Intelligent vending method and system for analyzing user consumption behavior
Through multi-channel data collection and timing topological analysis, consumer groups are divided and consumption model is constructed, which solves the problem of the lack of personalization and real-time nature of the traditional retail model, realizes intelligent sales recommendations, and improves market response capabilities.
Patent Information
- Application Number
- CN202411167661.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-23
- Publication Date
- 2025-01-03
- Estimated Expiration
- 2044-08-23
AI Technical Summary
The traditional retail model lacks personalization and real-timeness, making it difficult to meet consumers' growing personalized needs and rapidly changing market environment.
By collecting user consumption behavior data from multiple channels, preprocessing and timing topological analysis, dividing consumer groups, and building a consumption model mining model for intelligent sales recommendations.
It realizes personalized and real-time intelligent sales recommendations, improves the speed of consumer group classification and consumption model mining efficiency, and enhances market response capabilities.
Smart Images
Figure CN118691383B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent vending, and in particular to an intelligent vending method and system for analyzing user consumption behavior. Background Art
[0002] With the rapid development of the Internet and mobile technology, consumers' shopping behaviors are becoming increasingly diversified and personalized. As an important part of the field of marketing and business intelligence, consumer behavior analysis has undergone a transformation from traditional data analysis to big data and artificial intelligence technology. With the maturity of cloud computing, the Internet of Things and data mining technologies, companies can obtain a large amount of consumer behavior data from multiple channels (such as online shopping, social media, mobile applications, etc.). These data not only include consumers' purchase records and preferences, but also cover their social interactions, search behaviors, and geographic location information, providing a rich source of information for in-depth understanding of consumers. The traditional retail model is usually based on static product displays and unified promotion strategies, lacking personalization and real-time. This model seems to be unable to cope with the growing personalized needs of consumers and the rapidly changing market environment. Summary of the invention
[0003] In view of this, the present invention proposes an intelligent vending method based on user consumption behavior analysis, which obtains product information, compares prices and makes recommendation decisions through different channels to achieve intelligent vending.
[0004] To achieve the above object, the present invention provides an intelligent vending method for analyzing user consumption behavior, comprising the following steps:
[0005] S1: Collecting user consumption behavior data, where the user consumption behavior data includes transaction records of online shopping platforms, product browsing history, social media interaction data, and offline store purchase data, and preprocessing the collected user consumption behavior data to obtain preprocessed user consumption behavior data;
[0006] S2: grouping users based on the pre-processed user consumption behavior data, dividing users into different consumer groups, wherein time series topology structure analysis is an implementation method for the group division;
[0007] S3: Mining consumption patterns of consumption behavior data of users in different consumer groups to obtain consumption patterns of users in different consumer groups;
[0008] S4: Based on the user grouping results and consumption pattern mining results, a sales recommendation model is constructed to perform intelligent sales recommendations.
[0009] As a further improvement method of the present invention:
[0010] Optionally, collecting user consumption behavior data in step S1 includes:
[0011] Collect user consumption behavior data, including transaction records of online shopping platforms, product browsing history, social media interaction data, and offline store purchase data. The collected user consumption behavior data is in the form of:
[0012] ;
[0013] ;
[0014] ;
[0015] in:
[0016] Indicates the collected user consumption behavior data;
[0017] Indicates the consumption behavior data collected for the nth user;
[0018] represents the online shopping platform transaction records, product browsing history, social media interaction data, and offline store purchase data of the nth user in turn;
[0019] represents the M consumption behavior descriptions of the nth user in the ith consumption behavior data, where the mth consumption behavior description Corresponding to the description of the user's mth consumption behavior in the i-th consumption behavior data, ;
[0020] The collected user consumption behavior data X is preprocessed to obtain preprocessed user consumption behavior data.
[0021] Optionally, the preprocessing of the collected user consumption behavior data X includes:
[0022] The preprocessing process of user consumption behavior data X is as follows:
[0023] S11: Obtain any consumption behavior description in the user consumption behavior data X, and perform word segmentation on the consumption behavior description to obtain a word segmentation sequence of the consumption behavior description, where the consumption behavior description The corresponding word segmentation sequence is:
[0024] ;
[0025] in:
[0026] Describe consumer behavior The sth word segmentation result in Describe consumer behavior The total number of word segmentation results, ;
[0027] S12: Perform one-hot encoding on any word segmentation result in the word segmentation sequence, where the word segmentation result The one-hot encoding of ;
[0028] S13: performing type recognition on any segmentation results in the segmentation sequence, and screening segmentation results recognized as commodities;
[0029] S14: Perform word vector encoding processing based on type recognition on the one-hot encoding representation to obtain the word vector corresponding to the word segmentation result, where the word segmentation result The word vector encoding processing formula is:
[0030] ;
[0031] in:
[0032] Indicates word segmentation results The word vector encoding result of ;
[0033] represents the word vector encoding matrix, * represents the convolution operation;
[0034] represents an exponential function with a natural constant as base;
[0035] Description of consumer behavior The word vector representation of is:
[0036] ;
[0037] in:
[0038] Describe consumer behavior The word vector representation of ;
[0039] The pre-processed user consumption behavior data is:
[0040] ;
[0041] ;
[0042] ;
[0043] in:
[0044] Represents the pre-processed user consumption behavior data;
[0045] Consumer behavior data The preprocessing results are In order The preprocessing results.
[0046] Optionally, the user grouping based on the pre-processed user consumption behavior data in step S2 includes:
[0047] Based on the pre-processed user consumption behavior data, users are grouped into different consumer groups, wherein the temporal topological structure analysis is the implementation method of the group division, wherein the consumer group division process is as follows:
[0048] S21: Set the time series window length to L, and divide the pre-processed consumption behavior data of N users into window data sequence, where the preprocessed consumption behavior data The corresponding e-th window data sequence is:
[0049] ;
[0050] ;
[0051] in:
[0052] Represents the consumption behavior data after preprocessing The corresponding e-th window data sequence, ;
[0053] T stands for transpose;
[0054] Represents a window data sequence The j-th window data in ;
[0055] S22: Preprocessing of consumer behavior data The window data sequences are processed by time series transformation at different angles to obtain multi-angle window data sequences, where the window data sequences The corresponding multi-angle window data sequence is:
[0056] ;
[0057] ;
[0058] ;
[0059] ;
[0060] in:
[0061] Represents a window data sequence The corresponding multi-angle window data sequence;
[0062] Represents window data Three angle coefficients for timing transformation;
[0063] S23: Based on the multi-angle window data sequence, extract the topological features of the pre-processed consumer behavior data, where the pre-processed consumer behavior data The topological characteristics are:
[0064] ;
[0065] ;
[0066] in:
[0067] Represents the consumption behavior data after preprocessing The topological characteristics of
[0068] Represents a multi-angle window data sequence The corresponding standard deviation, Represents a multi-angle window data sequence The corresponding mean;
[0069] Represents a window data sequence Multi-angle temporal transformation gradient value;
[0070] S24: Calculate the topological feature distances of different users, and divide the N users into groups based on the topological feature distances to obtain K types of consumer group users.
[0071] Optionally, in step S24, the N users are divided into groups based on the topological feature distance to obtain K types of consumer group users, including:
[0072] S241: Calculate the topological characteristic distances between different users, where the nth user and the nth user have The topological characteristic distance between users is:
[0073] ;
[0074] in:
[0075] Indicates the relationship between the nth user and the The topological characteristic distance between users, , Indicates Topological features of the pre-processed consumption behavior data of users;
[0076] S242: Calculate the consumption behavior density of different users, where the consumption behavior density of the nth user is:
[0077] ;
[0078] in:
[0079] represents the consumption behavior density of the nth user;
[0080] Indicates the number of users whose topological characteristic distance to the nth user is lower than the preset distance threshold;
[0081] S243: Perform distance correction on the consumption behavior density of the user, where the consumption behavior density The distance correction formula is:
[0082] ;
[0083] in:
[0084] Indicates that the consumption behavior density of the nth user is higher than The minimum topological characteristic distance between any users, if the consumption behavior density is the maximum consumption behavior density of N users, then Set to positive infinity;
[0085] Indicates the density of consumption behavior The correction result of
[0086] S244: Select K users with the largest modified consumption behavior density as initial cluster centers;
[0087] S245: Calculate the topological feature distance from the user at the non-cluster center to the cluster center, and assign the user to the consumer group where the nearest cluster center is located;
[0088] S246: Calculate the mean of the topological features of all users in each type of consumer group, and use the mean of the topological features as the updated cluster center of the consumer group, and return to step S245 until the cluster center does not change and constitutes K types of consumer group users.
[0089] Optionally, in step S3, consumption pattern mining is performed on consumption behavior data of users in different consumer groups, including:
[0090] The consumption patterns of the consumption behavior data of users in different consumer groups are mined. The consumption pattern mining process of users in the kth consumer group is as follows:
[0091] S31: Construct product set R:
[0092] ;
[0093] in:
[0094] represents the u-th commodity in the commodity set R, and U represents the total number of commodities in the commodity set R;
[0095] S32: Calculate the frequency and density of each commodity in the k-th consumer group. The frequency of occurrence among users in the kth consumer group is:
[0096] ;
[0097] in:
[0098] Display products The frequency of occurrence among users in the kth consumer group;
[0099] represents the total number of commodities browsed, purchased and interacted by users in the kth consumer group, where the total number of commodities is the product of the commodity type and the number of browses, purchases and interactions of each commodity; in the embodiment of the present invention, the browses, purchases and interactions of each commodity correspond to the transaction records of the online shopping platform, the commodity browsing history, the social media interaction data and the purchase data of the offline store in the user consumption behavior data respectively;
[0100] Display products The number of times it appears in the users of the kth consumer group;
[0101] commodity The density among users in the kth consumer group is:
[0102] ;
[0103] in:
[0104] Display products Density among users in the kth consumer group;
[0105] represents the number of users in the kth consumer group;
[0106] Indicates the products browsed, purchased and interacted by users in the kth consumer group Number of users;
[0107] S33: Calculate the correlation degree of different commodities among users in the kth consumer group, where commodities With goods The degree of correlation between , ;
[0108] S34: Calculate the consumption pattern weights of different sales combinations, where the goods The consumption pattern weight is , sales combination The consumption pattern weights are:
[0109] ;
[0110] in:
[0111] Indicates sales combination The weight of consumption patterns;
[0112] S35: retaining the single commodity and the sales combination consisting of multiple commodities whose consumption pattern weight is greater than the preset threshold, and constructing the consumption pattern of the k-th consumer group user :
[0113] ;
[0114] in:
[0115] represents the dth sales combination retained, and D represents the consumption pattern The total number of sales combinations in .
[0116] Optionally, the calculation formula for the degree of association of the commodities is:
[0117] ;
[0118] in:
[0119] Indicates sales combination degree of association;
[0120] Represents the browsing, purchasing and interactive sales combinations among users in the kth consumer group Number of users;
[0121] Indicates sales combination The maximum time interval between user browsing, purchasing and interaction between any two products in ;
[0122] Indicates sales combination The maximum density of the commodity in Indicates sales combination The minimum density of the commodity.
[0123] Optionally, in step S4, constructing a sales recommendation model to perform intelligent sales recommendation includes:
[0124] Based on the user grouping results and consumption pattern mining results, a sales recommendation model is constructed for intelligent sales recommendation. The sales recommendation model is divided into a consumer group grouping part and a sales recommendation part. The consumer group grouping part is used to obtain the user's consumption behavior data and extract the topological features therein to group the user into consumer groups to obtain the user's consumer group categories. The sales recommendation part makes sales recommendations based on the consumption pattern mining results associated with the user's consumer group categories.
[0125] The intelligent vending recommendation process for user user is as follows:
[0126] S41: Obtain the consumption behavior data of user user and preprocess it to extract the topological features ;
[0127] S42: Calculate topological features The topological feature distance to the cluster centers of K consumer groups, and the consumer group with the closest topological feature distance is used as the consumer group grouping result of user user;
[0128] S43: Obtain corresponding consumption patterns according to the consumer group grouping result of the user user;
[0129] S44: querying the products that the user has recently browsed, purchased, and interacted with in sequence;
[0130] S45: The product is obtained by traversing the sales combinations in the consumption mode, and the sales combinations formed by the product are recommended.
[0131] In order to solve the above problems, the present invention provides an intelligent vending system for analyzing user consumption behavior, characterized in that the system comprises:
[0132] A data processing module is used to collect user consumption behavior data, pre-process the collected user consumption behavior data to obtain the pre-processed user consumption behavior data, and group users based on the pre-processed user consumption behavior data to divide users into different consumer groups;
[0133] The consumption pattern mining module is used to mine the consumption patterns of the consumption behavior data of users in different consumer groups and obtain the consumption patterns of users in different consumer groups;
[0134] The sales recommendation device is used to integrate user grouping results and consumption pattern mining results, build a sales recommendation model for intelligent sales recommendation.
[0135] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:
[0136] A memory storing at least one instruction;
[0137] Communication interface, enabling electronic equipment to communicate; and
[0138] The processor executes the instructions stored in the memory to implement the intelligent vending method for analyzing user consumption behavior as described above.
[0139] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned intelligent vending method for analyzing user consumption behavior.
[0140] Compared with the prior art, the present invention proposes an intelligent vending method based on user consumption behavior analysis, which has the following advantages:
[0141] First, this scheme proposes a classification method based on time series topological features. This scheme obtains different types of consumer behavior data from multiple channels to form a time series sequence. The time series sequence is divided into several window data sequences using time series windows, and the window data sequence is subjected to multi-angle transformation processing using different time series weighting methods. The time series information in the window data sequence is extracted from multiple angles to form a multi-angle window data sequence. The multi-angle time series transformation gradient value between the multi-angle window data sequence and the window data sequence is calculated, where the smaller the gradient value, the more all the commodities in the window data sequence are time-series robust, that is, they are not affected by time factors. The multi-angle time series transformation gradient value is constructed as the topological feature of the user consumption behavior data, which effectively captures the topological information of the high-dimensional data space, realizes the dimensionality reduction processing of the user consumption behavior data, improves the classification speed of consumer groups, and realizes the division of consumer groups.
[0142] At the same time, this solution proposes a consumption pattern mining method to mine the consumption pattern of consumption behavior data of users in different consumer groups. According to the associated purchase, browsing and interactive relationship of commodities in different consumer groups, the density of a single commodity and the degree of association between multiple commodities are obtained in combination with support calculation. A single commodity and multiple commodities are used as sales combinations to construct consumption pattern weights that characterize the importance of sales combinations, extract the consumption patterns in each consumer group, and build a sales recommendation model based on the user grouping results and consumption pattern mining results to perform intelligent sales recommendations. The model obtains the user's consumption behavior data and extracts the topological features therein, groups the users into consumer groups, obtains the user's consumer group categories, and performs sales recommendations based on the consumption pattern mining results associated with the user's consumer group categories. BRIEF DESCRIPTION OF THE DRAWINGS
[0143] Figure 1 A schematic diagram of a flow chart of an intelligent vending method for analyzing user consumption behavior provided by an embodiment of the present invention;
[0144] Figure 2 A functional module diagram of an intelligent vending system for analyzing user consumption behavior provided by an embodiment of the present invention;
[0145] Figure 2 In: 100 intelligent vending system for analyzing user consumption behavior, 101 data processing module, 102 consumption pattern mining module, 103 vending recommendation device;
[0146] 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
[0147] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.
[0148] The embodiment of the present application provides an intelligent vending method for analyzing user consumption behavior. The execution subject of the intelligent vending method for analyzing user consumption behavior includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided in the embodiment of the present application. In other words, the intelligent vending method for analyzing user consumption behavior can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.
[0149] S1: Collect user consumption behavior data, where the user consumption behavior data includes transaction records of online shopping platforms, product browsing history, social media interaction data, and offline store purchase data, and pre-process the collected user consumption behavior data to obtain pre-processed user consumption behavior data.
[0150] The user consumption behavior data collected in step S1 includes:
[0151] Collect user consumption behavior data, including transaction records of online shopping platforms, product browsing history, social media interaction data, and offline store purchase data. The collected user consumption behavior data is in the form of:
[0152] ;
[0153] ;
[0154] ;
[0155] in:
[0156] Indicates the collected user consumption behavior data;
[0157] Indicates the consumption behavior data collected for the nth user;
[0158] represents the online shopping platform transaction records, product browsing history, social media interaction data, and offline store purchase data of the nth user in turn;
[0159] represents the M consumption behavior descriptions of the nth user in the ith consumption behavior data, where the mth consumption behavior description Corresponding to the description of the user's mth consumption behavior in the i-th consumption behavior data, ;
[0160] The collected user consumption behavior data X is preprocessed to obtain preprocessed user consumption behavior data.
[0161] The preprocessing of the collected user consumption behavior data X includes:
[0162] The preprocessing process of user consumption behavior data X is as follows:
[0163] S11: Obtain any consumption behavior description in the user consumption behavior data X, and perform word segmentation on the consumption behavior description to obtain a word segmentation sequence of the consumption behavior description, where the consumption behavior description The corresponding word segmentation sequence is:
[0164] ;
[0165] in:
[0166] Describe consumer behavior The sth word segmentation result in Describe consumer behavior The total number of word segmentation results, ;
[0167] S12: Perform one-hot encoding on any word segmentation result in the word segmentation sequence, where the word segmentation result The one-hot encoding of ;
[0168] S13: performing type recognition on any segmentation results in the segmentation sequence, and screening segmentation results recognized as commodities;
[0169] S14: Perform word vector encoding processing based on type recognition on the one-hot encoding representation to obtain the word vector corresponding to the word segmentation result, where the word segmentation result The word vector encoding processing formula is:
[0170] ;
[0171] in:
[0172] Indicates word segmentation results The word vector encoding result of ;
[0173] represents the word vector encoding matrix, * represents the convolution operation;
[0174] represents an exponential function with a natural constant as base;
[0175] Constituent consumer behavior description The word vector representation of is:
[0176] ;
[0177] in:
[0178] Describe consumer behavior The word vector representation of ;
[0179] The pre-processed user consumption behavior data is:
[0180] ;
[0181] ;
[0182] ;
[0183] in:
[0184] Represents the pre-processed user consumption behavior data;
[0185] Consumer behavior data The preprocessing results are In order The preprocessing results.
[0186] S2: Group users based on the pre-processed user consumption behavior data and divide users into different consumer groups.
[0187] The step S2 includes grouping users based on the pre-processed user consumption behavior data, including:
[0188] Based on the pre-processed user consumption behavior data, users are grouped into different consumer groups, wherein the temporal topological structure analysis is the implementation method of the group division, wherein the consumer group division process is as follows:
[0189] S21: Set the time series window length to L, and divide the pre-processed consumption behavior data of N users into window data sequence, where the preprocessed consumption behavior data The corresponding e-th window data sequence is:
[0190] ;
[0191] ;
[0192] in:
[0193] Represents the consumption behavior data after preprocessing The corresponding e-th window data sequence, ;
[0194] T stands for transpose;
[0195] Represents a window data sequence The j-th window data in ;
[0196] S22: Preprocessing of consumer behavior data The window data sequences are processed by time series transformation at different angles to obtain multi-angle window data sequences, where the window data sequences The corresponding multi-angle window data sequence is:
[0197] ;
[0198] ;
[0199] ;
[0200] ;
[0201] in:
[0202] Represents a window data sequence The corresponding multi-angle window data sequence;
[0203] Represents window data Three angle coefficients for timing transformation;
[0204] S23: Based on the multi-angle window data sequence, extract the topological features of the pre-processed consumer behavior data, where the pre-processed consumer behavior data The topological characteristics are:
[0205] ;
[0206] ;
[0207] in:
[0208] Represents the consumption behavior data after preprocessing The topological characteristics of
[0209] Represents a multi-angle window data sequence The corresponding standard deviation, Represents a multi-angle window data sequence The corresponding mean;
[0210] Represents a window data sequence Multi-angle temporal transformation gradient value;
[0211] S24: Calculate the topological feature distances of different users, and divide the N users into groups based on the topological feature distances to obtain K types of consumer group users.
[0212] In the step S24, the N users are divided into groups based on the topological feature distance to obtain K types of consumer group users, including:
[0213] S241: Calculate the topological characteristic distances between different users, where the nth user and the nth user have The topological characteristic distance between users is:
[0214] ;
[0215] in:
[0216] Indicates the relationship between the nth user and the The topological characteristic distance between users, , Indicates Topological features of the pre-processed consumption behavior data of users;
[0217] S242: Calculate the consumption behavior density of different users, where the consumption behavior density of the nth user is:
[0218] ;
[0219] in:
[0220] represents the consumption behavior density of the nth user;
[0221] Indicates the number of users whose topological characteristic distance to the nth user is lower than the preset distance threshold;
[0222] S243: Perform distance correction on the consumption behavior density of the user, where the consumption behavior density The distance correction formula is:
[0223] ;
[0224] in:
[0225] Indicates that the consumption behavior density of the nth user is higher than The minimum topological characteristic distance between any users, if the consumption behavior density is the maximum consumption behavior density of N users, then Set to positive infinity;
[0226] Indicates the density of consumption behavior The correction result of
[0227] S244: Select K users with the largest modified consumption behavior density as initial cluster centers;
[0228] S245: Calculate the topological feature distance from the user at the non-cluster center to the cluster center, and assign the user to the consumer group where the nearest cluster center is located;
[0229] S246: Calculate the mean of the topological features of all users in each type of consumer group, and use the mean of the topological features as the updated cluster center of the consumer group, and return to step S245 until the cluster center does not change and constitutes K types of consumer group users.
[0230] .S3: Conduct consumption pattern mining on the consumption behavior data of users in different consumer groups to obtain the consumption patterns of users in different consumer groups.
[0231] In the step S3, consumption patterns are mined for consumption behavior data of users in different consumer groups, including:
[0232] The consumption patterns of the consumption behavior data of users in different consumer groups are mined. The consumption pattern mining process of users in the kth consumer group is as follows:
[0233] S31: Construct product set R:
[0234] ;
[0235] in:
[0236] represents the u-th commodity in the commodity set R, and U represents the total number of commodities in the commodity set R;
[0237] S32: Calculate the frequency and density of each commodity in the kth consumer group. The frequency of occurrence among users in the kth consumer group is:
[0238] ;
[0239] in:
[0240] Display products The frequency of occurrence among users in the kth consumer group;
[0241] represents the total number of products browsed, purchased, and interacted by users in the kth consumer group, where the total number of products is the product of the product type and the number of browses, purchases, and interactions for each product;
[0242] Display products The number of times it appears in the users of the kth consumer group;
[0243] commodity The density among users in the kth consumer group is:
[0244] ;
[0245] in:
[0246] Display products Density among users in the kth consumer group;
[0247] represents the number of users in the kth consumer group;
[0248] Indicates the products browsed, purchased and interacted by users in the kth consumer group Number of users;
[0249] S33: Calculate the correlation degree of different commodities among users in the kth consumer group, where commodities With goods The degree of correlation between , ;
[0250] S34: Calculate the consumption pattern weights of different sales combinations, where the goods The consumption pattern weight is , sales combination The consumption pattern weights are:
[0251] ;
[0252] in:
[0253] Indicates sales combination The weight of consumption patterns;
[0254] S35: retaining the single commodity and the sales combination consisting of multiple commodities whose consumption pattern weight is greater than the preset threshold, and constructing the consumption pattern of the k-th consumer group user :
[0255] ;
[0256] in:
[0257] represents the dth sales combination retained, and D represents the consumption pattern The total number of sales combinations in .
[0258] The calculation formula for the degree of association of the products is:
[0259] ;
[0260] in:
[0261] Indicates sales combination degree of association;
[0262] Represents the browsing, purchasing and interactive sales combinations among users in the kth consumer group Number of users;
[0263] Indicates sales combination The maximum time interval between user browsing, purchasing and interaction between any two products in ;
[0264] Indicates sales combination The maximum density of the commodity in Indicates sales combination The minimum density of the commodity.
[0265] .S4: Based on the user grouping results and consumption pattern mining results, a sales recommendation model is constructed to perform intelligent sales recommendations.
[0266] The step S4 constructs a sales recommendation model to perform intelligent sales recommendation, including:
[0267] Based on the user grouping results and consumption pattern mining results, a sales recommendation model is constructed for intelligent sales recommendation. The sales recommendation model is divided into a consumer group grouping part and a sales recommendation part. The consumer group grouping part is used to obtain the user's consumption behavior data and extract the topological features therein to group the user into consumer groups to obtain the user's consumer group categories. The sales recommendation part makes sales recommendations based on the consumption pattern mining results associated with the user's consumer group categories.
[0268] The intelligent vending recommendation process for user user is as follows:
[0269] S41: Obtain the consumption behavior data of user user and preprocess it to extract the topological features ;
[0270] S42: Calculate topological features The topological feature distance to the cluster centers of K consumer groups, and the consumer group with the closest topological feature distance is used as the consumer group grouping result of user user;
[0271] S43: Obtain corresponding consumption patterns according to the consumer group grouping result of the user user;
[0272] S44: querying the products that the user has recently browsed, purchased, and interacted with in sequence;
[0273] S45: The product is obtained by traversing the sales combinations in the consumption mode, and the sales combinations formed by the product are recommended.
[0274] like Figure 2 , which is a functional module diagram of an intelligent vending system for analyzing user consumption behavior provided by an embodiment of the present invention, which can implement the intelligent vending method for analyzing user consumption behavior in Example 1.
[0275] The intelligent vending system 100 for analyzing user consumption behavior of the present invention can be installed in an electronic device. According to the functions to be implemented, the intelligent vending system for analyzing user consumption behavior can include a data processing module 101, a consumption pattern mining module 102, and a vending recommendation device 103. The module of the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.
[0276] The data processing module 101 is used to collect user consumption behavior data, pre-process the collected user consumption behavior data to obtain the pre-processed user consumption behavior data, and group users based on the pre-processed user consumption behavior data to divide users into different consumer groups;
[0277] The consumption pattern mining module 102 is used to mine consumption patterns of consumption behavior data of users in different consumer groups to obtain consumption patterns of users in different consumer groups;
[0278] The product recommendation device 103 is used to integrate the user grouping results and the consumption pattern mining results, and construct a product recommendation model to perform intelligent product recommendation.
[0279] In detail, each module in the intelligent vending system 100 for analyzing user consumption behavior in the embodiment of the present invention is used in the same manner as described above. Figure 1 The intelligent vending method based on user consumption behavior analysis described in the present invention has the same technical means and can produce the same technical effects, so I will not go into details here.
[0280] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.
[0281] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.
[0282] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. An intelligent vending method based on user consumption behavior analysis, characterized in that: The method comprises: S1: Collecting user consumption behavior data, where the user consumption behavior data includes transaction records of online shopping platforms, product browsing history, social media interaction data, and offline store purchase data, and preprocessing the collected user consumption behavior data to obtain preprocessed user consumption behavior data; The user consumption behavior data collected in step S1 includes: Collect user consumption behavior data, including transaction records of online shopping platforms, product browsing history, social media interaction data, and offline store purchase data. The collected user consumption behavior data is in the form of: in: Indicates the collected user consumption behavior data; Indicates the consumption behavior data collected for the nth user; represents the online shopping platform transaction records, product browsing history, social media interaction data, and offline store purchase data of the nth user in turn; represents the M consumption behavior descriptions of the nth user in the ith consumption behavior data, where the mth consumption behavior description Corresponding to the description of the user's mth consumption behavior in the i-th consumption behavior data, ; Preprocessing the collected user consumption behavior data X to obtain preprocessed user consumption behavior data; S2: Group users based on the pre-processed user consumption behavior data and divide users into different consumer groups; The process of dividing consumer groups is as follows: S21: Set the time series window length to L, and divide the pre-processed consumption behavior data of N users into window data sequence, where the preprocessed consumption behavior data The corresponding e-th window data sequence is: ； in: Represents the consumption behavior data after preprocessing The corresponding e-th window data sequence, ; T stands for transpose; Represents a window data sequence The j-th window data in ; S22: Preprocessing of consumer behavior data The window data sequences are processed by time series transformation at different angles to obtain multi-angle window data sequences, where the window data sequences The corresponding multi-angle window data sequence is: ； ； ； ； in: Represents a window data sequence The corresponding multi-angle window data sequence; Represents window data Three angle coefficients for timing transformation; S23: Based on the multi-angle window data sequence, extract the topological features of the pre-processed consumer behavior data, where the pre-processed consumer behavior data The topological characteristics are: ； ； in: Represents the consumption behavior data after preprocessing The topological characteristics of Represents a multi-angle window data sequence The corresponding standard deviation, Represents a multi-angle window data sequence The corresponding mean; Represents a window data sequence Multi-angle temporal transformation gradient value; S24: Calculate the topological feature distances of different users, and divide the N users into groups based on the topological feature distances to obtain K types of consumer group users; S3: Mining consumption patterns of consumption behavior data of users in different consumer groups to obtain consumption patterns of users in different consumer groups; S4: Based on the user grouping results and consumption pattern mining results, a sales recommendation model is constructed to perform intelligent sales recommendations.
2. The intelligent vending method based on user consumption behavior analysis according to claim 1, characterized in that: The preprocessing of the collected user consumption behavior data X includes: The preprocessing process of user consumption behavior data X is as follows: S11: Obtain any consumption behavior description in the user consumption behavior data X, and perform word segmentation on the consumption behavior description to obtain a word segmentation sequence of the consumption behavior description, where the consumption behavior description The corresponding word segmentation sequence is: in: Describe consumer behavior The sth word segmentation result in Describe consumer behavior The total number of word segmentation results, ; S12: Perform one-hot encoding on any word segmentation result in the word segmentation sequence, where the word segmentation result The one-hot encoding of ; S13: performing type recognition on any segmentation results in the segmentation sequence, and screening segmentation results recognized as commodities; S14: Perform word vector encoding processing based on type recognition on the one-hot encoding representation to obtain the word vector corresponding to the word segmentation result, where the word segmentation result The word vector encoding processing formula is: ； in: Indicates word segmentation results The word vector encoding result of ; represents the word vector encoding matrix, * represents the convolution operation; represents an exponential function with a natural constant as base; Constituent consumer behavior description The word vector representation of is: ； in: Describe consumer behavior The word vector representation of ; The pre-processed user consumption behavior data is: ； ； ； in: Represents the pre-processed user consumption behavior data; Consumer behavior data The preprocessing results are In order The preprocessing results.
3. The intelligent vending method based on user consumption behavior analysis according to claim 1, characterized in that: In the step S24, the N users are divided into groups based on the topological feature distance to obtain K types of consumer group users, including: S241: Calculate the topological characteristic distances between different users, where the nth user and the nth user have The topological characteristic distance between users is: ； in: Indicates the relationship between the nth user and the The topological characteristic distance between users, , Indicates Topological features of the pre-processed consumption behavior data of users; S242: Calculate the consumption behavior density of different users, where the consumption behavior density of the nth user is: in: represents the consumption behavior density of the nth user; Indicates the number of users whose topological characteristic distance to the nth user is lower than the preset distance threshold; S243: Perform distance correction on the consumption behavior density of the user, where the consumption behavior density The distance correction formula is: ； in: Indicates that the consumption behavior density of the nth user is higher than The minimum topological characteristic distance between any users, if the consumption behavior density is the maximum consumption behavior density of N users, then Set to positive infinity; Indicates the density of consumption behavior The correction result of S244: Select K users with the largest modified consumption behavior density as initial cluster centers; S245: Calculate the topological feature distance from the user at the non-cluster center to the cluster center, and assign the user to the consumer group where the nearest cluster center is located; S246: Calculate the mean of the topological features of all users in each type of consumer group, and use the mean of the topological features as the updated cluster center of the consumer group, and return to step S245 until the cluster center does not change and constitutes K types of consumer group users.
4. The intelligent vending method based on user consumption behavior analysis as claimed in claim 3, characterized in that: In the step S3, consumption patterns are mined based on the consumption behavior data of users in different consumer groups, including: The consumption patterns of the consumption behavior data of users in different consumer groups are mined. The consumption pattern mining process of users in the kth consumer group is as follows: S31: Construct product set R: ； in: represents the u-th commodity in the commodity set R, and U represents the total number of commodities in the commodity set R; S32: Calculate the frequency and density of each commodity in the kth consumer group. The frequency of occurrence among users in the kth consumer group is: in: Display products The frequency of occurrence among users in the kth consumer group; represents the total number of products browsed, purchased, and interacted by users in the kth consumer group, where the total number of products is the product of the product type and the number of browses, purchases, and interactions for each product; Display products The number of times it appears in the users of the kth consumer group; commodity The density among users in the kth consumer group is: ； in: Display products Density among users in the kth consumer group; represents the number of users in the kth consumer group; Indicates the products browsed, purchased and interacted by users in the kth consumer group Number of users; S33: Calculate the correlation degree of different commodities among users in the kth consumer group, where commodity With goods The degree of correlation between , ; S34: Calculate the consumption pattern weights of different sales combinations, where the goods The consumption pattern weight is , sales combination The consumption pattern weights are: ； in: Indicates sales combination The weight of consumption patterns; S35: retaining the single commodity and the sales combination consisting of multiple commodities whose consumption pattern weight is greater than the preset threshold, and constructing the consumption pattern of the k-th consumer group user : ； in: represents the dth sales combination retained, and D represents the consumption pattern The total number of sales combinations in .
5. The intelligent vending method based on user consumption behavior analysis as claimed in claim 4, characterized in that: The calculation formula for the degree of association of the products is: ； in: Indicates sales combination degree of association; Represents the browsing, purchasing and interactive sales combinations among users in the kth consumer group Number of users; Indicates sales combination The maximum time interval between user browsing, purchasing and interaction between any two products in ; Indicates sales combination The maximum density of the commodity in Indicates sales combination The minimum density of the commodity.
6. The intelligent vending method based on user consumption behavior analysis according to claim 1, characterized in that: The step S4 constructs a sales recommendation model to perform intelligent sales recommendation, including: Based on the user grouping results and consumption pattern mining results, a sales recommendation model is constructed for intelligent sales recommendation. The sales recommendation model is divided into a consumer group grouping part and a sales recommendation part. The consumer group grouping part is used to obtain the user's consumption behavior data and extract the topological features therein to group the user into consumer groups to obtain the user's consumer group categories. The sales recommendation part makes sales recommendations based on the consumption pattern mining results associated with the user's consumer group categories. The intelligent vending recommendation process for user user is as follows: S41: Obtain the consumption behavior data of user user and preprocess it to extract the topological features ; S42: Calculate topological features The topological feature distance to the cluster centers of K consumer groups, and the consumer group with the closest topological feature distance is used as the consumer group grouping result of user user; S43: Obtain corresponding consumption patterns according to the consumer group grouping result of the user user; S44: querying the products that the user has recently browsed, purchased, and interacted with in sequence; S45: The product is obtained by traversing the sales combinations in the consumption mode, and the sales combinations formed by the product are recommended.
7. An intelligent vending system for analyzing user consumption behavior, characterized in that: The system comprises: A data processing module is used to collect user consumption behavior data, pre-process the collected user consumption behavior data to obtain the pre-processed user consumption behavior data, and group users based on the pre-processed user consumption behavior data to divide users into different consumer groups; The consumption pattern mining module is used to mine the consumption patterns of the consumption behavior data of users in different consumer groups and obtain the consumption patterns of users in different consumer groups; A sales recommendation device is used to integrate user grouping results and consumption pattern mining results, build a sales recommendation model for intelligent sales recommendation, so as to realize an intelligent sales method for user consumption behavior analysis as described in any one of claims 1-6.