Recommendation method and system considering commodity circle layer preference, product and electronic equipment

Abstract

The invention belongs to the technical field of recommendation systems, and provides a recommendation method and system considering commodity circle layer preferences, a product and electronic equipment. The recommendation method considering the commodity circle layer preference comprises the steps of obtaining user behavior data authorized by a user; obtaining a basic prediction score of the user on the commodity; determining the preference degree of the user for each commodity circle layer at the current moment through the user behavior data; predicting the preference degree of the user for each commodity circle layer at the next moment based on the preference degree of the user for each commodity circle layer at the current moment; adjusting the basic prediction score of the user on the commodity by using the preference degree of the user on the commodity circle layer to which the commodity belongs at the next moment, and obtaining the optimized score of the user on the commodity; and selecting M commodities with the highest optimization score to form a recommendation list of the user. The invention further provides a recommendation system, a computer program product and electronic equipment considering the commodity circle layer preference, the preference degree of the user for each commodity circle layer is introduced, the user demand is more accurately quantified, and more accurate recommendation is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of recommendation systems, and in particular to a recommendation method, system, product and electronic device that take into account the preferences of commodity circles. Background Art

[0002] The product recommendation method is to analyze the user's potential behavior pattern from a large amount of user behavior data, and recommend products of interest to users based on the potential behavior pattern. Products can include various physical products or software products or service products, such as books, tools, equipment, movies, music, online tutorials, home appliance repair services, etc.

[0003] Since the interaction record modeling of user products and the graph model have a high degree of match, the recommendation method based on graph neural network has become the main research direction. Although the existing graph neural network recommendation method has improved the recommendation accuracy to a certain extent, when recommending products in certain interest circles, it does not fully consider the influence of the circle characteristics of the products and their hierarchical relationship on the recommendation. In the existing social and consumer ecology, some products are often subdivided according to user interest circles (such as fishing circles, cycling circles, photography circles, etc.), and the products in different circles have a hierarchical system from entry-level to professional level. For each user interest circle, the typical user evolution path is from the initial interest exploration (preferring entry-level products in this circle) to gradually transitioning to more professional and high-end product preferences. This process reflects the strengthening and deepening of the user's association with a specific circle. However, the existing recommendation algorithms rarely explicitly consider the circle structure and hierarchical evolution relationship of products in modeling and strategy adjustment, resulting in an inaccurate description of the user's real purchase path and interest deepening process.

[0004] In addition, most current recommendation methods perform recommendation calculations under preset rules and fixed data sets. Such strategies may perform better under experimental conditions or when the data set is relatively static, but in actual applications, user preferences are often affected by multiple factors and change dynamically. Existing recommendation methods pay less attention to the evolution characteristics of user preferences in the time dimension, and fail to effectively track and adaptively adjust dynamic factors such as user usage cycles, changes in the external environment, and the influence of social trends. Summary of the invention

[0005] The present invention aims to at least solve the technical problem that existing recommendation methods have obvious deficiencies in quantifying and modeling the dynamic changes in product circle relationships and user interests, resulting in insufficient recommendation accuracy, and to provide a recommendation method, system, product and electronic device that takes product circle preferences into consideration.

[0006] In order to achieve the above-mentioned purpose of the present invention, the first aspect of the present invention provides a recommendation method that takes into account the preference of product circles, including: obtaining user behavior data authorized by the user, the user behavior data including at least one of the user's search, browsing, collection, purchase, comment and posting for multiple products; using a preset recommendation algorithm to process the user behavior data to obtain the user's basic predicted score for the product; determining the user's preference for each product circle at the current moment through the user behavior data, wherein a product circle corresponds to a user interest circle, and each product circle includes more than one product; based on the user's preference for each product circle at the current moment, predicting the user's preference for each product circle at the next moment; using the user's preference for the product circle to which the product belongs at the next moment to adjust the user's basic predicted score for the product, to obtain the user's optimized score for the product; selecting M products with the highest optimized scores to form a recommendation list for the user, and outputting the user's recommendation list, wherein M is a positive integer.

[0007] In order to achieve the above-mentioned purpose of the present invention, the second aspect of the present invention provides a recommendation system that considers product circle preferences, which is used to implement the above-mentioned recommendation method that considers product circle preferences, including: a data acquisition module, which acquires user behavior data authorized by the user, and the user behavior data includes at least one of the user's search, browsing, collection, purchase, comment and posting behavior on multiple products; a basic prediction score acquisition module, which uses a preset recommendation algorithm to process the user behavior data to obtain the user's basic prediction score for the product; a first preference determination module, which determines the user's preference for each product circle at the current moment through the user behavior data, wherein a product circle corresponds to a user interest circle, and each product circle includes more than one product; a second preference determination module, which predicts the user's preference for each product circle at the next moment based on the user's preference for each product circle at the current moment; a score optimization module, which adjusts the user's basic prediction score for the product by using the user's preference for the product circle to which the product belongs at the next moment, and obtains the user's optimized score for the product; an output module, which selects M products with the highest optimization scores to form a user's recommendation list, and outputs the user's recommendation list, wherein M is a positive integer.

[0008] In order to achieve the above-mentioned purpose of the present invention, the third aspect of the present invention provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the recommendation method considering product circle preferences described in the first aspect of the present invention.

[0009] In order to achieve the above-mentioned purpose of the present invention, the fourth aspect of the present invention provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the above-mentioned recommendation method considering the preferences of product circles.

[0010] The beneficial technical effect of the present invention is as follows: for the commodity field associated with the user's interest circle, first use a preset recommendation algorithm to obtain the user's basic predicted score for the commodity, and calculate the user's preference for each commodity circle at the current moment based on the user's behavior data, and then infer the user's preference for each commodity circle at the next moment based on the user's current preference for each commodity circle, and finally use the user's preference for each commodity circle at the next moment to adjust the basic predicted score to obtain an optimized score, and recommend commodities according to the optimized score sorting. The present invention is based on the user's social relationships and social behaviors, takes into account the circle attributes of the commodity, obtains the user's preference for each commodity circle, can more accurately quantify user needs, and provide more accurate recommendation services. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a flowchart of a recommendation method considering product circle preferences in a preferred embodiment of the present invention; Figure 2 It is a schematic diagram of the implementation process of a recommendation method considering product circle preferences in an application scenario of the present invention; Figure 3 It is a schematic structural diagram of an electronic device in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0012] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0013] In the description of the present invention, it is to be understood that the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicating orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0014] In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the internal connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0015] The execution subject of the recommendation method considering the commodity circle preference provided by the present invention includes but is not limited to at least one of the electronic devices such as the server and the terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the recommendation method considering the commodity circle preference can be executed by software or hardware installed on the terminal device or the server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms.

[0016] In a preferred embodiment, the recommendation method considering product circle preferences provided by the present invention is as follows: Figure 1 As shown, including: Step S1, obtaining user behavior data authorized by the user, the user behavior data includes at least one of the user's search, browsing, collection, purchase, comment and posting for multiple products. User behavior data authorized by the user means that the user behavior data involved in this application has obtained the user's permission and authorization, and does not involve violation of personal data privacy protection requirements.

[0017] In this embodiment, user behaviors include on-site behaviors and off-site behaviors. On-site behaviors refer to a series of interactive operations performed on the current commodity sales platform (such as Taobao, JD.com, etc.), such as commodity search, browsing, collection, purchase, etc., which are defined as: .

[0018] Off-site behavior refers to a series of product discussion behaviors on other social platforms (such as WeChat, Xiaohongshu, etc.), such as comments and posts, which are defined as: .

[0019] User behavior data includes on-site behavior and / or off-site behavior. Preferably, in order to more accurately capture user preferences, user behavior data includes at least one on-site behavior and at least one off-site behavior. User behavior data is the behavior data of the user in a period of time from the current moment to the present, including the name of each behavior and the execution time of each behavior.

[0020] Step S2: Use a preset recommendation algorithm to process the user behavior data to obtain the user's basic predicted rating for the product.

[0021] In this embodiment, the preset recommendation algorithm is an existing recommendation method. The preset recommendation algorithm can be a recommendation method based on a traditional mathematical model, such as a recommendation method based on matrix decomposition provided in the prior art "Y. Koren, R. Bell, and C. Volinsky, 'Matrixfactorization techniques for recommender systems,'Computer, vol. 42, no. 8, pp. 30–37, 2009.", or a recommendation method based on collaborative filtering disclosed in the prior art "D. Goldberg, D. Nichols, BM Oki, and D. Terry, 'Using collaborative filtering to weavean information tapestry,' Communications of the ACM, vol. 35, no. 12, pp. 61–70, 1992."

[0022] In this embodiment, the preset recommendation algorithm can also be a recommendation method based on a graph neural network. Step S2 is specifically: construct a user-product interaction graph based on user behavior data, input the user-product interaction graph into a pre-trained graph convolutional network model, and the graph convolutional network model outputs the user's basic predicted score for the product. The nodes in the user-product interaction graph are users and products, and there are only edges between user nodes and product nodes, that is, user-product interaction edges. Multiple users can share a user-product interaction graph, learn the hidden feature representation of users and products through a graph convolutional network, and then calculate each user's basic predicted score for each product through cosine similarity. The recommendation method based on graph neural network can also refer to existing solutions, such as the prior art "D. Goldberg, D. Nichols, BM Oki, and D. Terry, 'Using collaborative filtering to weave an information tapestry,'Communications ofthe ACM, vol. 35, no. 12, pp. 61–70, 1992.", and the prior art "R. vd Berg, TNKipf, and M. Welling, 'Graph convolutional matrix completion,'arXiv preprintarXiv:1706.02263, 2017.".

[0023] Step S3, determining the user's current preference for each product circle through user behavior data, wherein one product circle corresponds to one user interest circle, and each product circle includes more than one product.

[0024] In this embodiment, the user's current interest status is determined through user behavior data. The user's current interest status includes the user's current preference for each product circle. User interest circles include fishing circles, cycling circles, photography circles, etc. The products involved in each user interest circle are sorted out in advance through domain knowledge or prior settings to obtain a product set. Therefore, the products can be grouped Divided into multiple product circles, each product circle corresponds to a user interest circle, expressed as: Represents the product sets of the first, second, ..., and nth product circles respectively. n represents the number of user interest circles or product circles, which is a positive integer. time Interest status for: .

[0025] in, Respectively represent users At the moment Commodity circles preference.

[0026] In one example, a specific method for determining the user's current preference for each product circle through user behavior data may be: counting the number of product purchases or behaviors of the user in each product circle in the user behavior data (including on-site and off-site behaviors), normalizing the number of product purchases or behaviors of all product circles, and using the normalized value of each product circle as the user's current preference for each product circle.

[0027] Step S4, predicting the user's preference for each commodity circle at the next moment based on the user's preference for each commodity circle at the current moment.

[0028] In one example, the prediction method may be: based on user behavior data, statistics are collected on the user's purchase time period for each commodity circle (which may be the average of the purchase time intervals), and the time distance (i.e., the time difference, which is a positive number) between the user's last purchase of each commodity circle and the current time is obtained. The ratio of the time distance of each commodity circle to the purchase time period of the commodity circle is calculated, and the user's current preference for each commodity circle is multiplied by the aforementioned ratio corresponding to each commodity circle to obtain the user's preference for each commodity circle at the next moment.

[0029] Step S5, adjusting the user's basic predicted score for the product by using the user's preference for the product circle to which the product belongs at the next moment, and obtaining the user's optimized score for the product.

[0030] In this embodiment, the user's optimized score for the product can be obtained by multiplying the user's preference for the product circle to which the product belongs at the next moment by the user's basic predicted score for the product.

[0031] Step S6, select M products with the highest optimization scores to form a recommendation list for the user, and output the recommendation list for the user, where M is a positive integer. M is preferably, but not limited to, 5, 10, 15, or 20. Due to the diversity of user interest circles, the products in the recommendation list may not belong to the same product circle.

[0032] In a preferred embodiment, in order to more accurately obtain the user's preference for product circles, the user's behavior is comprehensively considered. In step S3, the user's current preference for each product circle is determined through user behavior data, including: Step S31, extract the user's behavior implementation record for each product from the user behavior data, and the behavior implementation record includes the behavior type. The user behavior data may not include the user's behavior for all product circles, may not completely include all products in a product circle, and may not include all types of behavior. Therefore, in step S31, extract the user's behavior implementation record for each product in the user behavior data. In addition to the behavior type, the behavior implementation record also includes the behavior time.

[0033] Step S32, calculating the user's current behavior score for each commodity based on the user's behavior record for each commodity.

[0034] Step S33, integrate the user's current behavior scores for the products in each product circle to obtain the user's current preference for each product circle. Specifically, in step S33, integrate the user's current behavior scores for all products in each product circle in the user's behavior data to obtain the user's current preference for each product circle.

[0035] In this embodiment, further preferably, in order to better capture the dynamic change characteristics of the user's preference for commodities, in step S32, based on the user's behavior record for each commodity, the user's current behavior score for each commodity is calculated, including: Step S321, obtaining the basic weight of each behavior in each commodity behavior implementation record according to the preset basic weight setting rule, and performing time decay adjustment and / or frequency enhancement adjustment on the basic weight to obtain the adjusted weight of each behavior in each commodity behavior implementation record.

[0036] Step S322, accumulating the adjustment weights of all behaviors in each commodity behavior implementation record to obtain the user's current behavior score for each commodity.

[0037] In this embodiment, the basic weight setting rule is: for the behavior within the site, the browsing behavior The user's interest signal is weak, but it may reflect potential attention. The basic weight should be set relatively low. Search behavior It is the behavior of users actively expressing their interest, and the basic weight should be set higher; collection behavior Expressing strong interest intention, the basic weight should be set higher; purchase behavior As the final decision-making behavior, the basic weight should be the highest. For off-site behavior, comment behavior and posting behavior Both reflect the attention paid to the circle, but the importance of posting behavior should be relatively high. Preferably, in order to facilitate the quantification of the basic weight of each behavior, the basic weight setting rule is set as a value table. In an example, the basic weight setting rule is shown in Table 1. According to Table 1, the basic weight of each behavior in each commodity behavior implementation record is obtained.

[0038] Table 1 Basic weight setting rules

[0039] In step S321, after obtaining the basic weight of each behavior in each commodity behavior implementation record, the following three methods can be used to obtain the adjustment weight of each behavior in each commodity behavior implementation record: Method 1: Adjust the basic weight by time decay to obtain the adjustment weight of each behavior in the implementation record of each commodity behavior. Adjustment weight : in, Expressing behavior The time decay factor, , Expressing behavior The time decay coefficient, Expressing behavior The basic weight of For behavior The length of the time interval with the current moment. For different types of behaviors, the time decay coefficient is set differently. For example, for search behavior: due to its strong immediacy, a larger time decay coefficient is set. ; Collection behavior: reflects long-term preference, set smaller ; The time decay coefficient of browsing behavior, purchasing behavior and other behaviors should be appropriately set according to actual conditions. The value of can be or or or or or In one example, refer to Table 2 for The value of .

[0040] Table 2 Time attenuation coefficient reference value setting table

[0041] The effectiveness of user interest expression in method 1 will gradually weaken over time, so it is necessary to introduce a time decay factor .

[0042] Method 2: Perform frequency enhancement adjustment on the basic weight to obtain the adjustment weight of each behavior in the implementation record of each commodity behavior. Adjustment weight : in, Expressing behavior The frequency factor of Expressing behavior In method 2, the frequency of a user's behavior reflects the degree of their interest. In order to prevent excessive interference caused by high-frequency data, it is necessary to use the frequency factor Adjust the base weight.

[0043] Method 3: Perform time decay and frequency enhancement adjustments on the basic weights to obtain the adjusted weights of each behavior in each product behavior implementation record. Adjustment weight : .

[0044] In a preferred implementation, in order to accurately predict the user's preference for multiple commodity circles at the next moment, the user's preference for multiple commodity circles is taken as the state, and a Kalman filter algorithm is used to predict the user's preference for multiple commodity circles at the next moment.

[0045] In this embodiment, the user interest state is modeled. The user interest changes dynamically, so Kalman filtering is used to perform state space modeling on the user's preference for each commodity circle (that constitutes the user interest state).

[0046] Defining Users In time The interest status is: in For users In time To the circle preference.

[0047] State transition model: Observation model: in, is the state transfer matrix. If the interest changes steadily over time, then Usually set to the identity matrix; is the input influence matrix, By calculating the user In time After calculating the preference of each product circle, the vectors obtained by normalizing each dimension are: Indicates the increment of the state caused by the user behavior. It's in time Users The state observation of interest. Represents process noise, which represents the random error in the transfer process. For example, Gaussian noise can be used to obtain a normally distributed random noise value; Represents observation noise, which indicates the error in user behavior data. For example, Gaussian noise can be used to obtain a normally distributed random noise value. is the observation matrix, which describes how the state is mapped to the observation value. If the observation value corresponds directly to the state, then is the identity matrix. According to Kalman filtering, the matrix is ​​updated through continuous learning , , , using the updated matrix , , , obtain users through the state transfer model Next moment Interest status , that is, the user The preference for multiple product circles at the next moment.

[0048] In this embodiment, Kalman filtering is used to correct the error between the actual observation value and the predicted value, so as to more realistically reflect the user's interest state. The Kalman filtering method can find the preference between the actual observation value and the predicted value that best reflects the real user's interest.

[0049] In a preferred embodiment, in step S5, the user's preference for the commodity circle to which the commodity belongs at the next moment is used to adjust the user's basic predicted score for the commodity, and in the step of obtaining the user's optimized score for the commodity, the user For Products have an optimization score of : in, Indicates user For The basic predicted score of each product; Indicates the circle interest adjustment coefficient, which can be set based on experience; Indicates user Next moment For The product circle to which the product belongs preference.

[0050] In a preferred embodiment, in order to achieve more accurate recommendations, in step S5, the user's preference for the product circle to which the product belongs at the next moment is used to adjust the user's basic predicted score for the product, and in the step of obtaining the user's optimized score for the product, the user For Products have an optimization score of : in, Indicates user For The basic predicted score of each product; Indicates the circle interest adjustment coefficient; Indicates user Next moment For The product circle to which the product belongs preference of; Indicates user Next moment For The product circle to which the product belongs preference level; Indicates Products in the product circle The product level within Indicates fusion and The obtained fusion level. and the level of the product in the product circle Fusion, using the fusion level to adjust the basic prediction score, can make and To achieve more accurate recommendations.

[0051] In this implementation, the basic prediction score is adjusted according to the user's preference for each product circle at the next moment, so that the obtained optimized score is more accurate, thereby obtaining a more accurate recommendation list. After adjustment, the obtained optimized score set is: ; Indicates product set The model, that is, the number of goods. Indicates user For product set The optimized score matrix is ​​used to sort the products, and the M products with the highest scores are selected as the recommendation list: .

[0052] Indicates the user group consisting of the M products with the highest optimization scores Recommended list of Indicates from The M products with the highest optimization scores are selected to form a recommendation list.

[0053] In this embodiment, based on the user Next moment Interest status , you can get the user For Product Circle Preference By setting the interest status The value range of preference is divided into different intervals to realize the mapping between preference and preference level, such as user The next moment The product circle to which the product belongs Preference level for: in, Indicates user Next moment For The product circle to which the product belongs The mapping function at this time is It can be determined based on business experience or statistical analysis, including the following two modes: Empirical setting: pre-define the first preference threshold , the second preference threshold , so that when hour (Primary), when hour (Intermediate), when hour (advanced).

[0054] Data-driven: Cluster or quantile the distribution of historical interest values ​​of users in the circle to automatically determine the threshold.

[0055] In this embodiment, in order to better capture the dynamic change characteristics of user preferences for products, thereby more accurately quantifying user needs and providing more accurate recommendation services, preferably, the method for obtaining the product level of a product in the product circle to which it belongs is: Step a, obtaining historical purchase records of products in the product circle by multiple users in the user interest circle corresponding to the product circle; Step b, determining a purchase sequence of each user based on the historical purchase records of each user, the purchase sequence comprising a plurality of purchase time-product pairs arranged in chronological order of purchase time; Step c, linearly mapping the purchase time in the purchase sequence to the reference time interval to obtain a mapped purchase sequence, wherein the mapped purchase sequence includes a plurality of mapped purchase time-product pairs arranged in chronological order of the mapped purchase time; in the linear mapping relationship, the earliest purchase time in the purchase sequence corresponds to the starting time of the reference time interval, and the latest purchase time in the purchase sequence corresponds to the ending time of the reference time interval; Step d: Perform K-Means cluster analysis on the mapping purchase time in the mapping purchase sequence of multiple users to obtain Clusters are clustered by combining the mapped purchase sequences of multiple users. The product level of all products corresponding to the purchase time in the clusters is set to , which will be The item level of all mapped purchase times in the clusters is set to . . is a positive integer, Indicates the product level index within the product circle. is a positive integer, . The preferred value is 3 but not limited to it.

[0056] In this embodiment, the fusion function Fusion and The method of obtaining the fusion level is not limited to: Method 1, piecewise linear function: When hour, , you can set this time , ;when When moderately improving Values, such as , , which encourages users to explore higher-level items slightly; when and The level gap is too large (i.e. the level difference is 2) or Below the user's current preference level When , avoid recommending incompatible products; represents the first coefficient, represents the second coefficient; Method 2: Based on similarity or distance function: each level is regarded as a discrete point. Use a decreasing function (such as a Gaussian kernel) for weighting, Indicates the difference The absolute value of The numerical value and Inversely proportional to , to smoothly control the weighting intensity caused by level differences.

[0057] In this embodiment, the commodities in the circle are clustered and stratified by features, and then three layers of circle attributes are distinguished to express the commodities that may be liked by the circle with different preferences.

[0058] Figure 2 The following is a flowchart showing an implementation of the recommendation method considering product circle preferences provided by the present invention. Figure 2 As shown in the figure, the recommendation method considering product circle preferences is in execution, which mainly includes system initialization, behavior data collection and processing, product category circle label calculation, user circle interest state modeling, recommendation generation and other modules. The description of each module is as follows: System initialization module: This module is used to initialize the system's pre-data, mainly to define the classification of user behaviors and initialize parameters. The most critical contents in this module include user behavior classification, user behavior weight definition, and product category circle initialization.

[0059] Behavior data collection module: This module is used to count the user's historical interactive behavior data and collect real-time data on the user's behavior. The collected data will be cleaned and preprocessed in this module to provide data support for subsequent modules.

[0060] Commodity category circle label calculation module: This module is based on the historical statistical data of the commodity circle and classifies the commodities therein through a clustering method, so as to obtain the hierarchical label of each commodity in the specified commodity circle.

[0061] User circle interest state modeling module: This module calculates the user's preference for each circle, uses the Kalman filter method to model the state space, and continuously observes, corrects and predicts to predict the user's preference changes for each circle.

[0062] Recommendation generation module: This module first obtains the user's basic predicted rating for each product based on the graph convolution method, and then adjusts the predicted rating based on the user's predicted interest state space.

[0063] The recommendation method considering product circle preferences proposed by the present invention has the following advantages: 1. Based on existing social relationships and social behaviors, the circle attributes of commodities are considered, and based on this, a method for dividing commodity circles and grading and labeling in circles is proposed. After the preliminary division of circles based on prior knowledge, the commodities in the circles are clustered and stratified, and then three layers of circle attributes are distinguished to express the commodities that may be preferred by different preferences in the circle. By introducing the idea of ​​circle stratification, the dynamic changes of users' preferences for commodities can be better captured, so as to more accurately quantify user needs and provide more accurate recommendation services; 2. The definition and weight classification of user behavior are proposed, and a precise weight adjustment method is designed after data analysis of various behaviors. A user interest state space modeling method based on user-product circle preference is designed. By quantifying the user's preference for each circle as the state space, and using the Kalman filter method for correction and prediction, the prediction of user circle preference is obtained, which is then used for subsequent recommendation adjustments.

[0064] The present invention also provides a recommendation system that considers product circle preferences, which is used to implement the above-mentioned recommendation method that considers product circle preferences provided by the present invention. In a preferred embodiment, the device includes: A data acquisition module is used to acquire user behavior data authorized by the user, where the user behavior data includes at least one of the following behaviors: searching, browsing, collecting, purchasing, commenting, and posting on multiple products by the user; The basic prediction score acquisition module uses a preset recommendation algorithm to process user behavior data to obtain the user's basic prediction score for the product; A first preference determination module determines the user's preference for each commodity circle at the current moment through user behavior data, wherein one commodity circle corresponds to one user interest circle, and each commodity circle includes more than one commodity; The second preference determination module predicts the user's preference for each commodity circle at the next moment based on the user's preference for each commodity circle at the current moment; The rating optimization module adjusts the user's basic predicted rating of the product by using the user's preference for the product circle to which the product belongs at the next moment, and obtains the user's optimized rating of the product; The output module selects M products with the highest optimization scores to form a recommendation list for the user, and outputs the recommendation list for the user, where M is a positive integer.

[0065] In this embodiment, the data acquisition module, the basic prediction score acquisition module, the first preference determination module, the second preference determination module, the score optimization module and the output module correspond one by one to step S1, step S2, step S3, step S4, step S5 and step S6 in the above-mentioned recommendation method considering product circle preferences provided by the present invention, and will not be repeated here.

[0066] The present invention also discloses a computer program product, including a computer program. When the computer program is executed by a processor, it implements the steps of the above-mentioned recommendation method considering product circle preferences provided by the present invention. The computer program product should be understood as a software product that mainly implements its solution through a computer program, such as a program product integrated in the cloud or a software library.

[0067] The present invention also discloses an electronic device. In one embodiment, the electronic device includes at least one processor; and a memory connected to the at least one processor in communication; wherein: The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the recommendation method considering the commodity circle preference provided by the present invention.

[0068] like Figure 3 , which is a schematic diagram of the structure of an electronic device for a recommendation method considering product circle preferences provided by an embodiment of the present invention. The electronic device may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a recommendation method program considering product circle preferences.

[0069] Among them, the processor 10 can be composed of an integrated circuit in some embodiments, for example, it can be composed of a single packaged integrated circuit, or it can be composed of multiple integrated circuits with the same function or different functions, including one or more central processing units (CPU), microprocessors, digital processing chips, graphics processors and various control chips. The processor 10 is the control core (Control Unit) of the electronic device, which uses various interfaces and lines to connect the various components of the entire electronic device, and executes or executes the program or module stored in the memory 11 (for example, executing the recommendation method considering the preference of the commodity circle, etc.), and calls the data stored in the memory 11 to execute various functions of the electronic device and process data.

[0070] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (for example, SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of an electronic device, such as a mobile hard disk of the electronic device. In other embodiments, the memory 11 may also be an external storage device of an electronic device, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Furthermore, the memory 11 may also include both an internal storage unit of the electronic device and an external storage device. The memory 11 can not only be used to store application software and various types of data installed in the electronic device, such as the code of the recommendation method program considering the preference of the commodity circle, but also can be used to temporarily store data that has been output or is to be output.

[0071] The communication bus 12 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0072] The communication interface 13 is used for communication between the above-mentioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device and other electronic devices. The user interface may be a display (Display), an input unit (such as a keyboard (Keyboard)), and optionally, the user interface may also be a standard wired interface, a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device and to display a visual user interface.

[0073] Figure 3 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 3The structure shown does not constitute a limitation on the electronic device, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0074] For example, although not shown, the electronic device may also include a power source (such as a battery) for supplying power to various components. Preferably, the power source may be logically connected to at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include any components such as one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, and power status indicators. The electronic device may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0075] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited by this structure.

[0076] Furthermore, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM).

[0077] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", "an implementation", "a preferred implementation" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0078] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A recommendation method that considers product circle preferences, characterized in that: include: Obtaining user behavior data authorized by the user, where the user behavior data includes at least one of the following behaviors: searching, browsing, collecting, purchasing, commenting, and posting on multiple products; Use the preset recommendation algorithm to process user behavior data to obtain the user's basic predicted rating of the product; Determine the user's current preference for each product circle through user behavior data, where one product circle corresponds to one user interest circle, and each product circle includes more than one product; Predict the user's preference for each product circle at the next moment based on the user's current preference for each product circle; Use the user's preference for the product circle to which the product belongs at the next moment to adjust the user's basic predicted score for the product, and obtain the user's optimized score for the product; Select the M products with the highest optimization scores to form the user's recommendation list, and output the user's recommendation list, where M is a positive integer.

2. The method for recommending products taking into account the preferences of commodity circles as claimed in claim 1, characterized in that: Determining the user's current preference for each commodity circle through user behavior data includes: Extracting the user's behavior records for each product from the user behavior data, where the behavior records include the behavior types; Calculate the user's current behavior score for each product based on the user's behavior record for each product; Integrate the user's current behavioral ratings of the products in each product circle to obtain the user's current preference for each product circle.

3. The method for recommending products taking into account the preferences of commodity circles as claimed in claim 2, characterized in that: The step of calculating the user's current behavior score for each commodity based on the user's behavior record for each commodity includes: The basic weight of each behavior in each commodity behavior implementation record is obtained according to the preset basic weight setting rules, and the basic weight is adjusted by time decay and / or frequency enhancement to obtain the adjusted weight of each behavior in each commodity behavior implementation record; The adjustment weights of all behaviors in each product behavior implementation record are accumulated to obtain the user's current behavior score for each product.

4. The method for recommending products taking into account the preferences of commodity circles as claimed in claim 1, characterized in that: Taking the user's preference for multiple commodity circles as the state, the Kalman filter algorithm is used to predict the user's preference for multiple commodity circles at the next moment.

5. The method for recommending products taking into account the preferences of product circles according to any one of claims 1 to 4, characterized in that: In the step of adjusting the user's basic predicted score for the product by using the user's preference for the product circle to which the product belongs at the next moment to obtain the user's optimized score for the product, the user For Products have an optimization score of : ； in, Indicates user For The basic predicted score of each product; Indicates the circle interest adjustment coefficient; Indicates user Next moment For The product circle to which the product belongs preference.

6. The method for recommending products taking into account the preferences of product circles according to any one of claims 1 to 4, characterized in that: In the step of adjusting the user's basic predicted score for the product by using the user's preference for the product circle to which the product belongs at the next moment to obtain the user's optimized score for the product, the user For Products have an optimization score of : ； in, Indicates user For The basic predicted score of each product; Indicates the circle interest adjustment coefficient; Indicates user Next moment For The product circle to which the product belongs preference of; Indicates user Next moment For The product circle to which the product belongs preference level; Indicates Products in the product circle The product level within Indicates fusion and The obtained fusion level.

7. The method for recommending products taking into account the preferences of commodity circles as claimed in claim 6, characterized in that: The method to obtain the product level of a product within its corresponding product circle is as follows: Obtain historical purchase records of products in the product circle by multiple users in the user interest circle corresponding to the product circle; Determine a purchase sequence for each user based on the historical purchase records of each user, the purchase sequence comprising a plurality of purchase time-product pairs arranged in chronological order of purchase time; Linearly mapping the purchase time in the purchase sequence to a reference time interval to obtain a mapped purchase sequence, wherein the mapped purchase sequence includes a plurality of mapped purchase time-product pairs arranged in chronological order of the mapped purchase time; Perform K-Means cluster analysis on the mapping purchase time in the mapping purchase sequence of multiple users to obtain Clusters are clustered by combining the mapped purchase sequences of multiple users. The product level of all products corresponding to the purchase time in the clusters is set to , is a positive integer, Indicates the product level index within the product circle. is a positive integer, .

8. A recommendation system that considers product circle preferences, characterized by: A method for implementing the recommendation method considering product circle preferences as described in any one of claims 1 to 7, characterized by comprising: A data acquisition module is used to acquire user behavior data authorized by the user, where the user behavior data includes at least one of the following behaviors: searching, browsing, collecting, purchasing, commenting, and posting on multiple products by the user; The basic prediction score acquisition module uses a preset recommendation algorithm to process user behavior data to obtain the user's basic prediction score for the product; A first preference determination module determines the user's preference for each commodity circle at the current moment through user behavior data, wherein one commodity circle corresponds to one user interest circle, and each commodity circle includes more than one commodity; The second preference determination module predicts the user's preference for each commodity circle at the next moment based on the user's preference for each commodity circle at the current moment; The rating optimization module adjusts the user's basic predicted rating of the product by using the user's preference for the product circle to which the product belongs at the next moment, and obtains the user's optimized rating of the product; The output module selects M products with the highest optimization scores to form a recommendation list for the user, and outputs the recommendation list for the user, where M is a positive integer.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the recommendation method considering product circle preferences as described in any one of claims 1 to 7 are implemented.

10. An electronic device, characterized in that: The electronic device comprises: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the recommendation method considering the commodity circle preference as described in any one of claims 1 to 7.

Images