A personalized coupon recommendation method and system based on user characteristics

Abstract

The present invention discloses a personalized coupon push method and system based on user characteristics, including: obtaining user historical behavior information, performing user preference analysis based on the user historical behavior information, and obtaining user preference analysis information; obtaining user real-time behavior information, performing user interest prediction based on the user real-time behavior information, and obtaining user interest prediction information; performing attribute analysis on target users in combination with user preference analysis information and user interest prediction information, and obtaining attribute analysis result information; obtaining coupon information, performing attribute analysis and category classification on various coupons, and performing personalized coupon push for target users, and obtaining candidate coupon push information; performing store analysis and evaluation based on the candidate coupon push information, and selecting recommended coupons using the evaluation results as weights, and obtaining final coupon push information. The quality of coupons is guaranteed, and the preferences and usage habits of users are met, user satisfaction is improved, and the use, receipt and purchase of coupons are promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of coupon recommendation, and in particular to a personalized coupon recommendation method and system based on user characteristics. Background Art

[0002] In the modern business environment, with the booming development of e-commerce and mobile applications, personalized recommendation systems have gradually become one of the important means to improve user satisfaction and promote sales. The traditional way of issuing coupons is usually to push the same discount information to a large number of users in batches, lacking a deep understanding of individual differences. At the same time, traditional coupon push is usually based on global rules or mass trends, which cannot well meet the unique needs of different individual users. In addition, the traditional coupon push system mainly relies on the user's purchase history, but lacks a comprehensive analysis of user characteristics, such as user interests, preferences, shopping time preferences, etc.

[0003] With the continuous development of big data analysis and artificial intelligence, the large amount of data generated by users on the Internet platform, such as search history, shopping behavior, social interaction, etc., enables a more comprehensive and accurate understanding of the individual characteristics of users. By mining user behavior data, we can gain an in-depth understanding of user interests, shopping preferences, usage habits and other aspects, providing a rich information basis for personalized coupons, and more accurately understanding the individual differences of users, so as to push coupons that meet the user's personalization. Improve user satisfaction, enhance brand loyalty, and drive sales growth for merchants. Summary of the invention

[0004] The present invention overcomes the defects of the prior art and provides a personalized coupon promotion method and system based on user characteristics, the important purpose of which is to improve user satisfaction and promote merchant sales growth.

[0005] To achieve the above-mentioned purpose, the present invention provides a personalized coupon recommendation method based on user characteristics in a first aspect, comprising:

[0006] Obtaining user historical behavior information, performing user preference analysis based on the user historical behavior information, and obtaining user preference analysis information;

[0007] Acquire user real-time behavior information, and predict user interests based on the user real-time behavior information to obtain user interest prediction information;

[0008] Combining the user preference analysis information and the user interest prediction information, the attribute analysis is performed on the target user to obtain the attribute analysis result information;

[0009] Obtain coupon information, perform attribute analysis and category classification on various coupons, and recommend personalized coupons to target users to obtain candidate coupon information;

[0010] Store analysis and evaluation are performed based on the candidate coupon information, and the evaluation results are used as weights to select recommended coupons to obtain final coupon information.

[0011] In this solution, the user historical behavior information is obtained, and the user preference analysis is performed based on the user historical behavior information, specifically:

[0012] Acquire user historical behavior information, wherein the user historical behavior information includes: user historical purchase behavior, user historical browsing behavior and user historical interaction behavior;

[0013] Extracting features from the user's historical behavior information to extract the user's historical purchase type, frequency and price features, historical browsing behavior features and historical interaction behavior features to obtain the user's historical behavior feature information;

[0014] Based on the clustering algorithm and combined with the user's historical behavior characteristics information, the user's purchase preference analysis is carried out, and the user's purchase behavior is divided into two categories: purchase type and purchase price, and the purchase frequency is combined into the two categories to obtain the user's historical purchase analysis information;

[0015] Taking the purchase frequency as a preference judgment indicator, the user preference analysis is performed in combination with the user's historical purchase analysis information to obtain the first preference analysis information;

[0016] Analyze the user's browsing preferences through the user's historical behavior feature information, preset browsing categories, calculate the Manhattan distance between the user's historical browsing behavior and the browsing category, and judge it with the preset threshold to obtain the user's browsing category information;

[0017] Extracting the browsing time and frequency of the user through the historical behavior feature information of the user, setting the corresponding weights, performing weighted calculation on the browsing category information of the user, analyzing the browsing preference of the user through the weighted calculation result, and obtaining the second preference analysis information;

[0018] Constructing a sentiment analysis model, inputting the user's historical behavior feature information into the sentiment analysis model to analyze the sentiment of the user's historical interactive behavior, and obtaining historical interactive behavior sentiment analysis information;

[0019] The user's historical interaction objects and interaction frequency are extracted through the historical interaction behavior sentiment analysis information, and the interaction frequency is used as a preference judgment indicator to obtain the third preference analysis information;

[0020] The first preference analysis information, the second preference analysis information and the third preference analysis information are integrated to obtain the user preference analysis information.

[0021] In this solution, the real-time user behavior information is obtained, and user interest prediction is performed based on the real-time user behavior information to obtain user interest prediction information, specifically:

[0022] Acquire user real-time behavior information, wherein the user real-time behavior information includes: user real-time browsing behavior information and user real-time interactive behavior information;

[0023] Perform feature extraction on the user's real-time behavior information to extract the user's real-time browsing behavior object information, dwell time and click frequency, and user's real-time interactive behavior features to obtain user's real-time behavior feature information;

[0024] Preset interest tags, calculate the Mahalanobis distance between the user's real-time browsing object and the interest tag, use it as a correlation judgment indicator and a preset threshold for judgment, and merge the corresponding stay time and click frequency into the corresponding interest tag to obtain preliminary analysis information;

[0025] Using the dwell time and click frequency as interest judgment indicators, setting judgment thresholds corresponding to different interest scores respectively, judging the preliminary analysis information and the judgment thresholds, calculating the interest scores of different interest tags, judging the user's interest preference by the interest scores, and obtaining the first interest analysis information;

[0026] Inputting the real-time behavior feature information into the emotion sub-model to analyze the user's real-time interactive behavior emotion to obtain real-time interactive behavior emotion analysis information;

[0027] The real-time interactive behavior sentiment analysis information and the first interest analysis information are used to predict and analyze the real-time interest preference of the target user to obtain user interest prediction information.

[0028] In this solution, the attribute analysis of the target user is performed in combination with the user preference analysis information and the user interest prediction information to obtain the attribute analysis result information, specifically:

[0029] Obtain user preference analysis information and user interest prediction information, and build a user attribute analysis model based on the random forest algorithm;

[0030] Preset user attribute labels, obtain feature information of various user attribute labels based on big data retrieval, build a training data set, and perform deep learning and training on the user attribute analysis model;

[0031] Setting different analysis weights for user preferences and user interests, and inputting the user preference analysis information and user interest prediction information into the user attribute analysis model for analysis;

[0032] The user preference analysis information and user interest prediction information are used as attribute features, and multiple decision trees are constructed for attribute analysis. Several features are randomly selected and node splitting is performed. The cosine similarity between each node and various attribute labels is calculated, and the cosine similarity is compared with the preset threshold to select the available nodes.

[0033] Preset the selection threshold, extract the attribute labels of the available nodes and count the occurrence frequency, judge the occurrence frequency of each attribute label with the selection threshold, and obtain the attribute analysis result information;

[0034] The attribute analysis result information includes: user long-term preference attributes and user short-term preference attributes.

[0035] In this solution, the coupon information is obtained, the attributes of various coupons are analyzed and classified, and personalized coupons are recommended to target users to obtain candidate coupon information, specifically:

[0036] Obtain the user's historical coupon information, perform feature extraction on the user's historical coupon information, extract coupon type features, coupon validity features, discount features, and starting amount features, and obtain historical coupon feature information;

[0037] Performing user usage incentive analysis based on the historical coupon usage feature information, and classifying the coupon usage categories based on the coupon type features using a clustering algorithm to obtain usage category classification information;

[0038] According to the coupon validity characteristics, discount characteristics and starting amount characteristics, corresponding distribution graphs are constructed and divided into intervals. The number of coupons used is used as a usage tendency analysis indicator. The user's usage tendency is analyzed in combination with the constructed distribution graph to obtain usage tendency analysis information.

[0039] Combine the usage category classification information and the usage tendency analysis information to perform usage inducement analysis to obtain usage inducement analysis information;

[0040] Obtaining coupon information, performing attribute analysis on the coupon information, analyzing type attributes, time validity attributes, discount attributes, and starting amount attributes of various coupons, and obtaining coupon attribute analysis information;

[0041] According to the usage incentive analysis information, four matching weights of type, time limit, discount and starting amount are set, and weighted calculation is performed on the coupon attribute analysis information to calculate the matching weights of the attributes of various coupons and the usage incentives to obtain matching degree information;

[0042] The matching degree information is judged against a preset threshold, and coupons are recommended based on the judgment result to obtain candidate coupon recommendation information.

[0043] In this solution, store evaluation is performed based on the candidate coupon information, and coupon screening and selection is performed in combination with the attribute analysis result information to obtain the final coupon information, specifically:

[0044] Acquire candidate coupon information, extract store information corresponding to the coupon according to the candidate coupon information, and obtain store information;

[0045] Based on big data retrieval and combined with store information, data collection is performed to collect the target store's rating information, evaluation information, and store attribute information to obtain store feature information;

[0046] Constructing a store evaluation model, inputting the store feature information into the store evaluation model for evaluation, and obtaining store evaluation result information;

[0047] Acquire attribute analysis result information, pre-screen the candidate coupon information according to the store evaluation result information, and obtain pre-screening result information;

[0048] The attribute analysis result information is used as a selection index and a corresponding weight is set. The pre-screening result information is weighted and sorted. Coupon selection is performed based on the sorting result to obtain final coupon recommendation information.

[0049] A second aspect of the present invention provides a personalized coupon push system based on user characteristics, the system comprising: a memory, a processor, the memory containing a personalized coupon push method program based on user characteristics, the personalized coupon push method program based on user characteristics when executed by the processor implements the following steps:

[0050] Obtaining user historical behavior information, performing user preference analysis based on the user historical behavior information, and obtaining user preference analysis information;

[0051] Acquire user real-time behavior information, and predict user interests based on the user real-time behavior information to obtain user interest prediction information;

[0052] Combining the user preference analysis information and the user interest prediction information, the attribute analysis is performed on the target user to obtain the attribute analysis result information;

[0053] Obtain coupon information, perform attribute analysis and category classification on various coupons, and recommend personalized coupons to target users to obtain candidate coupon information;

[0054] Store analysis and evaluation are performed based on the candidate coupon information, and the evaluation results are used as weights to select recommended coupons to obtain final coupon information.

[0055] The present invention discloses a personalized coupon push method and system based on user characteristics, including: obtaining user historical behavior information, performing user preference analysis based on the user historical behavior information, and obtaining user preference analysis information; obtaining user real-time behavior information, performing user interest prediction based on the user real-time behavior information, and obtaining user interest prediction information; performing attribute analysis on target users in combination with user preference analysis information and user interest prediction information, and obtaining attribute analysis result information; obtaining coupon information, performing attribute analysis and category classification on various coupons, and performing personalized coupon push for target users, and obtaining candidate coupon push information; performing store analysis and evaluation based on the candidate coupon push information, and selecting recommended coupons using the evaluation results as weights, and obtaining final coupon push information. The quality of coupons is guaranteed, and the preferences and usage habits of users are met, user satisfaction is improved, and the use, receipt and purchase of coupons are promoted. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments or exemplary embodiments of the present invention, the drawings required for use in the embodiments or exemplary descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained according to the drawings without paying creative work.

[0057] Figure 1 A flowchart of a personalized coupon recommendation method based on user characteristics provided by an embodiment of the present invention;

[0058] Figure 2 A coupon recommendation flow chart provided by an embodiment of the present invention;

[0059] Figure 3 A block diagram of a personalized coupon recommendation system based on user characteristics provided by an embodiment of the present invention;

[0060] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0061] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0062] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0063] Figure 1 A flowchart of a personalized coupon recommendation method based on user characteristics provided by an embodiment of the present invention;

[0064] like Figure 1 As shown, the present invention provides a flow chart of a personalized coupon recommendation method based on user characteristics, including:

[0065] S102, obtaining user historical behavior information, performing user preference analysis based on the user historical behavior information, and obtaining user preference analysis information;

[0066] Acquire user historical behavior information, wherein the user historical behavior information includes: user historical purchase behavior, user historical browsing behavior and user historical interaction behavior;

[0067] Extracting features from the user's historical behavior information to extract the user's historical purchase type, frequency and price features, historical browsing behavior features and historical interaction behavior features to obtain the user's historical behavior feature information;

[0068] Based on the clustering algorithm and combined with the user's historical behavior characteristics information, the user's purchase preference analysis is carried out, and the user's purchase behavior is divided into two categories: purchase type and purchase price, and the purchase frequency is combined into the two categories to obtain the user's historical purchase analysis information;

[0069] Taking the purchase frequency as a preference judgment indicator, the user preference analysis is performed in combination with the user's historical purchase analysis information to obtain the first preference analysis information;

[0070] Analyze the user's browsing preferences through the user's historical behavior feature information, preset browsing categories, calculate the Manhattan distance between the user's historical browsing behavior and the browsing category, and judge it with the preset threshold to obtain the user's browsing category information;

[0071] Extracting the browsing time and frequency of the user through the historical behavior feature information of the user, setting the corresponding weights, performing weighted calculation on the browsing category information of the user, analyzing the browsing preference of the user through the weighted calculation result, and obtaining the second preference analysis information;

[0072] Constructing a sentiment analysis model, inputting the user's historical behavior feature information into the sentiment analysis model to analyze the sentiment of the user's historical interactive behavior, and obtaining historical interactive behavior sentiment analysis information;

[0073] The user's historical interaction objects and interaction frequency are extracted through the historical interaction behavior sentiment analysis information, and the interaction frequency is used as a preference judgment indicator to obtain the third preference analysis information;

[0074] The first preference analysis information, the second preference analysis information and the third preference analysis information are integrated to obtain the user preference analysis information.

[0075] It should be noted that the target user's preference analysis is performed through the user's historical behavior information. First, based on the clustering algorithm, combined with the user's historical behavior feature information, the user's purchase preference analysis is performed. The user's purchase behavior is divided into two main categories, namely purchase type and purchase price. The user's purchase type and purchase price are analyzed, and the purchase frequency is used as an analysis indicator to understand the user's purchase item preference and price preference, and understand the user's purchasing habits. Next, the user's browsing preference is analyzed through the user's historical browsing behavior, and the user's attention to the browsing content is judged according to the user's browsing time and frequency, and the user's preference is further analyzed from the browsing content. Then, the user's historical interaction behavior is analyzed, and the target user is subjected to sentiment analysis. The target user's emotional tendency towards the interaction object is judged from the target user's interaction behavior, thereby judging the user's preference degree. Finally, the user's preferences are analyzed from the three perspectives of purchasing behavior, browsing behavior, and interaction behavior, so as to map the user's long-term preference content and help to understand the target user more deeply.

[0076] S104, obtaining user real-time behavior information, and predicting user interests based on the user real-time behavior information to obtain user interest prediction information;

[0077] Acquire user real-time behavior information, wherein the user real-time behavior information includes: user real-time browsing behavior information and user real-time interactive behavior information;

[0078] Perform feature extraction on the user's real-time behavior information to extract the user's real-time browsing behavior object information, dwell time and click frequency, and user's real-time interactive behavior features to obtain user's real-time behavior feature information;

[0079] Preset interest tags, calculate the Mahalanobis distance between the user's real-time browsing object and the interest tag, use it as a correlation judgment indicator and a preset threshold for judgment, and merge the corresponding stay time and click frequency into the corresponding interest tag to obtain preliminary analysis information;

[0080] Using the dwell time and click frequency as interest judgment indicators, setting judgment thresholds corresponding to different interest scores respectively, judging the preliminary analysis information and the judgment thresholds, calculating the interest scores of different interest tags, judging the user's interest preference by the interest scores, and obtaining the first interest analysis information;

[0081] Inputting the real-time behavior feature information into the emotion sub-model to analyze the user's real-time interactive behavior emotion to obtain real-time interactive behavior emotion analysis information;

[0082] The real-time interactive behavior sentiment analysis information and the first interest analysis information are used to predict and analyze the real-time interest preference of the target user to obtain user interest prediction information.

[0083] It should be noted that, first of all, the user's real-time behavior information is concerned, including real-time browsing behavior information and user real-time interactive behavior information. The user's real-time behavior information reflects the user's activities on the platform, thereby providing an analysis basis for analyzing the user's short-term preferences. Subsequently, the user's real-time behavior information is feature extracted. By extracting the object information, dwell time and click frequency of the user's real-time browsing behavior, as well as the characteristics of the user's real-time interactive behavior, the user's real-time behavior feature information is obtained. Further, the interest tag is preset, and the Mahalanobis distance between the user's real-time browsing object and the preset interest tag is calculated as the correlation judgment index to judge the association between the user's real-time browsing object and the interest tag, and analyze the user's interest from the user's browsing behavior. Next, the dwell time and click frequency are used as interest judgment indicators, and the judgment threshold of different interest scores is set. The preliminary analysis information is compared with the judgment threshold to calculate the interest scores of different interest tags. Through the judgment of the interest score, the user's first interest analysis information is obtained, which intuitively reflects the user's real-time interest. Then, the real-time behavior feature information is input into the sentiment analysis model, the sentiment of the user's real-time interactive behavior is analyzed, and the real-time interactive behavior sentiment analysis information is obtained. The user's interest tendency is judged from the user's real-time interactive behavior, such as user likes, user comments and other interactive behaviors. Finally, the real-time interactive behavior sentiment analysis information and the first interest analysis information are combined to predict the real-time interests of the target users and judge the users’ short-term preferences. Through multi-level analysis, more accurate user information is provided for personalized recommendations.

[0084] S106, performing attribute analysis on the target user in combination with the user preference analysis information and the user interest prediction information to obtain attribute analysis result information;

[0085] Obtain user preference analysis information and user interest prediction information, and build a user attribute analysis model based on the random forest algorithm;

[0086] Preset user attribute labels, obtain feature information of various user attribute labels based on big data retrieval, build a training data set, and perform deep learning and training on the user attribute analysis model;

[0087] Setting different analysis weights for user preferences and user interests, and inputting the user preference analysis information and user interest prediction information into the user attribute analysis model for analysis;

[0088] The user preference analysis information and user interest prediction information are used as attribute features, and multiple decision trees are constructed for attribute analysis. Several features are randomly selected and node splitting is performed. The cosine similarity between each node and various attribute labels is calculated, and the cosine similarity is compared with the preset threshold to select the available nodes.

[0089] Preset the selection threshold, extract the attribute labels of the available nodes and count the occurrence frequency, judge the occurrence frequency of each attribute label with the selection threshold, and obtain the attribute analysis result information;

[0090] The attribute analysis result information includes: user long-term preference attributes and user short-term preference attributes.

[0091] It should be noted that, first, a user attribute analysis model is constructed, and the random forest algorithm is used for modeling. User attribute labels are preset, and feature information of various user attribute labels is obtained through big data retrieval. Subsequently, a training data set is constructed to perform deep learning and training on the user attribute analysis model. By introducing a large number of label features, the model has the ability to understand and predict the multi-dimensional attributes of users. In order to better understand the user, different analysis weights are set, and user preference analysis information and user interest prediction information are input into the user attribute analysis model for in-depth analysis. The user's attributes are better understood by weight setting, and the differences in user preferences and interests are fully considered. Next, the user preference analysis information and user interest prediction information are used as attribute features and analyzed in the user attribute analysis model. In each decision tree, several features are randomly selected for node splitting, and the cosine similarity of each node and various attribute labels is calculated. By judging with the preset threshold, available nodes are selected, and each node represents the prediction result for the user attribute; by presetting the selection threshold, available prediction results are selected, and the occurrence frequency is counted, and the selection threshold is judged to obtain the attribute analysis result for the target user. The attribute analysis result information includes user long-term preference attributes and user short-term preference attributes. By analyzing the user's attributes, we can define target users, such as food lovers, sports enthusiasts, etc. Based on the target user's attributes, we can better recommend coupons related to the user's attributes, in line with the user's preferences and interests, and encourage users to use coupons for consumption.

[0092] S108, obtaining coupon information, performing attribute analysis and category classification on various coupons, and performing personalized coupon promotion for target users to obtain candidate coupon promotion information;

[0093] Obtain the user's historical coupon information, perform feature extraction on the user's historical coupon information, extract coupon type features, coupon validity features, discount features, and starting amount features, and obtain historical coupon feature information;

[0094] Performing user usage incentive analysis based on the historical coupon usage feature information, and classifying the coupon usage categories based on the coupon type features using a clustering algorithm to obtain usage category classification information;

[0095] According to the coupon validity characteristics, discount characteristics and starting amount characteristics, corresponding distribution graphs are constructed and divided into intervals. The number of coupons used is used as a usage tendency analysis indicator. The user's usage tendency is analyzed in combination with the constructed distribution graph to obtain usage tendency analysis information.

[0096] Combine the usage category classification information and the usage tendency analysis information to perform usage inducement analysis to obtain usage inducement analysis information;

[0097] Obtaining coupon information, performing attribute analysis on the coupon information, analyzing type attributes, time validity attributes, discount attributes, and starting amount attributes of various coupons, and obtaining coupon attribute analysis information;

[0098] According to the usage incentive analysis information, four matching weights of type, time limit, discount and starting amount are set, and weighted calculation is performed on the coupon attribute analysis information to calculate the matching weights of the attributes of various coupons and the usage incentives to obtain matching degree information;

[0099] The matching degree information is judged against a preset threshold, and coupons are recommended based on the judgment result to obtain candidate coupon recommendation information.

[0100] It should be noted that the user's historical coupon information is first focused on, and the multi-dimensional feature information of the user's historical coupons is obtained by extracting the coupon type feature, coupon time feature, discount feature and starting amount feature. Then, the user's use incentive analysis is performed based on the historical coupon feature information. The clustering algorithm is used to classify the use of coupons based on the coupon type feature, and the use category classification information is obtained. At the same time, the corresponding distribution map is constructed according to the coupon time feature, discount feature and starting amount feature, and the interval division is performed. The number of coupons used is used as the use tendency analysis indicator, and the user's use tendency is analyzed in combination with the constructed distribution map to form the use tendency analysis information. Combined with the use category classification information and the use tendency analysis information, the use incentive analysis is performed to determine whether the reason for the user to use the coupon is the type of coupon, the validity period, the discount strength or the starting amount, so as to understand what kind of coupons can be used by the target user. Subsequently, the coupon information is obtained and the attribute analysis of this information is performed. The type attribute, time attribute, discount attribute and starting amount attribute of various coupons are analyzed. Four matching weights of type, time, discount and starting amount are set according to the use incentive analysis information. By performing weighted calculation on the coupon attribute analysis information, the matching weights of various coupon attributes and usage incentives are calculated to obtain the matching information. Finally, the matching information is judged with the preset threshold, and coupons are recommended based on the judgment results to obtain candidate coupon information. Based on the user's incentives to use coupons, the existing coupons that meet these incentives are judged and recommended, which is closer to the user's coupon usage preferences and promotes the use, collection and purchase of coupons.

[0101] S110, analyzing and evaluating stores according to the candidate coupon information, selecting recommended coupons using the evaluation results as weights, and obtaining final coupon information;

[0102] Acquire candidate coupon information, extract store information corresponding to the coupon according to the candidate coupon information, and obtain store information;

[0103] Based on big data retrieval and combined with store information, data collection is performed to collect the target store's rating information, evaluation information, and store attribute information to obtain store feature information;

[0104] Constructing a store evaluation model, inputting the store feature information into the store evaluation model for evaluation, and obtaining store evaluation result information;

[0105] Acquire attribute analysis result information, pre-screen the candidate coupon information according to the store evaluation result information, and obtain pre-screening result information;

[0106] The attribute analysis result information is used as a selection index and a corresponding weight is set. The pre-screening result information is weighted and sorted. Coupon selection is performed based on the sorting result to obtain final coupon recommendation information.

[0107] It should be noted that, for the candidate coupon information, only the user's coupon usage preference is considered, but there are still coupons that do not fit the user's preferences and interests. At the same time, for the stores corresponding to the candidate coupons, there are also some stores with poor reviews. Therefore, by evaluating the stores corresponding to the candidate coupon information, evaluating the quality and scale of the stores through the scores and reviews of the target stores, and then combining the user's attribute analysis results information, coupon selection is carried out to fit the user's preferences, and bad stores and stores that do not meet the user's preferences are filtered out. For example, users may like large and clean stores, etc. By comprehensively considering multiple factors such as store evaluation and user attribute analysis, users are provided with final recommendations that are more in line with their needs and preferences, thereby improving the quality of personalized recommendations and user satisfaction.

[0108] Figure 2 A coupon recommendation flow chart provided by an embodiment of the present invention;

[0109] like Figure 2 As shown, the present invention provides a coupon recommendation flow chart, including:

[0110] S202, analyzing the historical behavior and real-time behavior of the target user, and analyzing the user's long-term preference and short-term preference;

[0111] S204, analyzing user attributes according to the user's long-term preference and short-term preference, and defining attributes of the target user;

[0112] S206, analyzing the coupon usage preference of the target user and the coupon usage inducement of the target user through the user's historical coupon usage information;

[0113] S208, performing attribute analysis on the existing coupons, analyzing the type attribute, time attribute, discount attribute, and starting amount attribute of the coupons;

[0114] S210, selecting a coupon that meets the coupon usage incentive of the target user from the existing coupons as a candidate recommended coupon;

[0115] S212, evaluating the stores corresponding to the candidate recommended coupons, and selecting and recommending coupons based on the attribute analysis result information of the target user.

[0116] Furthermore, user feedback information is obtained, and feature extraction is performed on the user feedback information to obtain user feedback feature information; text extraction is performed on the user feedback features based on NLP technology to obtain feedback text extraction information; a user feedback evaluation model is constructed, and the user feedback feature information and feedback text extraction information are input into the user feedback evaluation model for evaluation, the user's usage experience is evaluated, and user usage experience evaluation information is obtained; factors of user satisfaction and dissatisfaction are identified and analyzed based on the user usage experience evaluation information to obtain first analysis information; a sentiment analysis model is obtained, and the feedback text information is input into the sentiment analysis model for feedback sentiment analysis to obtain feedback sentiment analysis information; a feedback sentiment level judgment threshold is constructed, and the feedback sentiment analysis information is judged with the feedback sentiment level judgment threshold to obtain feedback sentiment level information; user satisfaction and dissatisfaction are analyzed based on the first analysis information, the feedback sentiment analysis information, and the feedback sentiment level information to obtain second analysis information; and recommendation weights of corresponding types of coupons are adjusted based on the second analysis information to improve user experience.

[0117] Figure 3 A personalized coupon push system framework based on user characteristics provided by an embodiment of the present invention Figure 3 The system includes: a memory 31 and a processor 32. The memory 31 includes a personalized coupon push method program based on user characteristics. When the personalized coupon push method program based on user characteristics is executed by the processor 32, the following steps are implemented:

[0118] Obtaining user historical behavior information, performing user preference analysis based on the user historical behavior information, and obtaining user preference analysis information;

[0119] Acquire user real-time behavior information, and predict user interests based on the user real-time behavior information to obtain user interest prediction information;

[0120] Combining the user preference analysis information and the user interest prediction information, the attribute analysis is performed on the target user to obtain the attribute analysis result information;

[0121] Obtain coupon information, perform attribute analysis and category classification on various coupons, and recommend personalized coupons to target users to obtain candidate coupon information;

[0122] Store analysis and evaluation are performed based on the candidate coupon information, and the evaluation results are used as weights to select recommended coupons to obtain final coupon information.

[0123] It should be noted that the present invention provides a personalized coupon recommendation method and system based on user characteristics. The long-term preferences of target users are analyzed and determined through the user's historical behaviors including purchasing behaviors, browsing behaviors and interactive behaviors. The short-term interests of target users are analyzed and determined through the user's real-time behaviors including browsing behaviors and interactive behaviors. The user's emotions towards the interactive object are analyzed from the interactive behaviors, thereby mapping the user's real-time interests, that is, short-term preferences. The user's attributes are defined through the user's short-term preferences and long-term preferences, and attribute labels are defined for the target user, representing the user's preference and interest type and direction. Furthermore, the user's coupon usage incentives are analyzed through the user's historical coupon usage information, and it is determined which coupon the user prefers to use, and then the coupon that meets the user's usage preferences is selected. At the same time, the store corresponding to the coupon is evaluated, the quality and scale of the store are evaluated, and the coupon is recommended in combination with the user's attributes to ensure the quality of the coupon, and fit the user's preferences and usage habits, improve user satisfaction, and promote the use, collection and purchase of coupons.

[0124] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0125] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0126] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0127] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0128] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0129] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A personalized coupon recommendation method based on user characteristics, characterized in that: include: Obtaining user historical behavior information, performing user preference analysis based on the user historical behavior information, and obtaining user preference analysis information; Acquire user real-time behavior information, and predict user interests based on the user real-time behavior information to obtain user interest prediction information; Combining the user preference analysis information and the user interest prediction information, the attribute analysis is performed on the target user to obtain the attribute analysis result information; Obtain coupon information, perform attribute analysis and category classification on various coupons, and recommend personalized coupons to target users to obtain candidate coupon information; Perform store analysis and evaluation based on the candidate coupon information, and use the evaluation results as weights to select recommended coupons to obtain final coupon information; The acquiring of user real-time behavior information, and performing user interest prediction according to the user real-time behavior information to obtain user interest prediction information specifically includes: Acquire user real-time behavior information, wherein the user real-time behavior information includes: user real-time browsing behavior information and user real-time interactive behavior information; Perform feature extraction on the user's real-time behavior information to extract the user's real-time browsing behavior object information, dwell time and click frequency, and user's real-time interactive behavior features to obtain user's real-time behavior feature information; Preset interest tags, calculate the Mahalanobis distance between the user's real-time browsing object and the interest tag, use it as a correlation judgment indicator and a preset threshold for judgment, and merge the corresponding stay time and click frequency into the corresponding interest tag to obtain preliminary analysis information; Using the dwell time and click frequency as interest judgment indicators, setting judgment thresholds corresponding to different interest scores respectively, judging the preliminary analysis information and the judgment thresholds, calculating the interest scores of different interest tags, judging the user's interest preference by the interest scores, and obtaining the first interest analysis information; Inputting the real-time behavior feature information into a sentiment analysis model to analyze the user's real-time interactive behavior sentiment, and obtaining real-time interactive behavior sentiment analysis information; Performing a prediction analysis on the real-time interest preference of the target user through the real-time interactive behavior sentiment analysis information and the first interest analysis information to obtain user interest prediction information; The attribute analysis of the target user is performed in combination with the user preference analysis information and the user interest prediction information to obtain the attribute analysis result information, specifically including: Obtain user preference analysis information and user interest prediction information, and build a user attribute analysis model based on the random forest algorithm; Preset user attribute labels, obtain feature information of various user attribute labels based on big data retrieval, build a training data set, and perform deep learning and training on the user attribute analysis model; Setting different analysis weights for user preferences and user interests, and inputting the user preference analysis information and user interest prediction information into the user attribute analysis model for analysis; The user preference analysis information and user interest prediction information are used as attribute features, and multiple decision trees are constructed for attribute analysis. Several features are randomly selected and node splitting is performed. The cosine similarity between each node and various attribute labels is calculated, and the cosine similarity is compared with the preset threshold to select the available nodes. Preset the selection threshold, extract the attribute labels of the available nodes and count the occurrence frequency, judge the occurrence frequency of each attribute label with the selection threshold, and obtain the attribute analysis result information; The attribute analysis result information includes: user long-term preference attributes and user short-term preference attributes; The obtaining of coupon information, attribute analysis and category classification of various coupons, and personalized coupon promotion for target users to obtain candidate coupon promotion information specifically includes: Obtain the user's historical coupon information, perform feature extraction on the user's historical coupon information, extract coupon type features, coupon validity features, discount features, and starting amount features, and obtain historical coupon feature information; Performing user usage incentive analysis based on the historical coupon usage feature information, and classifying the coupon usage categories based on the coupon type features using a clustering algorithm to obtain usage category classification information; According to the coupon validity characteristics, discount characteristics and starting amount characteristics, corresponding distribution graphs are constructed and divided into intervals. The number of coupons used is used as a usage tendency analysis indicator. The user's usage tendency is analyzed in combination with the constructed distribution graph to obtain usage tendency analysis information. Combine the usage category classification information and the usage tendency analysis information to perform usage inducement analysis to obtain usage inducement analysis information; Obtaining coupon information, performing attribute analysis on the coupon information, analyzing type attributes, time validity attributes, discount attributes, and starting amount attributes of various coupons, and obtaining coupon attribute analysis information; Generate four matching weights of the type attribute, time attribute, discount attribute and starting amount attribute of the coupon according to the usage incentive analysis information, perform weighted calculation on the coupon attribute analysis information, calculate the matching weights of the attributes of various coupons and the usage incentives, and obtain matching degree information; The matching degree information is judged against a preset threshold, and coupons are recommended based on the judgment result to obtain candidate coupon recommendation information.

2. A personalized coupon recommendation method based on user characteristics according to claim 1, characterized in that: The obtaining of user historical behavior information and performing user preference analysis based on the user historical behavior information specifically includes: Acquire user historical behavior information, wherein the user historical behavior information includes: user historical purchase behavior, user historical browsing behavior and user historical interaction behavior; Extracting features from the user's historical behavior information to extract the user's historical purchase type, frequency and price features, historical browsing behavior features and historical interaction behavior features to obtain the user's historical behavior feature information; Based on the clustering algorithm and combined with the user's historical behavior characteristics information, the user's purchase preference analysis is carried out, and the user's purchase behavior is divided into two categories: purchase type and purchase price, and the purchase frequency is combined into the two categories to obtain the user's historical purchase analysis information; Taking the purchase frequency as a preference judgment indicator, the user preference analysis is performed in combination with the user's historical purchase analysis information to obtain the first preference analysis information; Analyze the user's browsing preferences through the user's historical behavior feature information, preset browsing categories, calculate the Manhattan distance between the user's historical browsing behavior and the browsing category, and judge it with the preset threshold to obtain the user's browsing category information; Extracting the browsing time and frequency of the user through the historical behavior feature information of the user, setting the corresponding weights, performing weighted calculation on the browsing category information of the user, analyzing the browsing preference of the user through the weighted calculation result, and obtaining the second preference analysis information; Constructing a sentiment analysis model, inputting the user's historical behavior feature information into the sentiment analysis model to analyze the sentiment of the user's historical interactive behavior, and obtaining historical interactive behavior sentiment analysis information; The user's historical interaction objects and interaction frequency are extracted through the historical interaction behavior sentiment analysis information, and the interaction frequency is used as a preference judgment indicator to obtain the third preference analysis information; The first preference analysis information, the second preference analysis information and the third preference analysis information are integrated to obtain the user preference analysis information.

3. A personalized coupon recommendation method based on user characteristics according to claim 1, characterized in that: The step of evaluating a store based on the candidate coupon information and screening and selecting coupons based on the attribute analysis result information to obtain the final coupon information specifically includes: Acquire candidate coupon information, extract store information corresponding to the coupon according to the candidate coupon information, and obtain store information; Based on big data retrieval and combined with store information, data collection is performed to collect the target store's rating information, evaluation information, and store attribute information to obtain store feature information; Constructing a store evaluation model, inputting the store feature information into the store evaluation model for evaluation, and obtaining store evaluation result information; Acquire attribute analysis result information, pre-screen the candidate coupon information according to the store evaluation result information, and obtain pre-screening result information; The attribute analysis result information is used as a selection index and a corresponding weight is set. The pre-screening result information is weighted and sorted. Coupon selection is performed based on the sorting result to obtain final coupon recommendation information.

4. A personalized coupon recommendation system based on user characteristics, characterized in that: The system includes: a memory and a processor, wherein the memory contains a personalized coupon push method program based on user characteristics, and when the personalized coupon push method program based on user characteristics is executed by the processor, the following steps are implemented: Obtaining user historical behavior information, performing user preference analysis based on the user historical behavior information, and obtaining user preference analysis information; Acquire user real-time behavior information, and predict user interests based on the user real-time behavior information to obtain user interest prediction information; Combining the user preference analysis information and the user interest prediction information, the attribute analysis is performed on the target user to obtain the attribute analysis result information; Obtain coupon information, perform attribute analysis and category classification on various coupons, and recommend personalized coupons to target users to obtain candidate coupon information; Perform store analysis and evaluation based on the candidate coupon information, and use the evaluation results as weights to select recommended coupons to obtain final coupon information; The acquiring of user real-time behavior information, and performing user interest prediction according to the user real-time behavior information to obtain user interest prediction information specifically includes: Acquire user real-time behavior information, wherein the user real-time behavior information includes: user real-time browsing behavior information and user real-time interactive behavior information; Perform feature extraction on the user's real-time behavior information to extract the user's real-time browsing behavior object information, dwell time and click frequency, and user's real-time interactive behavior features to obtain user's real-time behavior feature information; Preset interest tags, calculate the Mahalanobis distance between the user's real-time browsing object and the interest tag, use it as a correlation judgment indicator and a preset threshold for judgment, and merge the corresponding stay time and click frequency into the corresponding interest tag to obtain preliminary analysis information; Using the dwell time and click frequency as interest judgment indicators, setting judgment thresholds corresponding to different interest scores respectively, judging the preliminary analysis information and the judgment thresholds, calculating the interest scores of different interest tags, judging the user's interest preference by the interest scores, and obtaining the first interest analysis information; Inputting the real-time behavior feature information into a sentiment analysis model to analyze the user's real-time interactive behavior sentiment, and obtaining real-time interactive behavior sentiment analysis information; Performing a prediction analysis on the real-time interest preference of the target user through the real-time interactive behavior sentiment analysis information and the first interest analysis information to obtain user interest prediction information; The attribute analysis of the target user is performed in combination with the user preference analysis information and the user interest prediction information to obtain the attribute analysis result information, specifically including: Obtain user preference analysis information and user interest prediction information, and build a user attribute analysis model based on the random forest algorithm; Preset user attribute labels, obtain feature information of various user attribute labels based on big data retrieval, build a training data set, and perform deep learning and training on the user attribute analysis model; Setting different analysis weights for user preferences and user interests, and inputting the user preference analysis information and user interest prediction information into the user attribute analysis model for analysis; The user preference analysis information and user interest prediction information are used as attribute features, and multiple decision trees are constructed for attribute analysis. Several features are randomly selected and node splitting is performed. The cosine similarity between each node and various attribute labels is calculated, and the cosine similarity is compared with the preset threshold to select the available nodes. Preset the selection threshold, extract the attribute labels of the available nodes and count the occurrence frequency, judge the occurrence frequency of each attribute label with the selection threshold, and obtain the attribute analysis result information; The attribute analysis result information includes: user long-term preference attributes and user short-term preference attributes; The obtaining of coupon information, attribute analysis and category classification of various coupons, and personalized coupon promotion for target users to obtain candidate coupon promotion information specifically includes: Obtain the user's historical coupon information, perform feature extraction on the user's historical coupon information, extract coupon type features, coupon validity features, discount features, and starting amount features, and obtain historical coupon feature information; Performing user usage incentive analysis based on the historical coupon usage feature information, and classifying the coupon usage categories based on the coupon type features using a clustering algorithm to obtain usage category classification information; According to the coupon validity characteristics, discount characteristics and starting amount characteristics, corresponding distribution graphs are constructed and divided into intervals. The number of coupons used is used as a usage tendency analysis indicator. The user's usage tendency is analyzed in combination with the constructed distribution graph to obtain usage tendency analysis information. Combine the usage category classification information and the usage tendency analysis information to perform usage inducement analysis to obtain usage inducement analysis information; Obtaining coupon information, performing attribute analysis on the coupon information, analyzing type attributes, time validity attributes, discount attributes, and starting amount attributes of various coupons, and obtaining coupon attribute analysis information; Generate four matching weights of the type attribute, time attribute, discount attribute and starting amount attribute of the coupon according to the usage incentive analysis information, perform weighted calculation on the coupon attribute analysis information, calculate the matching weights of the attributes of various coupons and the usage incentives, and obtain matching degree information; The matching degree information is judged against a preset threshold, and coupons are recommended based on the judgment result to obtain candidate coupon recommendation information.

5. A personalized coupon recommendation system based on user characteristics according to claim 4, characterized in that: The obtaining of user historical behavior information and performing user preference analysis based on the user historical behavior information specifically includes: Acquire user historical behavior information, wherein the user historical behavior information includes: user historical purchase behavior, user historical browsing behavior and user historical interaction behavior; Extracting features from the user's historical behavior information to extract the user's historical purchase type, frequency and price features, historical browsing behavior features and historical interaction behavior features to obtain the user's historical behavior feature information; Based on the clustering algorithm and combined with the user's historical behavior characteristics information, the user's purchase preference analysis is carried out, and the user's purchase behavior is divided into two categories: purchase type and purchase price, and the purchase frequency is combined into the two categories to obtain the user's historical purchase analysis information; Taking the purchase frequency as a preference judgment indicator, the user preference analysis is performed in combination with the user's historical purchase analysis information to obtain the first preference analysis information; Analyze the user's browsing preferences through the user's historical behavior feature information, preset browsing categories, calculate the Manhattan distance between the user's historical browsing behavior and the browsing category, and judge it with the preset threshold to obtain the user's browsing category information; Extracting the browsing time and frequency of the user through the historical behavior feature information of the user, setting the corresponding weights, performing weighted calculation on the browsing category information of the user, analyzing the browsing preference of the user through the weighted calculation result, and obtaining the second preference analysis information; Constructing a sentiment analysis model, inputting the user's historical behavior feature information into the sentiment analysis model to analyze the sentiment of the user's historical interactive behavior, and obtaining historical interactive behavior sentiment analysis information; The user's historical interaction objects and interaction frequency are extracted through the historical interaction behavior sentiment analysis information, and the interaction frequency is used as a preference judgment indicator to obtain the third preference analysis information; The first preference analysis information, the second preference analysis information and the third preference analysis information are integrated to obtain the user preference analysis information.

Images