

The missing value, the repeated value and the abnormal value are processed through data cleaning, so that the integrity and the accuracy of the data are ensured, and the data quality is improved. And unifying the data formats and types, ensuring that data of different sources and different types can be seamlessly integrated, and improving the consistency of the data. The normalized and encoded data is convenient for subsequent machine learning and data mining, and the stability and accuracy of the model are improved. Useful features are extracted and generated, feature information of data is enriched, and a solid foundation is provided for subsequent customer portrait generation and accurate recommendation.

S3: generating an initial customer portrait based on the feature standard data and preset historical feature data;

Specifically, customer characteristic data subjected to standardized processing is extracted from a database, and the data comprises basic information, behavior characteristics, relationship characteristics and the like of an enterprise. Preset customer history feature data is extracted from the history database, and the data generally comprises historical transaction records, past purchasing behavior, opinion feedback and the like of enterprises. Ensuring that the extracted feature standard data can be matched with the client identification (such as client ID) in the preset historical feature data correctly. And fusing the characteristic standard data with preset historical characteristic data to form a comprehensive data set containing the current characteristics and the historical characteristics of the client.

Further, key features that have an important influence on the generation of the customer representation, such as financial information, operation information, enterprise customer group information, legal compliance information, and the like, are selected from the fused dataset. The distribution and relation of the selected features are analyzed by using data analysis and statistical methods (such as correlation analysis, principal component analysis and the like), so that the accuracy and the effectiveness of the data are ensured. Clustering algorithms (e.g., K-means, hierarchical clustering, etc.) are applied to divide the clients into different groups, each group representing a class of clients with similar characteristics and patterns of behavior. By analyzing the characteristics of each customer group, the rationality of the grouping result is verified, and the customer grouping can be ensured to effectively reflect the difference of customers. And constructing a customer portrait model based on the feature extraction and grouping results. The model may employ a rules engine or machine learning algorithm (e.g., decision tree, random forest, etc.) to generate the customer representation. An initial customer representation is generated from the representation model, including basic information, behavioral characteristics, relational characteristics, preference characteristics, and the like of the customer.

Optionally, the generated initial customer representation is stored in a database for subsequent queries and applications. The initial customer portrayal can be applied to a plurality of scenes such as customer demand prediction, accurate marketing, personalized recommendation and the like, and the accuracy and the efficiency of business decision making are improved.

The feature standard data is the customer feature data after standardized processing, and has consistency and usability. The preset historical characteristic data is preset customer characteristic data in a historical database, and generally comprises historical transaction records and behavior data of customers. The customer representation is a customer description generated based on customer characteristic data reflecting the basic information, behavioral characteristics, and preference characteristics of the customer.

The current characteristic data and the historical characteristic data are combined, the generated customer portrait is more comprehensive and accurate, and the overall and behavior change of the customer can be reflected. Through cluster analysis, clients are divided into different groups, so that the differentiation and individuation requirements of the clients can be recognized, and the pertinence of marketing and service is improved. By adopting data analysis and machine learning algorithm, the initial customer portrait is generated quickly, and the efficiency and accuracy of portrait generation are improved. The generated initial client image can be applied to accurate marketing and personalized recommendation, and the client satisfaction degree and the marketing success rate are improved.

S4: and acquiring customer interaction data, and updating the initial customer portrait based on the customer interaction data and the customer potential demand data to obtain a target customer portrait.

It should be noted that the client interaction data includes online interaction data, offline interaction data, and the like. The online interaction data refers to interaction data of clients in online channels such as bank websites, mobile applications, emails and the like, and comprises browsing records, clicking records, search queries, online consultations and the like. The offline interaction data refers to interaction data of customers at banking sites or through telephone customer service, and comprises visit records, telephone records and interview records.

It should be noted that, the relevant data (e.g., online interaction data, offline interaction data, etc.) of the client interaction data related to the present application are all obtained after obtaining the permission or consent of the user; that is, when the present application is applied to a specific product or technology, it is necessary to obtain a user license to achieve the acquisition and processing of the related data, and the processing of the related data is required to comply with the related laws and regulations of the related country and region.

Specifically, the missing value, the repeated value and the abnormal value in the interactive data are processed, so that the data quality is ensured. The data format is unified, so that the interactive data of different sources can be integrated together for analysis. And integrating the online interaction data and the offline interaction data of the clients into a unified database. The integrated interaction data is correlated with the initial customer representation to ensure that all interaction data matches the correct customer. Based on the initial customer representation and the interaction data, potential needs of the customer are identified. Machine learning models (e.g., collaborative filtering, decision trees, etc.) are used to predict potential needs of customers. And classifying the identified potential demands of the clients to form the characteristic data of the demands of the clients.

Further, new interactive data and potential demand data are input into the customer representation model to update the initial customer representation. Key features are extracted from the new interactive data and are added to the customer representation. And optimizing a customer portrait model according to the new feature data, and ensuring the accuracy and the real-time performance of the portrait. A target customer representation is generated based on the updated customer representation model. The target customer representation should contain up-to-date customer characteristics, behavioral patterns, and demand characteristics. And the accuracy of the target client image is verified by means of feedback of a client manager, observation of client behaviors and the like, so that the latest requirements and behaviors of the client can be accurately reflected.

It should be noted that the customer interaction data refers to data generated when a customer interacts with a bank in various channels. The customer potential demand data is customer likely demand data predicted based on customer historical behavior and characteristics. The target customer portrait is a customer portrait generated by combining the latest customer interaction data and the potential demand data on the basis of the initial customer portrait, and can accurately reflect the current demand and behavior mode of the customer.

The customer image is dynamically updated, so that the latest requirements and behavior changes of the customer can be reflected in real time, and timeliness and accuracy of the image are ensured. And the latest interaction data and potential demand data are combined, so that a customer portrait model is optimized, and the accuracy and the comprehensiveness of the portrait are improved. The generated target customer portrait provides a reliable data base for accurate marketing and personalized service, and improves customer satisfaction and marketing success rate. The customer portraits are updated and optimized in real time, so that the bank can respond to the customer demands and market changes more quickly, and the agility and effect of business decisions are improved.

S5: and determining a target recommended product according to the target customer portrait.

Specifically, key features of the customer are extracted from the target customer representation, including basic information (e.g., business name, registered funds, industry category, registered address), behavioral features (e.g., transaction frequency, purchase history), preference features (e.g., product preferences, interests), etc. Clients are separated into different groups (e.g., high value clients, potential clients, frequently purchased clients, etc.) based on the extracted features, so that different recommendation strategies are formulated for each group. The detailed information of all products of the bank is collected, including product names, categories, functions, prices, applicable customer groups, historical sales records, etc. The extraction of key features from the product data ensures that it can be matched to the needs and preferences of the customer. For example, the risk level, the yield, the applicable scene, etc. of the product are extracted.

Further, a recommendation engine based on business rules is constructed, and products are matched according to preset rules (such as customer preference, historical purchasing behavior, product similarity and the like). Product recommendations are made using machine learning algorithms (e.g., collaborative filtering, matrix factorization, deep learning, etc.). The model may make recommendations based on historical behavior of the customer, purchase records of similar customers, product similarity, and so on.

Optionally, a recommended product list is generated based on the target customer representation and the product matching result. The recommendation list may be ordered by relevance, popularity, customer preference, etc. The recommendation reasons are generated for each recommended product, the reason is explained why the product is recommended, and the trust degree and the acceptance degree of the client on the recommendation result are increased. The recommended proposal is displayed by using a chart, a graph and other visualization tools, so that a customer can intuitively see the recommended product and the characteristics thereof. Friendly user interfaces are designed on the banking official network, mobile application and other platforms, recommended products and reasons are displayed, and customer experience is improved.

Further, feedback of the customer on the recommended products is collected, and satisfaction degree and actual requirements of the customer are known. Feedback means may include questionnaires, online evaluations, customer manager feedback, and so forth. And analyzing behavior data of the client on the recommended products, such as clicking, purchasing, browsing and the like, and evaluating the recommended effect. And continuously optimizing a recommendation algorithm and rules according to feedback and behavior data of the clients, and improving the accuracy and effect of the recommendation model.

Based on the latest portrait of the customer, products which are most suitable for the customer needs are intelligently matched and recommended, so that the recommended products are ensured to be highly matched with the customer needs, and the customer satisfaction degree and the marketing success rate are improved. By generating the personalized recommendation list and recommendation reasons, personalized experience of the client is improved, and trust and viscosity of the client to banking service are enhanced. And collecting customer feedback and behavior data, continuously optimizing recommendation algorithms and rules, ensuring that a recommendation system can adapt to the change of customer requirements in real time, and improving recommendation effect and accuracy. The automatic and intelligent recommendation flow reduces the workload of a customer manager, improves the efficiency and effect of business decision making, and helps banks to respond to market and customer demands more quickly.

In the embodiment, the preset product data and the client data are classified to obtain multi-type data; then, carrying out standardization processing on the multi-type data to obtain characteristic standard data; then generating an initial customer portrait based on the feature standard data and the preset historical feature data; acquiring customer interaction data, and updating an initial customer portrait based on the customer interaction data and the potential demand data of the customer to obtain a target customer portrait; and finally, determining a target recommended product according to the target customer portrait. According to the method, the initial customer portrait is generated through data classification and standardization processing, the customer portrait is dynamically updated to reflect customer demands in real time, and finally products are accurately recommended based on target customer portraits, so that data processing consistency and recommendation accuracy are improved, high matching between recommended products and the customer demands is ensured, product recommendation effect is improved, and marketing efficiency is improved.

Based on the first embodiment described above, a second embodiment of the product recommendation method of the present application is presented. Referring to fig. 2, fig. 2 is a schematic flow chart of a second embodiment of the product recommendation method according to the present application.

As shown in fig. 2, in the present embodiment, step S2 includes:

s21: performing data cleaning on the multi-type data to obtain cleaning data, wherein the data cleaning comprises one or more of missing value processing, repeated value processing and abnormal value processing;

s22: normalizing the cleaning data to obtain normalized data;

s23: and integrating the normalized data into a preset formatting table to obtain the characteristic standard data.

Specifically, the dataset is scanned, and records containing missing values are identified. The missing values are processed by deleting the missing value records, filling in the missing values or interpolation. Scanning the data set, identifying the repeated records, deleting the repeated records, and ensuring the uniqueness of each record. Identifying outliers using statistical methods (e.g., box graphs) or machine learning methods (e.g., isolated forest algorithms), and deleting the identified outlier records.

Further, scaling the numeric data to a uniform range, e.g., [0,1] range, converts the category data to numeric data. The processed numeric data and category data are integrated into a unified data table. And according to a preset formatting standard, ensuring the consistency of the formats of all fields in the data table, including data types, column name specifications and the like. And storing the integrated characteristic standard data into a database or a file, so that the subsequent analysis and use are convenient.

The data cleansing refers to a process of processing missing values, repeated values, and abnormal values in data to ensure data quality. Normalization refers to the process of scaling data to a uniform scale for subsequent analysis and modeling. The preset formatting table is a data table with unified format according to a preset standard.

And the missing value, the repeated value and the abnormal value are cleaned and processed by the data, so that the integrity and the accuracy of the data are ensured, and the data quality is improved. The normalization process scales the data to a uniform scale, unifies the data format, and ensures the consistency and availability of the data. The processed characteristic standard data format is uniform, so that subsequent machine learning and data mining are facilitated, and the stability and accuracy of the model are improved.

Based on the above-described first embodiment, in the present embodiment, step S3 includes:

S31: extracting data, of which the similarity with the characteristic standard data exceeds a preset similarity threshold value, in the preset historical characteristic data to obtain data to be fused;

s32: carrying out weighted fusion on the data to be fused and the characteristic standard data to obtain fusion data;

s33: and carrying out feature extraction on the fusion data, and generating the initial customer portrait based on a feature extraction result and a preset portrait generation model.

Specifically, a proper similarity measurement method is selected according to the data characteristics, such as euclidean distance, cosine similarity, pearson correlation coefficient and the like, and similarity calculation is performed on the characteristic standard data and the preset historical characteristic data to obtain a similarity value of each pair of data. And setting a similarity threshold according to service requirements, and extracting historical characteristic data with similarity exceeding a preset similarity threshold to form a data set to be fused.

Further, the fusion weight is defined according to the importance or reliability of the data, and a preset weight can be set or the weight can be learned through a model. And carrying out weighted calculation on the characteristic standard data and the data to be fused to generate fused data. Key features from the fused data that have an important impact on customer representation generation, such as customer age, gender, transaction frequency, product preferences, etc., are selected. The distribution and relation of the selected features are analyzed by using data analysis and statistical methods (such as correlation analysis, principal component analysis and the like), so that the accuracy and the effectiveness of the data are ensured. And constructing a customer portrait model based on the feature extraction result. The model may employ a rules engine or machine learning algorithm (e.g., decision tree, random forest, etc.) to generate the customer representation. An initial customer representation is generated from the representation model, including basic information, behavioral characteristics, relational characteristics, preference characteristics, and the like of the customer.

And the accuracy and the reliability of the initial customer portrait are improved by utilizing the weighted fusion of the similar historical characteristic data and the current characteristic data. The weighted fusion method can effectively integrate data from different sources, fully utilize useful information in historical data and improve the utilization rate and value of the data. Through the feature extraction technology, key features are extracted, so that more accurate and comprehensive customer portraits can be constructed, and data support is provided for follow-up accurate marketing and personalized services. The machine learning algorithm and the portrait generation model are adopted to automatically generate the initial customer portrait, so that the efficiency and accuracy of portrait generation are improved, and the manual intervention is reduced.

Referring to fig. 3, fig. 3 is a schematic view of a sub-process in a second embodiment of the product recommendation method according to the present application.

As shown in fig. 3, in the present embodiment, step S4 includes:

S41: identifying the initial customer portrait and determining potential demand data of customers;

Specifically, key features in the initial customer representation are analyzed, including basic information, behavioral features, and preference features. Based on the feature data in the representation, a demand prediction model (e.g., collaborative filtering, classification algorithms, etc.) is used to identify potential demands of the customer. And generating potential demand data of the client according to the output of the demand prediction model, and recording the types of products or services which the client may be interested in. The customer's potential demand data is classified to form structured demand feature data (customer potential demand data).

It should be noted that the demand prediction model is an algorithm model for predicting potential demands of clients based on client history data and feature data.

S42: acquiring the customer interaction data, and carrying out the standardized processing on the customer interaction data and the customer potential demand data to obtain supplementary feature data;

Specifically, online interaction data and offline interaction data of clients are collected, missing values, repeated values and abnormal values are processed, the integrity and the accuracy of the data are ensured, and invalid data are removed. And normalizing the interactive data and the demand data to ensure the consistency and comparability of the data. And integrating the standardized customer interaction data and the demand data to form the supplementary feature data.

S43: based on the supplementary feature data, adjusting parameters and weights of the preset portrait generation model;

S44: and obtaining the target customer portrait according to the adjusted preset portrait generation model, the supplementary feature data and the feature extraction result.

Specifically, the supplemental feature data is divided into a training set and a verification set for model adjustment and verification. And adjusting parameters and weights of the portrait generation model by using training set data, so as to ensure that the model can better adapt to new supplementary feature data. And evaluating the performance of the adjusted model by using the verification set data, and ensuring the accuracy and generalization capability of the model. And repeatedly adjusting model parameters and weights according to the verification result, optimizing the model performance, and ensuring that the model can accurately generate a client image.

Further, the adjusted portrait generation model, the supplementary feature data and the previously extracted key feature data are fused to form a complete input data set. And predicting the integrated data set by using the adjusted image generation model to generate the target customer image. The target customer representation should contain the latest features, behavior patterns, and demand features of the customer.

Optionally, the accuracy of the target client image is verified by means of client manager feedback, client behavior observation and the like, so that the latest requirements and behaviors of the client can be accurately reflected. The generated target customer portraits are applied to the scenes of customer demand prediction, accurate marketing, personalized recommendation and the like, so that the accuracy and the efficiency of business decision are improved.

Customer interaction data is acquired and integrated in real time, and customer portraits are dynamically updated, so that timeliness and accuracy of the portraits are ensured. And combining the potential demand data and the interaction data of the clients, improving the accuracy of demand prediction and providing a solid foundation for accurate marketing and personalized recommendation. By continuously adjusting and optimizing parameters and weights of the portrait generation model, the prediction performance of the model is improved, and the generated customer portrait is ensured to be more comprehensive and accurate. The generated target client image can accurately reflect the latest demands and behaviors of clients, support business decisions such as accurate marketing, client service and the like, and promote client satisfaction and marketing success rate.

Based on the above second embodiment, a third embodiment of the product recommendation method of the present application is presented. Referring to fig. 4, fig. 4 is a schematic flow chart of a third embodiment of the product recommendation method according to the present application.

In the present embodiment, step S41 includes:

S411: based on a preset feature recognition algorithm, performing feature recognition on the initial customer portrait to obtain behavior features, relationship features and preference features;

Specifically, initial customer representation data including basic information of the customer, historical behavior data, interactive records, and the like is prepared. Feature recognition algorithms are selected or developed, which may be rule-based systems or machine learning models. Customer behavior data such as transaction frequency, purchase period, browse records, etc. are extracted. Behavior characteristics of the client, such as consumption habits, liveness and the like, are identified based on the algorithm. Relationship data of the customer with other entities, such as customer interactions with customer managers, customer relationship networks with other customers, etc., are extracted. Relationship characteristics of the customer are identified, such as contact frequency with the customer manager, interaction relationship with other enterprises, and so forth. Customer preference data such as product selection, feedback records, etc. are extracted. Customer preference characteristics are identified, such as preferences for certain types of products, types of products commonly purchased, and the like.

S412: predicting the client behavior and the client demand according to the behavior characteristics, the relation characteristics and the preference characteristics to obtain behavior prediction data and demand prediction data;

Taking the identified behavior characteristics as input, constructing a behavior prediction model, wherein the model can adopt time sequence analysis, regression analysis, machine learning algorithm and the like. And predicting the future behavior of the client by using a behavior prediction model to obtain behavior prediction data such as the future purchase times, the future transaction amount, the future access frequency and the like. And taking the preference characteristics and the relation characteristics obtained by recognition as input, constructing a demand prediction model, wherein the model can adopt a classification algorithm, association rule mining, collaborative filtering and the like. The demand prediction model is used to predict the demand of the customer to obtain demand prediction data, such as products of possible interest, services that may be required, potential purchase intent, and the like.

S413: and carrying out demand probability evaluation on the behavior prediction data and the demand prediction data, and taking the behavior prediction data or the demand prediction data corresponding to a probability evaluation result exceeding a preset demand probability threshold value as the potential demand data of the client.

Specifically, the behavior prediction data and the demand prediction data are taken as inputs, a demand probability evaluation model is constructed, and the model can adopt a probability theory method, a Bayesian network, a machine learning classification model and the like. And carrying out probability evaluation on each piece of prediction data, and calculating the occurrence probability value of each piece of prediction data. A demand probability threshold, such as 0.7 or 0.8, is set according to the service demand and is used as a standard for judging the demand of the clients. And screening out the behavior prediction data and the demand prediction data corresponding to the probability evaluation result exceeding the preset demand probability threshold value, and taking the behavior prediction data and the demand prediction data as potential demand data of clients. And integrating the screened data to form a data set of potential demands of the clients, and recording the types of products or services which the clients may be interested in.

It should be noted that, the feature recognition algorithm is an algorithm for recognizing and extracting specific features in data, and may be a rule-based system or a machine learning model. The behavior prediction data is a prediction result of the future behavior of the client through a prediction model, such as the number of future purchases, the amount of future transactions, and the like. Demand forecast data is the result of a forecast of the future demands of a customer, such as products of possible interest, services that may be needed, etc., by means of a forecast model. The demand probability evaluation refers to probability evaluation of the predicted data and calculation of the occurrence probability value thereof. The demand probability threshold is a standard probability value for judging the demand of the customer.

And through the feature recognition and prediction model, the behavior and the demand of the client are accurately predicted, and the accuracy of demand prediction is improved. Through the identification and analysis of the behavior characteristics, the relation characteristics and the preference characteristics, the client requirements are comprehensively known, and personalized recommendation and service are provided. Through the demand probability evaluation, high-probability potential demand data are screened out, the marketing strategy is optimized, and the marketing effect and the customer satisfaction are improved. Based on the real-time updating of the customer interaction data and the demand prediction data, the customer portrait is dynamically adjusted, and timeliness and accuracy of the portrait are ensured.

Referring to fig. 5, fig. 5 is a schematic view of a sub-process in a third embodiment of the product recommendation method according to the present application.

In this embodiment, step S5 includes:

s51: extracting key features of the target customer portrait to obtain key feature data;

S52: normalizing the key characteristic data and the preset product data to obtain normalized key characteristic data and normalized product data;

S53: matching the normalized key feature data with each product feature data in the normalized product data;

S54: and determining the target recommended product according to the product characteristic data corresponding to the matching degree higher than the preset matching degree threshold.

Specifically, target customer portrait data including basic information, behavior characteristics, relationship characteristics, preference characteristics, and the like of the customer is prepared. And determining key characteristics which have significant influence on product recommendation, such as enterprise financial report information, supply chain information, digitization degree and the like. Selected key features are extracted from the target customer representation to form a key feature dataset.

Further, banking product data is prepared, including information on product names, categories, functions, prices, applicable customer groups, and the like. And carrying out normalization processing on the key characteristic data and the product data, and ensuring that the data are convenient to compare on a unified scale. And selecting a proper matching algorithm, such as cosine similarity, euclidean distance, pearson correlation coefficient and the like, according to the data characteristics. And carrying out similarity calculation on the normalized key feature data and the normalized product data to obtain the matching degree between each piece of customer feature data and each piece of product feature data. A matching degree threshold value, such as 0.7 or 0.8, is set as a recommendation standard according to service requirements. And screening out the product characteristic data with the matching degree higher than a preset threshold value, and determining the product characteristic data as a target recommended product. And generating a recommended product list according to the screening result, wherein the recommended product list comprises all products meeting the matching degree requirement.

By extracting the key feature data, the high matching of the recommended products and the requirements of clients is ensured, and the accuracy of recommendation is improved. The normalization processing ensures the comparability of the customer characteristic data and the product data, and reduces the influence caused by the data isomerism. By adopting the similarity matching algorithm, the efficiency and accuracy of product recommendation are improved, and the reliability of a recommendation result is ensured. Accurate product recommendation enhances the personalized experience of the clients, improves the satisfaction degree and the loyalty degree of the clients, and improves the marketing effect.

Based on the above-described second embodiment, in the present embodiment, after step S5, further includes:

S5a: acquiring purchase detail information, and judging whether the purchase detail information comprises the target recommended product or not;

specifically, purchasing detail information of the customer including the name of the product purchased, the time of purchase, the amount of purchase, and the like is collected from a transaction system of a bank, a CRM system, and the like. And the collected purchasing detail information is arranged and stored, so that the integrity and accuracy of the data are ensured. Comparing the purchase detail information with the target recommended product, and judging whether the customer purchases the recommended product. Recording the comparison result, and marking which recommended products are purchased by clients and which are not purchased.

S5b: if not, acquiring customer feedback information, and determining a purchase decision influencing factor according to the customer feedback information;

Specifically, customer feedback information is collected through a variety of channels, such as customer questionnaires, telephone return visits, online evaluations, and the like. And sorting and classifying the collected customer feedback information, and recording the opinion and suggestion of the customer on the recommended product. And analyzing the feedback information of the clients to identify the reasons that the clients do not purchase the recommended products, such as excessive price, non-conforming to the requirements of the products and unequal recommending time. Based on the feedback analysis, the primary factors that influence the customer's purchasing decision are determined.

S5c: and generating a product recommendation optimization strategy based on the purchase decision influencing factors and the key feature data.

Specifically, the purchase decision influencing factors are integrated with key feature data of the clients to form a comprehensive data set. And analyzing the relation between the key characteristic data and the purchase decision influencing factors, and identifying key variables influencing the recommendation effect. Based on the comprehensive data set, parameters and weights of the product recommendation model are adjusted, and the recommendation strategy is optimized. And adjusting recommendation rules and logic, such as adjusting the price range, recommendation time, product characteristics and the like of the recommended products according to the identified key variables and influencing factors. And applying the optimized recommendation strategy to a product recommendation system to generate a new recommendation scheme. And verifying the effect of the optimization strategy through customer feedback and purchasing detail information, and continuously adjusting and perfecting the recommendation strategy.

The purchasing detail information is detailed data for recording purchasing behavior of the customer, and includes the name, time, amount and the like of the purchased product. Customer feedback information is opinion and advice of the customer on products and services, and can be collected by means of questionnaires, telephone return visits, online evaluation and the like. Purchase decision influencing factors refer to the main reasons that influence the customer's purchase decision, such as price, product characteristics, recommendation opportunities, etc. The product recommendation optimization strategy is a strategy for adjusting a recommendation model and rules and optimizing a recommendation effect based on an analysis result.

By analyzing the reasons of not purchasing the recommended products, the recommendation strategy is optimized in a targeted manner, and the matching degree and the customer acceptance degree of the recommended products are improved. And collecting and analyzing the customer feedback, and timely adjusting the recommendation strategy to provide a recommendation scheme which meets the customer requirements better, so that the customer satisfaction is improved. And by combining purchase detail information and customer feedback, the recommendation model and rules are continuously optimized, so that the recommendation system can adapt to changes of market and customer requirements in real time, and the recommendation effect is improved. Through a data analysis and feedback mechanism, product recommendation optimization is driven based on real data, scientificity and effectiveness of a recommendation strategy are enhanced, and accuracy of business decision is improved.

For an exemplary purpose of understanding the technical concept or principle of the product recommendation method according to the above embodiment, please refer to fig. 6, and fig. 6 is a system architecture diagram of the product recommendation method according to an embodiment of the present application.

As shown in fig. 6, in this embodiment, the system architecture is divided into two parts, namely a client and a server, and requests are distributed through nginnx, so that the load is balanced to a plurality of application servers. The server comprises a plurality of modules including a customer information management service, a customer portrait feature service and a marketing service proposal service. The client terminal includes a client's computer and mobile device, accessing the system via the internet. And carrying out request distribution through Nginx, and balancing the load to an application server. The CRM service gateway is responsible for receiving and forwarding client requests, performing authority control and request routing. The customer information management service provides customer information query and analysis APIs. The customer representation feature service provides feature analysis and customer representation generation. The marketing service proposal service intelligently recommends products and generates service proposal according to customer portraits and demands. The application server cluster is used for bearing various application services of the system. The MySQL database is used for storing client information, image data and marketing schemes and providing data backup. The RabbitMQ message queue is used for processing high concurrency requests, and system stability is guaranteed. Redis cache is used for caching hot spot data, and database pressure is relieved. The log system records the operation log of the system, and is convenient for monitoring and analysis. ClickHouse and Gauss databases are used for high performance data analysis, storing customer portraits and behavioral data. The file server is used for uploading and downloading files, storing service schemes and related documents. And the modules are subjected to data synchronization through service call, and the MySQL database periodically backs up data. And (3) cleaning and normalizing the data from different sources, and integrating the data into a characteristic information database. And generating a corresponding product recommendation scheme according to the customer portrait, and storing and transmitting the product recommendation scheme through a file server. The system architecture comprises load balancing, authority control, a high-performance database, a message queue and a caching mechanism, and provides an efficient and stable client information management and product recommendation platform. Through cleaning and integrating the feature data, the customer portraits are accurately generated, the product is intelligently recommended based on the portraits, the product recommendation strategy is optimized, and the customer satisfaction degree and the product recommendation effect are improved.

Further, the call service refers to a mechanism for exchanging data and calling functions between the modules of the system. In the system architecture, the call and data interaction between the modules is realized through the service gateway and the micro-service architecture. Each micro service is registered and discovered through the service gateway, so that the accessibility and dynamic calling of the service are ensured. Rights control is performed through the service gateway to ensure that only authorized requests can access a particular service. The service gateway and the Nginx jointly realize load balancing, and ensure that the requests are uniformly distributed to each application server. The retry mechanism refers to that if a service fails temporarily during a system call, the retry mechanism can automatically reinitiate a request to attempt to call the service again. The retry mechanism may improve the fault tolerance of the system and reduce request failures due to temporary failures. By means of a retry mechanism, it can be ensured that the request is successful during short network fluctuations or service unavailability. The fusing mechanism is used to protect the system from a large number of failed requests. When a certain service continuously fails to exceed a preset threshold, the system temporarily stops calling the service, and system resources are prevented from being exhausted. And a large number of failed requests are prevented from consuming system resources, and normal operation of other services is ensured. Through the fusing mechanism, after service recovery is detected, the call to the service can be recovered step by step, and the system is ensured to be recovered to be normal quickly.

Optionally, referring to fig. 7, fig. 7 is a diagram illustrating data processing according to an embodiment of the product recommendation method of the present application.

As shown in fig. 7, after collecting the multi-channel data, the customer and product data required in each scene are sorted (i.e., the marketing-related system is sorted). Data cleaning refers to processing missing, duplicate, outliers and normalizing. The unified data format (definition data template) refers to an attribute template defining three types of data of label, relation and unstructured. Data association integration refers to unifying customer ID and product ID, integrating fragmented data into a formatting table of customer granularity and product granularity. It should be noted that the data comes from different systems and scenarios, including internal and external data sources. The quality and consistency of the data are ensured through multiple times of processing, and a reliable data base is provided for subsequent customer portrait generation. And through data cleaning and normalization processing, various forms of data are processed uniformly, so that the subsequent analysis and application are facilitated.

An embodiment of the present application further provides a product recommendation device, referring to fig. 8, fig. 8 is a schematic block diagram of the product recommendation device according to the embodiment of the present application, where the product recommendation device includes:

the classification module 801 is configured to classify preset product data and customer data to obtain multiple types of data;

A normalization module 802, configured to perform normalization processing on the multiple types of data to obtain feature standard data;

an initial portrait generation module 803, configured to generate an initial customer portrait based on the feature standard data and preset historical feature data;

The portrait update module 804 is configured to obtain customer interaction data, and update the initial customer portrait based on the customer interaction data and customer potential demand data to obtain a target customer portrait;

And the product recommending module 805 is configured to determine a target recommended product according to the target customer portrait.

The product recommendation device provided by the embodiment of the application can solve the technical problem of how to improve the accuracy of product recommendation and the product recommendation effect by adopting the product recommendation method in the embodiment. Compared with the prior art, the product recommending device provided by the embodiment of the application has the same beneficial effects as the product recommending method provided by the embodiment, and other technical features in the product recommending device are the same as the features disclosed by the method of the embodiment, and are not described in detail herein.

The present application provides a product recommendation device, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the product recommendation method of the above embodiment.

Referring now to FIG. 9, a schematic diagram of a product recommendation device suitable for use in implementing embodiments of the present application is shown. The product recommendation device in the embodiment of the present application may include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal DIGITAL ASSISTANT: personal digital assistants), PADs (Portable Application Description: tablet computers), PMPs (Portable MEDIA PLAYER: portable multimedia players), vehicle-mounted terminals (e.g., vehicle-mounted navigation terminals), and the like, and fixed terminals such as digital TVs, desktop computers, and the like. The product recommendation device shown in fig. 9 is only an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present application.

As shown in fig. 9, the product recommendation apparatus may include a processing device 1001 (e.g., a central processing unit, a graphic processor, etc.) that may perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access Memory (RAM: random Access Memory) 1004. In the RAM1004, various programs and data required for the operation of the product recommendation device are also stored. The processing device 1001, the ROM1002, and the RAM1004 are connected to each other by a bus 1005. An input/output (I/O) interface 1006 is also connected to the bus. In general, the following systems may be connected to the I/O interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, and the like; an output device 1008 including, for example, a Liquid crystal display (LCD: liquid CRYSTAL DISPLAY), a speaker, a vibrator, and the like; storage device 1003 including, for example, a magnetic tape, a hard disk, and the like; and communication means 1009. The communication means 1009 may allow the product recommendation device to communicate wirelessly or by wire with other devices to exchange data. While a product recommendation device having various systems is shown in the figures, it should be understood that not all of the illustrated systems are required to be implemented or provided. More or fewer systems may alternatively be implemented or provided.

In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. The above-described functions defined in the method of the disclosed embodiment of the application are performed when the computer program is executed by the processing device 1001.

The product recommending equipment provided by the application can solve the technical problem of how to improve the accuracy of product recommendation and the effect of product recommendation by adopting the product recommending method in the embodiment. Compared with the prior art, the product recommending device provided by the application has the same beneficial effects as the product recommending method provided by the embodiment, and other technical features in the product recommending device are the same as the features disclosed by the method of the previous embodiment, and are not repeated herein.

It is to be understood that portions of the present disclosure may be implemented in hardware, software, firmware, or a combination thereof. In the description of the above embodiments, particular features, structures, materials, or characteristics may be combined in any suitable manner in any one or more embodiments or examples.

The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

The present application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon for performing the product recommendation method in the above-described embodiments.

The computer readable storage medium provided by the present application may be, for example, a U disk, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access Memory (RAM: random Access Memory), a Read-Only Memory (ROM: read Only Memory), an erasable programmable Read-Only Memory (EPROM: erasable Programmable Read Only Memory or flash Memory), an optical fiber, a portable compact disc Read-Only Memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In this embodiment, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, or device. Program code embodied on a computer readable storage medium may be transmitted using any appropriate medium, including but not limited to: wire, fiber optic cable, RF (Radio Frequency), and the like, or any suitable combination of the foregoing.

The above-mentioned computer-readable storage medium may be contained in a product recommendation device; or may exist alone without being assembled into the product recommendation device.

The computer-readable storage medium carries one or more programs that, when executed by the product recommendation device, cause the product recommendation device to: classifying preset product data and customer data to obtain multi-type data; carrying out standardization processing on the multi-type data to obtain characteristic standard data; generating an initial customer portrait based on the feature standard data and preset historical feature data; acquiring customer interaction data, and updating the initial customer portrait based on the customer interaction data and the customer potential demand data to obtain a target customer portrait; and determining a target recommended product according to the target customer portrait. Computer program code for carrying out operations of the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of remote computers, the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN: local Area Network) or a wide area network (WAN: wide Area Network), or may be connected to an external computer (for example, through the Internet using an Internet service provider).

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The modules involved in the embodiments of the present application may be implemented in software or in hardware. Wherein the name of the module does not constitute a limitation of the unit itself in some cases.

The readable storage medium provided by the application is a computer readable storage medium, and the computer readable storage medium stores computer readable program instructions (namely computer programs) for executing the product recommendation method, so that the technical problem of how to improve the accuracy of product recommendation and the product recommendation effect can be solved. Compared with the prior art, the beneficial effects of the computer readable storage medium provided by the application are the same as those of the product recommendation method provided by the above embodiment, and are not described herein.

An embodiment of the application provides a computer program product comprising a computer program which, when executed by a processor, implements the steps of a product recommendation method as described above.

The computer program product provided by the application can solve the technical problem of how to improve the accuracy of product recommendation and the product recommendation effect. Compared with the prior art, the beneficial effects of the computer program product provided by the embodiment of the application are the same as those of the product recommending method provided by the embodiment, and are not described in detail herein.

The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the application, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein, or any application, directly or indirectly, within the scope of the application.

Claims

1. A method of product recommendation, comprising:

classifying preset product data and customer data to obtain multi-type data;

2. The method of claim 1, wherein the step of normalizing the multi-type data to obtain feature standard data comprises:

Normalizing the cleaning data to obtain normalized data;

3. The method of claim 1, wherein the step of generating an initial customer representation based on the feature criteria data and preset historical feature data comprises:

4. The method of claim 3, wherein the step of obtaining customer interaction data and updating the initial customer representation based on the customer interaction data and customer potential demand data to obtain a target customer representation comprises:

identifying the initial customer representation and determining the customer potential demand data;

5. The method of claim 4, wherein the step of identifying the initial customer representation and determining the customer potential demand data comprises:

6. The method of any one of claims 1 to 5, wherein the step of determining a target recommended product from the target customer representation comprises:

7. The method of claim 6, further comprising, after the step of determining a target recommended product from the target customer representation:

8. A product recommendation device, comprising:

the portrait updating module is used for acquiring customer interaction data, updating the initial customer portrait based on the customer interaction data and the customer potential demand data, and obtaining a target customer portrait;

9. A computer device, the device comprising: a memory, a processor and a product recommendation program stored on the memory and executable on the processor, the product recommendation program being configured to implement the steps of the product recommendation method of any of claims 1 to 7.

10. A storage medium having stored thereon a product recommendation program which when executed by a processor implements the steps of the product recommendation method according to any one of claims 1 to 7.