Commodity recommendation system and method based on artificial intelligence

Abstract

The invention discloses a commodity recommendation system based on artificial intelligence, and the system comprises a user behavior tracking module which is used for collecting and processing the multi-source behavior data of a user; the commodity feature extraction and fusion module is used for acquiring and analyzing multi-modal information of commodities; the dynamic updating module is used for adjusting a recommendation list in real time according to instant user behaviors and long-term preferences in combination with short-term memory and long-term memory; the situation awareness module takes external factors such as time, place and weather as input; the explanatory enhancement module is used for adding an explanatory layer for a black box model by using an explanatory machine learning model; the personalized sorting optimization module is used for dynamically adjusting the feature weight and improving the click rate and the conversion rate; and the privacy protection module is used for protecting user privacy by adopting a differential privacy technology. According to the commodity recommendation system and method provided by the invention, many problems existing in a traditional recommendation system are overcome, and a more efficient, more intelligent and safer commodity recommendation service is realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent business technology, and in particular to an artificial intelligence-based commodity recommendation system and method. Background Art

[0002] With the rapid development of the Internet and e-commerce, product recommendation systems have become an indispensable part of online shopping platforms. Traditional recommendation systems mainly rely on collaborative filtering, content-based filtering and other methods, which predict users' interests by analyzing their historical behaviors (such as purchase records, browsing history) or product features (such as categories, tags) and recommend products that may be of interest to them. However, traditional recommendation systems have several significant defects.

[0003] First, the cold start problem is a common challenge. For new users or new products, traditional recommendation systems find it difficult to provide effective personalized recommendations due to the lack of sufficient historical data. Second, the data sparsity problem limits the performance of the recommendation system, especially when the user base is large but the interaction data is limited, resulting in insufficient accuracy and relevance of the recommendation results. In addition, traditional recommendation systems often lack sensitivity to users' real-time behavior and changes in contextual environment, and cannot meet users' dynamic needs well, thus affecting the user experience. At the same time, personalized sorting is not accurate enough, and the recommendation list fails to be dynamically adjusted according to the unique situation of each user, affecting the click-through rate and conversion rate; privacy protection measures are not perfect, and the risk of personal information leakage still exists during data collection, storage and use, which has caused users to worry about privacy security. Summary of the invention

[0004] 1. Technical Problems Solved

[0005] The technical problems to be solved by the present invention are the various problems mentioned in the above background technology, and a product recommendation system and method based on artificial intelligence is provided.

[0006] 2. Technical Solution

[0007] In order to solve the above technical problems, the present invention provides a technical solution: a commodity recommendation system based on artificial intelligence, comprising:

[0008] A user behavior tracking module, which is used to collect and process multi-source behavior data of users, the data at least including purchase records, browsing history, evaluation content and social network interaction;

[0009] The product feature extraction and fusion module is used to obtain and analyze the multimodal information of the product, which includes at least text descriptions, pictures, and videos, and perform feature extraction and fusion through a deep neural network;

[0010] Dynamic update module, combining short-term memory and long-term memory, to adjust the recommendation list in real time based on immediate user behavior and long-term preferences;

[0011] The context-aware module takes external factors such as time, location, and weather as input to improve the relevance and timeliness of recommendations;

[0012] Explanation enhancement module, which uses explainable machine learning models or adds an explanation layer to black box models to increase users’ trust in the recommendation system;

[0013] Personalized sorting optimization module generates personalized sorting functions for each user, dynamically adjusts feature weights, and improves click-through rate and conversion rate;

[0014] The privacy protection module uses differential privacy technology to protect user privacy and ensure that the risk of personal information leakage is strictly controlled during data collection, storage, and use.

[0015] As an improvement, a data preprocessing module is also included to clean and standardize data from different sources to ensure data consistency and accuracy.

[0016] As an improvement, a feedback collection module is also included to receive user feedback on the recommendation results and use the feedback to improve the recommendation algorithm.

[0017] As an improvement, a method for a commodity recommendation system based on artificial intelligence is characterized by comprising the following steps:

[0018] (a) Collect and process multi-source behavioral data of users, which at least includes the user's purchase records, browsing history, evaluation content, and social network interactions;

[0019] (b) obtaining and analyzing multimodal information of the product, wherein the information includes at least a text description, a picture, and a video of the product;

[0020] (c) using a deep neural network to extract and fuse features of the user behavior data collected in step (a) and the product information obtained in step (b) to construct a user profile and product representation;

[0021] (d) Based on immediate user behavior and long-term preferences, a dynamic update mechanism combining short-term and long-term memory is used to adjust the recommendation list in real time;

[0022] (e) Incorporate external factors such as time, location, and weather into the recommendation model to achieve context-aware recommendations;

[0023] (f) Use interpretable machine learning models or add an explanation layer to black box models to enhance the transparency of recommendation results;

[0024] (g) Generate a personalized ranking function for each user, dynamically adjust the importance weight of each feature according to the user's specific situation, and achieve accurate ranking;

[0025] (h) While ensuring the recommendation effect, differential privacy technology is used to protect user privacy and prevent personal information leakage.

[0026] As an improvement, the deep neural network includes but is not limited to convolutional neural networks, recurrent neural networks and their variants such as long short-term memory networks and gated recurrent units.

[0027] As an improvement, the context-aware recommendation also includes: providing localized recommendations based on the user's geographic location; adjusting recommended content based on current time or seasonal activities; and recommending related products based on weather conditions.

[0028] As an improvement, the dynamic update mechanism combining short-term memory and long-term memory further includes: a fast response component for capturing instant changes in user behavior, and a stable memory component for maintaining user historical preferences.

[0029] 3. Beneficial Effects

[0030] The advantages of the present invention compared with the prior art are:

[0031] 1. The system product recommendation system of the present invention collects and processes multi-source behavioral data such as users' purchase records, browsing histories, evaluation content, and social network interactions, and simultaneously obtains multi-modal information such as text descriptions, pictures, and videos of products, and uses deep neural networks for feature extraction and fusion to construct a more comprehensive and accurate user portrait and product representation, significantly improving the accuracy and relevance of recommendations. It introduces a dynamic update mechanism that combines short-term memory with long-term memory, which can quickly adjust the recommendation list according to the user's immediate behavior, and maintains the consistency and coherence of the recommendation in combination with the user's historical preferences, ensuring that the recommendation results are always close to the user's current interests and needs.

[0032] 2. The system product recommendation system of the present invention incorporates external factors such as time, place, and weather into the recommendation model, which not only takes into account the user's behavior patterns, but also combines changes in the external environment, making the recommendations more timely and targeted, improving user experience and satisfaction. At the same time, it adopts an explainable machine learning model or adds an explanation layer to the black box model, which can give specific reasons why a certain product is recommended to a specific user, increasing the transparency of the system and the user's trust.

[0033] 3. The product recommendation system of the present invention generates a personalized ranking function for each user, dynamically adjusts the importance weight of each feature according to the user's specific situation, achieves accurate sorting, effectively improves click-through rate and conversion rate, and enhances commercial value. On the premise of ensuring the recommendation effect, the system adopts differential privacy technology to strictly control the risk of personal information leakage, fully respect the user's privacy, and ensure the security and compliance of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a structural diagram of a commodity recommendation system based on artificial intelligence in the present invention.

[0035] Figure 2 The present invention is a recommendation flow chart of a commodity recommendation system based on artificial intelligence. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0037] Embodiment 1

[0038] like Figure 1 As shown, a commodity recommendation system based on artificial intelligence, its specific system architecture includes:

[0039] User behavior tracking module: This module is responsible for collecting and processing multi-source user behavior data, including but not limited to purchase records, browsing history, evaluation content and social network interactions. It is integrated into the e-commerce platform through API interfaces or SDKs to capture various user interactions on the platform in real time and store these data in a distributed database for subsequent analysis and processing.

[0040] Product feature extraction and fusion module: This module is used to obtain and analyze multimodal information of products, such as text descriptions, pictures, videos, etc. It uses advanced deep neural networks (such as convolutional neural networks (CNN) for image processing, recurrent neural networks (RNN) and its variants (LSTM / GRU) for text analysis) to extract features from different forms of data, and fuse features through the Transformer model with self-attention mechanism or other appropriate algorithms to construct a comprehensive product representation. This step can not only capture the basic attributes of the product, but also explore potential correlations and patterns, providing a rich information basis for subsequent recommendations.

[0041] Dynamic update module: Combining short-term memory and long-term memory, the recommendation list is adjusted in real time according to the user's immediate behavior (short-term memory) and historical preferences (long-term memory). The short-term memory part uses a fast response component to quickly capture the user's latest interest changes; while the long-term memory part uses a stable memory component to ensure the consistency and coherence of the recommendation. This mechanism enables the recommendation system to adapt to the user's short-term needs while maintaining an understanding of long-term preferences, improving the relevance and accuracy of the recommendation.

[0042] Context-aware module: External factors such as time, location, and weather are taken as inputs and incorporated into the recommendation model. For example, localized recommendations are provided based on the user's geographic location, the recommended content is adjusted based on the current time or seasonal activities, or suitable products are recommended based on weather conditions. In this way, the recommendation results are more in line with the user's actual life scenarios, enhancing the timeliness and pertinence of the recommendations.

[0043] Explanation Enhancement Module: In order to increase users' trust in the recommendation system, this module uses interpretable machine learning models (such as decision trees, random forests, GBDT, linear regression) or adds explanation layers to black box models (such as LIME, SHAP). These technologies can help users understand why a certain product is recommended to them, thereby improving users' recognition and acceptance of the recommendation results.

[0044] Personalized sorting optimization module: Generates a personalized sorting function for each user, and dynamically adjusts the importance weight of each feature according to the user's specific situation. The sorting function takes into account factors such as the user's historical interaction frequency, the time interval between the last interaction, and the user's interest in different categories of goods to achieve accurate sorting. Through this personalized approach, the system can significantly improve click-through rate and conversion rate, and increase commercial value.

[0045] Privacy protection module: Under the premise of ensuring the recommendation effect, differential privacy technology (such as Laplace mechanism, exponential mechanism, Gaussian mechanism) is used to protect user privacy, ensuring that the risk of personal information leakage is strictly controlled during data collection, storage, and use. At the same time, the system also supports data anonymization and encrypted transmission, further ensuring the security of user data.

[0046] Data preprocessing module (see claim 2): This module is responsible for cleaning and standardizing data from different sources to ensure data consistency and accuracy. By performing operations such as deduplication, filling missing values, and format conversion on the original data, the preprocessing module provides a high-quality data foundation for subsequent data analysis and model training.

[0047] Feedback collection module (see claim 3): This module is used to receive user feedback on recommendation results and use this feedback to improve the recommendation algorithm. Users can express their opinions on recommended products through ratings, comments, likes, etc. The system will regularly analyze this feedback data, identify common problems and improvement suggestions, and then optimize the recommendation algorithm to form a closed-loop learning process.

[0048] Embodiment 2

[0049] like Figure 2 As shown in the figure, a commodity recommendation system method based on artificial intelligence has the following process:

[0050] Step (a): The user behavior tracking module collects the user's multi-source behavior data in real time, including purchase records, browsing history, evaluation content and social network interactions.

[0051] Step (b): The product feature extraction and fusion module obtains the product's multimodal information such as text description, pictures, videos, etc., and performs feature extraction and fusion through a deep neural network.

[0052] Step (c): The user behavior data collected in step (a) and the product information obtained in step (b) are fed into a deep neural network to construct a user profile and product representation.

[0053] Step (d): The dynamic update module uses a mechanism that combines short-term memory and long-term memory to adjust the recommendation list in real time based on the user's immediate behavior and long-term preferences.

[0054] Step (e): The context-aware module incorporates external factors such as time, location, and weather into the recommendation model to achieve more accurate context-aware recommendations.

[0055] Step (f): The explainability enhancement module uses an explainable machine learning model or adds an explanation layer to the black box model to enhance the transparency of the recommendation results so that users can understand the reasons for the recommendation.

[0056] Step (g): The personalized sorting optimization module generates a personalized sorting function for each user, dynamically adjusts the importance weight of each feature according to the user's specific situation, and achieves accurate sorting.

[0057] Step (h): The privacy protection module adopts differential privacy technology throughout the whole process to ensure the security and privacy protection of user data and prevent personal information leakage.

[0058] Through the above-mentioned specific implementation methods, the product recommendation system and method provided by the present invention overcome the cold start problem, data sparsity problem, insufficient diversity of recommendation results and other problems existing in traditional recommendation systems, and realize a more efficient, smarter and safer product recommendation service, which greatly meets the personalized needs of users and promotes the development of e-commerce platforms and the improvement of user experience.

[0059] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0060] Although 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 appended claims and their equivalents.

[0061] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A commodity recommendation system based on artificial intelligence, characterized in that: include: A user behavior tracking module, which is used to collect and process multi-source behavior data of users, the data at least including purchase records, browsing history, evaluation content and social network interaction; The product feature extraction and fusion module is used to obtain and analyze the multimodal information of the product, which includes at least text descriptions, pictures, and videos, and perform feature extraction and fusion through a deep neural network; Dynamic update module, combining short-term memory and long-term memory, to adjust the recommendation list in real time based on immediate user behavior and long-term preferences; The context-aware module takes external factors such as time, location, and weather as input to improve the relevance and timeliness of recommendations; Explanation enhancement module, which uses explainable machine learning models or adds an explanation layer to black box models to increase users’ trust in the recommendation system; Personalized sorting optimization module generates personalized sorting functions for each user, dynamically adjusts feature weights, and improves click-through rate and conversion rate; The privacy protection module uses differential privacy technology to protect user privacy and ensure that the risk of personal information leakage is strictly controlled during data collection, storage, and use.

2. The product recommendation system based on artificial intelligence according to claim 1, characterized in that: It also includes a data preprocessing module to clean and standardize data from different sources to ensure data consistency and accuracy.

3. The product recommendation system based on artificial intelligence according to claim 1, characterized in that: It also includes a feedback collection module for receiving user feedback on recommendation results and using the feedback to improve the recommendation algorithm.

4. The method of a commodity recommendation system based on artificial intelligence according to claim 1, characterized in that: The following steps are involved: (a) Collect and process multi-source behavioral data of users, which at least includes the user's purchase records, browsing history, evaluation content, and social network interactions; (b) obtaining and analyzing multimodal information of the product, wherein the information includes at least a text description, a picture, and a video of the product; (c) using a deep neural network to extract and fuse features of the user behavior data collected in step (a) and the product information obtained in step (b) to construct a user profile and product representation; (d) Based on immediate user behavior and long-term preferences, a dynamic update mechanism combining short-term and long-term memory is used to adjust the recommendation list in real time; (e) Incorporate external factors such as time, location, and weather into the recommendation model to achieve context-aware recommendations; (f) Use interpretable machine learning models or add an explanation layer to black box models to enhance the transparency of recommendation results; (g) Generate a personalized ranking function for each user, dynamically adjust the importance weight of each feature according to the user's specific situation, and achieve accurate ranking; (h) While ensuring the recommendation effect, differential privacy technology is used to protect user privacy and prevent personal information leakage.

5. The method of a commodity recommendation system based on artificial intelligence according to claim 4, characterized in that: The deep neural network includes but is not limited to convolutional neural networks, recurrent neural networks and their variants such as long short-term memory networks and gated recurrent units.

6. The method of a commodity recommendation system based on artificial intelligence according to claim 4, characterized in that: The context-aware recommendations also include: providing localized recommendations based on the user's geographic location; adjusting recommendations based on current time or seasonal activities; and recommending related products based on weather conditions.

7. The method of a commodity recommendation system based on artificial intelligence according to claim 4, characterized in that: The dynamic update mechanism combining short-term memory and long-term memory further includes: a fast response component for capturing instant user behavior changes, and a stable memory component for maintaining user historical preferences.

Images