Intelligent shopping system with ranking recommendation function

Abstract

The invention relates to the technical field of shopping recommendation, in particular to an intelligent shopping system with a ranking recommendation function, and the system comprises a user price sensitivity modeling module which is used for building a price sensitivity hierarchical model and extracting price sensitivity features of a user in different time periods through time sequence analysis; performing price sensitivity layering on the users by using a clustering algorithm; the commodity price feature analysis module is used for extracting the price interval, discount information, promotion frequency and historical price fluctuation data of the commodity; constructing a commodity price feature matrix, and generating a commodity price attraction factor; and the recommendation sorting decision module is used for generating a personalized recommendation score R through weighted calculation of the price sensitivity factor and the commodity price attraction factor, and carrying out commodity sorting based on the personalized recommendation score R. According to the method, the matching degree of the recommended content and the user demand is improved through multi-dimensional dynamic modeling, and the recommendation accuracy and the user satisfaction degree are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of shopping recommendation, and in particular to an intelligent shopping system with a recommendation ranking function. Background Art

[0002] With the rapid development of e-commerce, recommendation systems have become an important tool for major platforms to improve user experience and product exposure. By analyzing user behavior data, recommendation systems can provide users with personalized product recommendations. However, existing recommendation systems still have the following significant deficiencies in practical applications: Most recommendation systems rely mainly on users’ historical behavior data (such as click and purchase records) for recommendations. However, this method has a weak ability to capture users’ dynamic needs, and the match between recommended content and users’ real-time needs is low, resulting in inaccurate recommendations.

[0003] Users’ sensitivity to product prices varies from person to person, but existing systems often lack in-depth modeling of users’ price preferences. They only use price as an auxiliary feature for sorting, but fail to fully explore users’ price preferences in different time periods (such as low price preference, high cost-performance preference, and high price acceptance). In addition, product price-related features (such as discount information and price fluctuations) have not been effectively integrated, resulting in limited performance of the system in processing price-related recommendations.

[0004] Existing recommendation systems usually use static rules or preset weights to sort recommended content, which makes it difficult to adapt to real-time changes in user behavior and market dynamics (such as promotions, price fluctuations). This not only reduces the timeliness of recommended content, but also weakens the user experience and purchase conversion rate. Summary of the invention

[0005] The invention provides an intelligent shopping system with a recommendation ranking function.

[0006] Smart shopping system with recommendation ranking function, including: User price sensitivity modeling module: collects users' historical purchase records, product browsing behavior and price click preference data; establishes a price sensitivity stratification model, including extracting users' price sensitivity characteristics in different time periods through time series analysis, including low price preference, high cost performance preference and high price acceptance; uses clustering algorithms to stratify users' price sensitivity, including high-sensitivity users, medium-sensitivity users and low-sensitivity users, and generates price sensitivity factors; Product price feature analysis module: extracts product price range, discount information, promotion frequency and historical price fluctuation data; constructs a product price feature matrix and generates product price attractiveness factors; Recommendation sorting decision module: Generates a personalized recommendation score R through weighted calculation of price sensitivity factor and commodity price attractiveness factor; sorts commodities based on the personalized recommendation score R, and continuously optimizes the recommendation sorting strategy through reinforcement learning to ensure that the recommendation results match user preferences and market dynamics.

[0007] Optionally, the user price sensitivity modeling module specifically includes: User behavior data collection: Use distributed data to collect user behavior data, including collecting users' historical purchase records, product browsing behavior and price click preference data from e-commerce platform transaction records, product browsing logs and user click behavior data, converting the format of the collected user behavior data and performing deduplication operations, associating user behavior data through timestamps, and building a user behavior sequence database; User price-sensitive feature extraction: Based on the time series analysis method, the user behavior data is grouped by time period, the price preferences in each time period are counted, and price-sensitive features are extracted, including: Low price preference characteristic factor : The proportion of products purchased or clicked by users whose prices are lower than the average price of similar products on the platform; High cost-performance preference characteristic factor : The frequency with which users tend to browse or purchase heavily discounted products; High price acceptance characteristic factor : The frequency with which users click on or purchase products that are priced higher than the average price of similar products on the platform; Take the price sensitivity features in different time periods as input data to generate the user price sensitivity feature matrix , expressed as: ,in, is the user price sensitivity feature matrix; Establishment of price sensitivity stratification model: Based on the user price sensitivity feature matrix, a clustering algorithm is used to stratify users into high-sensitivity users, medium-sensitivity users, and low-sensitivity users; clustering performance evaluation indicators (such as silhouette coefficient and clustering uniformity) are introduced to optimize the stratification results and ensure the robustness and accuracy of the stratification model; according to the stratification results, the price sensitivity factor of the user is allocated , Indicates user The price sensitivity factor describes the user's sensitivity to product prices in the current time period.

[0008] Optionally, the clustering algorithm adopts Means clustering algorithm, user price sensitive feature matrix To layer: ,in, represents the clustering objective function, Indicates user The price-sensitive feature vector of Represents the cluster center , Indicates the preset number of clusters (such as high-sensitivity users, medium-sensitivity users, and low-sensitivity users); Based on the clustering results, assign price sensitivity factors : ,in, Respectively represent users low price preference, high cost performance preference and high price acceptance, Indicates the weight of the corresponding factor, which is preset by the system.

[0009] Optionally, after the price sensitivity hierarchical model is established, it also includes dynamic updating of price sensitivity factors; Combined with user behavior data collected in real time, the price sensitivity feature matrix is ​​updated, and the real-time feature factors are merged with the historical feature factors to generate updated price sensitivity factors; the user's classification into high sensitivity, medium sensitivity and low sensitivity layers is dynamically adjusted to ensure the accuracy of the price sensitivity stratification model.

[0010] Optionally, the updating of the price sensitivity feature matrix specifically includes: Low price preference characteristic factor Update: ; High cost-performance preference characteristic factor Update: ; High price acceptance characteristic factor Update: ; in, The updated low price preference, high cost performance preference and high price acceptance; is a new factor calculated in real time; is the weight coefficient, which controls the fusion ratio of real-time feature factors and historical feature factors; Recalculate the price sensitivity factor using the updated characteristic factors : ,in, For users Update price sensitivity factor.

[0011] Optionally, the commodity price feature analysis module specifically includes: Product price information data extraction: Extract the price range of the product : ,in, is the price range of the product, Indicates the lowest price in history of the product. Indicates the highest price in history of a product. The data sources include historical transaction records of the product, promotion information, and platform price adjustment logs. Extract discount information of products : ,in, Indicates the current discount information of the product. The original price of the product. The current selling price of the product; Extract promotion frequency of products : ,in, is the product promotion frequency, Indicates the time point set of the product participating in the promotion. Indicates the total online time of the product; Extract commodity price fluctuations : ,in, Indicates the fluctuation range of commodity prices. For time point The corresponding price, Calculated as standard deviation, used to reflect price volatility; Construct commodity price feature matrix: commodity price feature matrix It consists of the following features: ,in, Represents the price feature matrix of the product, is the price range of the product, The current discount information for the product. is the product promotion frequency, The fluctuation range of commodity prices.

[0012] Optionally, the generating commodity price attractiveness factor Based on the price feature matrix, It is a commodity The price attractiveness factor is expressed as: ,in, A normalization function representing a price interval, used to standardize the price range, They are the weights of price range, discount information, promotion frequency and price fluctuation. The weights are preset by the system. It represents the inverse contribution of price fluctuations. The smaller the fluctuation, the higher the attractiveness.

[0013] Optionally, the personalized recommendation score in the recommendation ranking decision module By user's price sensitivity factor and the price attractiveness factor of the product The weighted calculation generates: ,in, Indicates the personalized recommendation score of the product to the user. : They are the weights of the user price sensitivity factor and the product price attractiveness factor respectively. The weight values ​​are adjusted through reinforcement learning to ensure that the recommendation score is in line with user preferences and platform operation goals.

[0014] Optionally, the recommendation ranking decision module also includes score regularization processing: to ensure the recommendation score To ensure comparability, the ratings of all products are regularized and standardized ratings are calculated. : ,in, Represents the standardized recommendation score, ranging from [0,1], They are the highest and lowest recommendation scores in the current candidate product set respectively; According to the standardized recommendation score Sort the products and generate a recommended ranking list of the products.

[0015] Optionally, the reinforcement learning is used to dynamically adjust The value of makes the recommendation results continuously optimized, which can be expressed as: ,in, are the current weight values ​​respectively; User feedback for and The score adjustment increment is Represents the learning rate parameter, which is used to control the magnitude of weight adjustment.

[0016] Beneficial effects of the present invention: The present invention, through the user price sensitivity modeling module, accurately captures the price sensitivity characteristics of users in different time periods, including low price preference, high cost performance preference and high price acceptance, and uses a clustering algorithm to establish a price sensitivity hierarchical model. Combined with the price attractiveness factor generated in the commodity price feature analysis module, the recommendation ranking decision module can dynamically generate a personalized recommendation score R with the user price sensitivity factor and the commodity price attractiveness factor as the core. Compared with the existing recommendation system that relies on single-dimensional data or static rules, the multi-dimensional dynamic modeling improves the matching degree between the recommended content and the user's needs, thereby improving the accuracy of the recommendation and the user's satisfaction.

[0017] The present invention dynamically adjusts the recommendation ranking weights through a reinforcement learning algorithm, so that the system can optimize the recommendation strategy in real time. The user price sensitivity factor and the commodity price attractiveness factor are continuously updated through real-time behavior data, ensuring that the system can quickly respond to changes in user interests and market dynamics. Through score regularization and dynamic ranking adjustment mechanism, the problems of ranking fixation and delayed response in traditional recommendation systems are effectively solved, especially showing extremely high adaptability during promotional activities or changes in user short-term interests.

[0018] The present invention introduces a price sensitivity hierarchical model, combines time series analysis and clustering algorithms, deeply matches the user's price sensitivity characteristics with the commodity price characteristics, and constructs a new recommendation logic framework. The price attractiveness factor integrates multi-dimensional characteristics such as price range, discount information, promotion frequency and price fluctuation, especially in exploring the attractiveness of cost-effective commodities and stable-priced commodities, optimizes the shortcomings of traditional single-dimensional price characteristics, and overcomes the defects of the prior art in insufficient support for long-term changes in user demand and unpopular commodity recommendations by dynamically updating the price sensitivity factor and adjusting the sorting rules in real time, thereby promoting the leap of the recommendation system from a single rule-oriented to intelligent dynamic optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 A schematic diagram of a recommendation function module of a smart shopping system according to an embodiment of the present invention; Figure 2 Schematic diagram of a price sensitivity feature matrix according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0022] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0023] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0024] like Figure 1-Figure 2 As shown, the intelligent shopping system with recommendation ranking function embeds a user price sensitivity modeling module, a commodity price feature analysis module, and a recommendation ranking decision module in the intelligent shopping system to obtain the final commodity recommendation ranking, including: User price sensitivity modeling module: collects users' historical purchase records, product browsing behavior and price click preference data; establishes a price sensitivity stratification model, including extracting users' price sensitivity characteristics in different time periods through time series analysis, including low price preference, high cost performance preference and high price acceptance; uses clustering algorithms to stratify users' price sensitivity, including high-sensitivity users, medium-sensitivity users and low-sensitivity users, and generates price sensitivity factors; Product price feature analysis module: extracts product price range, discount information, promotion frequency and historical price fluctuation data; constructs a product price feature matrix and generates product price attractiveness factors; Recommendation sorting decision module: Generates a personalized recommendation score R through weighted calculation of price sensitivity factor and commodity price attractiveness factor; sorts commodities based on the personalized recommendation score R, and continuously optimizes the recommendation sorting strategy through reinforcement learning to ensure that the recommendation results match user preferences and market dynamics.

[0025] The user price sensitivity modeling module specifically includes: User behavior data collection: Use distributed data to collect user behavior data, including collecting users' historical purchase records, product browsing behavior and price click preference data from e-commerce platform transaction records, product browsing logs and user click behavior data, converting the format of the collected user behavior data and performing deduplication operations, associating user behavior data through timestamps, and building a user behavior sequence database; Purchase history: ,in Indicates the purchase time, Indicates the price of the purchased goods; Product browsing behavior: ,in Indicates browsing time. Indicates the category of the product being browsed; Price Click Preference: ,in Indicates the click time. Indicates the price of the item clicked; User price-sensitive feature extraction: Based on the time series analysis method, the user behavior data is grouped by time period, the price preferences in each time period are counted, and price-sensitive features are extracted, including: Low price preference characteristic factor : The proportion of products purchased or clicked by users whose prices are lower than the average price of similar products on the platform, a characteristic factor of low price preference It is expressed as: ,in, Represents the collection of all purchase records of a user. Indicates the price of the item purchased by the user satisfy A subset of Indicates the product category The average price, Representing a collection Total number of High cost-performance preference characteristic factor : The frequency of users’ tendency to browse or purchase heavily discounted products, characteristic factor of preference for high cost performance It is expressed as: ,in, , indicating the product The discount strength, Represents a collection of items browsed or purchased by a user with discount information. Indicates the total number of all user behavior records; High price acceptance characteristic factor : The frequency of users clicking or purchasing products with a price higher than the average price of similar products on the platform, a characteristic factor of high price acceptance It is expressed as: ,in, Indicates the price of the item that the user clicked or purchased satisfy The remaining parameters are defined as above; Take the price sensitivity features in different time periods as input data to generate the user price sensitivity feature matrix , expressed as: ,in, is the user price sensitivity feature matrix; Establishment of price sensitivity stratification model: Based on the user price sensitivity feature matrix, a clustering algorithm is used to stratify users into high-sensitivity users, medium-sensitivity users, and low-sensitivity users; clustering performance evaluation indicators (such as silhouette coefficient and clustering uniformity) are introduced to optimize the stratification results and ensure the robustness and accuracy of the stratification model; according to the stratification results, the price sensitivity factor of the user is allocated , Indicates user The price sensitivity factor describes the user's sensitivity to product prices in the current time period.

[0026] Clustering algorithm used Means clustering algorithm, user price sensitive feature matrix To layer: ,in, represents the clustering objective function, Indicates user The price-sensitive feature vector of Represents the cluster center , Indicates the preset number of clusters (such as high-sensitivity users, medium-sensitivity users, and low-sensitivity users); Based on the clustering results, assign price sensitivity factors : ,in, Respectively represent users low price preference, high cost performance preference and high price acceptance, Indicates the weight of the corresponding factor, which is preset by the system.

[0027] After the price sensitivity stratification model is established, it also includes dynamic updating of price sensitivity factors; Combined with user behavior data collected in real time, the price sensitivity feature matrix is ​​updated, and the real-time feature factors are merged with the historical feature factors to generate updated price sensitivity factors; the user's classification into high sensitivity, medium sensitivity and low sensitivity layers is dynamically adjusted to ensure the accuracy of the price sensitivity stratification model.

[0028] Updating the price sensitivity feature matrix specifically includes: Low price preference characteristic factor Update: ; High cost-performance preference characteristic factor Update: ; High price acceptance characteristic factor Update: ; in, The updated low price preference, high cost performance preference and high price acceptance; is a new factor calculated in real time; is the weight coefficient, which controls the fusion ratio of real-time feature factors and historical feature factors; Recalculate the price sensitivity factor using the updated characteristic factors : ,in, For users Update price sensitivity factor.

[0029] The commodity price feature analysis module specifically includes: Product price information data extraction: Extract the price range of the product : ,in, is the price range of the product, Indicates the lowest price in history of the product. Indicates the highest price in history of a product. The data sources include historical transaction records of the product, promotion information, and platform price adjustment logs. Extract discount information of products : ,in, Indicates the current discount information of the product. The original price of the product. The current selling price of the product; Extract promotion frequency of products : ,in, is the product promotion frequency, Indicates the time point set of the product participating in the promotion. Indicates the total online time of the product; Extract commodity price fluctuations : ,in, Indicates the fluctuation range of commodity prices. For time point The corresponding price, Calculated as standard deviation, used to reflect price volatility; Construct commodity price feature matrix: commodity price feature matrix It consists of the following features: ,in, Represents the price feature matrix of the product, is the price range of the product, The current discount information for the product. is the product promotion frequency, The fluctuation range of commodity prices.

[0030] Generate commodity price attractiveness factor Based on the price feature matrix, It is a commodity The price attractiveness factor is expressed as: ,in, A normalization function representing a price interval, used to standardize the price range, They are the weights of price range, discount information, promotion frequency and price fluctuation. The weights are preset by the system. It represents the inverse contribution of price fluctuations. The smaller the fluctuation, the higher the attractiveness.

[0031] In order to normalize the price range of commodities and make them suitable for the calculation of price attractiveness factor, we define The expression is as follows: ,in, Represents the normalized value of the price range, ranging from [1,1], which is used to measure the attractiveness of the commodity price range. Indicates the midpoint of the price range of a commodity; The reference price that represents the user's preference is usually the average price of similar products that the user has recently purchased. Indicates the highest historical price of a commodity; Indicates the historical lowest price of a commodity; is a very small positive value to avoid the denominator being zero ( ).

[0032] Normalization logic explanation: when near When , it means that the price range of the product matches the user's reference price well, and the normalized value is close to 0; when When , it means the price is high and the normalized value is negative; when When , it means the price is low and the normalized value is positive.

[0033] Personalized recommendation scoring in the recommendation ranking decision module By user's price sensitivity factor and the price attractiveness factor of the product The weighted calculation generates: ,in, Indicates the personalized recommendation score of the product to the user. : They are the weights of the user price sensitivity factor and the product price attractiveness factor respectively. The weight values ​​are adjusted through reinforcement learning to ensure that the recommendation score is in line with user preferences and platform operation goals.

[0034] The recommendation ranking decision module also includes score regularization processing: to ensure the recommendation score To ensure comparability, the ratings of all products are regularized and standardized ratings are calculated. : ,in, Represents the standardized recommendation score, ranging from [0,1], They are the highest and lowest recommendation scores in the current candidate product set respectively; According to the standardized recommendation score Sort the products and generate a recommended ranking list of products: ,in, Display products Ranking, Indicates according to The values ​​are sorted from high to low, and the ranking results are output to the front-end interface of the smart shopping system for display to users.

[0035] Dynamic ranking adjustment mechanism: Combined with real-time user behavior data, it is used for dynamic adjustment through reinforcement learning. The value of makes the recommendation results continuously optimized, which can be expressed as: ,in, are the current weight values ​​respectively; User feedback for and The score adjustment increment is Represents the learning rate parameter, which is used to control the magnitude of weight adjustment.

[0036] Through the above, the recommendation ranking decision module can be based on the price sensitivity factor and commodity price attractiveness factor Generate user personalized recommendation scores , and generates an accurate product ranking list through standardized scoring and dynamic weight adjustment mechanism, thereby improving the personalization of recommendation results and user satisfaction.

[0037] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0038] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An intelligent shopping system with a recommendation ranking function, characterized in that: include: User price sensitivity modeling module: collects user's historical purchase records, product browsing behavior and price click preference data; Establish a price sensitivity stratification model, including extracting the price sensitivity characteristics of users in different time periods through time series analysis, including low price preference, high cost performance preference and high price acceptance; use clustering algorithms to stratify users by price sensitivity, including high-sensitivity users, medium-sensitivity users and low-sensitivity users, and generate price sensitivity factors; Product price feature analysis module: extracts product price range, discount information, promotion frequency and historical price fluctuation data; constructs a product price feature matrix and generates product price attractiveness factors; Recommendation sorting decision module: Generates a personalized recommendation score R through weighted calculation of price sensitivity factor and commodity price attractiveness factor; sorts commodities based on the personalized recommendation score R, and continuously optimizes the recommendation sorting strategy through reinforcement learning.

2. The intelligent shopping system with recommendation ranking function according to claim 1, characterized in that: The user price sensitivity modeling module specifically includes: User behavior data collection: Use distributed data to collect user behavior data, including collecting users' historical purchase records, product browsing behavior and price click preference data from e-commerce platform transaction records, product browsing logs and user click behavior data, converting the format of the collected user behavior data and performing deduplication operations, associating user behavior data through timestamps, and building a user behavior sequence database; User price-sensitive feature extraction: Based on the time series analysis method, the user behavior data is grouped by time period, the price preferences in each time period are counted, and price-sensitive features are extracted, including: Low price preference characteristic factor : The proportion of products purchased or clicked by users whose prices are lower than the average price of similar products on the platform; High cost-performance preference characteristic factor : The frequency with which users tend to browse or purchase heavily discounted products; High price acceptance characteristic factor : The frequency with which users click on or purchase products that are priced higher than the average price of similar products on the platform; Take the price sensitivity features in different time periods as input data to generate the user price sensitivity feature matrix , expressed as: ,in, is the user price sensitivity feature matrix; Establishment of price sensitivity stratification model: Based on the user price sensitivity feature matrix, a clustering algorithm is used to stratify users into high-sensitivity users, medium-sensitivity users, and low-sensitivity users; clustering performance evaluation indicators are introduced to optimize the stratification results; according to the stratification results, the price sensitivity factor of the user is allocated , Indicates user The price sensitivity factor describes the user's sensitivity to product prices in the current time period.

3. The intelligent shopping system with recommendation ranking function according to claim 2, characterized in that: The clustering algorithm uses Means clustering algorithm, user price sensitive feature matrix To layer: ,in, represents the clustering objective function, Indicates user The price-sensitive feature vector of Represents the cluster center , Indicates the preset number of clusters; Based on the clustering results, assign price sensitivity factors : ,in, Respectively represent users low price preference, high cost performance preference and high price acceptance, Indicates the weight of the corresponding factor.

4. The intelligent shopping system with recommendation ranking function according to claim 3, characterized in that: After the price sensitivity hierarchical model is established, it also includes dynamic updating of price sensitivity factors; Combined with the user behavior data collected in real time, the price sensitivity feature matrix is ​​updated, and the real-time feature factors are integrated with the historical feature factors to generate an updated price sensitivity factor.

5. The intelligent shopping system with recommendation ranking function according to claim 4, characterized in that: The updating price sensitivity feature matrix specifically includes: Low price preference characteristic factor Update: ; High cost-performance preference characteristic factor Update: ; High price acceptance characteristic factor Update: ; in, for the updated low price preference, high cost performance preference and high price acceptance; is a new factor calculated in real time; is the weight coefficient, which controls the fusion ratio of real-time feature factors and historical feature factors; Recalculate the price sensitivity factor using the updated characteristic factors .

6. The intelligent shopping system with recommendation ranking function according to claim 3, characterized in that: The commodity price feature analysis module specifically includes: extracting commodity price information data, extracting commodity price range , extract discount information of products , extract the promotion frequency of the product And extract commodity price fluctuations ; Construct commodity price feature matrix: commodity price feature matrix It consists of the following features: ,in, Represents the price feature matrix of the product, is the price range of the product, The current discount information for the product. is the product promotion frequency, The fluctuation range of commodity prices.

7. The intelligent shopping system with recommendation ranking function according to claim 6, characterized in that: Generating commodity price attractiveness factor Based on the price feature matrix, it is expressed as: ,in, A normalization function representing a price interval, used to standardize the price range, are the weights of price range, discount information, promotion frequency and price fluctuation, It represents the inverse contribution of price fluctuations. The smaller the fluctuation, the higher the attractiveness.

8. The intelligent shopping system with recommendation ranking function according to claim 7, characterized in that: Personalized recommendation scoring in the recommendation ranking decision module By user's price sensitivity factor and the price attractiveness factor of the product The weighted calculation generates: ,in, Indicates the personalized recommendation score of the product to the user. : They are the weights of the user price sensitivity factor and the commodity price attractiveness factor, and the weight values ​​are adjusted through reinforcement learning.

9. The intelligent shopping system with recommendation ranking function according to claim 8, characterized in that: The recommendation ranking decision module also includes score regularization processing, which performs regularization processing on the scores of all products and calculates the standardized scores. : ,in, Represents the standardized recommendation score, ranging from [0,1], They are the highest and lowest recommendation scores in the current candidate product set respectively; According to the standardized recommendation score Sort the products and generate a recommended ranking list of the products.

10. The intelligent shopping system with recommendation ranking function according to claim 9, characterized in that: The reinforcement learning is used to dynamically adjust The value of makes the recommendation results continuously optimized, which can be expressed as: ,in, are the current weight values ​​respectively; User feedback for and The score adjustment increment is Represents the learning rate parameter, which is used to control the magnitude of weight adjustment.

Images