Content Recommendation System Using Explicit and Implicit Feedback Interaction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content recommendation systems have low accuracy due to limited consideration of user behavior beyond click and non-click interactions, resulting in a deviation between analyzed user preferences and true interests, leading to suboptimal content delivery.
Innovation Solution
A content recommendation method that integrates explicit and implicit feedback, including sharing and grading behaviors, using a deep feedback interaction component, wide component, and DNN component to capture user preferences, and considers time information for historical behavior sequences, thereby improving the accuracy of predicted click-through rates (CTR).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only click and non-click interactions are considered for content recommendation, then the system complexity remains low, but the recommendation accuracy deteriorates due to limited user behavior consideration
Solution Approach 1:
The patent segments user feedback into distinct types (explicit feedback like sharing and grading, implicit feedback like clicking) and processes each through specialized components (wide component for explicit feedback, DNN component for implicit feedback). This segmentation allows the system to handle multiple feedback types without overwhelming complexity, while still achieving high recommendation accuracy through the feature interaction component that combines these segmented features.
Solution Approach 2:
The patent adds a new dimension to user behavior analysis by incorporating explicit feedback dimensions (sharing, grading) beyond the traditional click dimension. This is achieved through the wide component that processes explicit feedback features and the feature interaction component that multiplies explicit and implicit feedback features, creating a multi-dimensional user preference representation that improves accuracy without proportionally increasing complexity.
2Measurement precision
If user feedback is limited to basic click behavior, then the data processing complexity remains low, but the user preference analysis accuracy deteriorates
Solution Approach 1:
The patent introduces a feature interaction component as an intermediary that bridges explicit feedback features (from wide component) and implicit feedback features (from DNN component). This intermediary performs feature multiplication to capture the interaction effects between different feedback types, enabling accurate user preference analysis without requiring direct complex processing of all raw feedback data, thus balancing accuracy and processing complexity.
3Measurement precision
If historical behavior sequences ignore time information, then the computational requirements remain low, but the accuracy of capturing current user interests deteriorates
Solution Approach 1:
The patent performs preliminary action by pre-processing historical behavior sequences to extract time-related features and user preference patterns before the actual recommendation task. The DNN component processes historical implicit feedback sequences to capture temporal patterns, and these pre-extracted features are then fed into the feature interaction component. This preliminary processing of time information reduces the computational burden during real-time recommendation while maintaining high accuracy in capturing current user interests.
Data Source
AI summary
This application provides a content recommendation method and apparatus, an electronic device, and a storage medium. The content recommendation method includes obtaining content feedback information of a target object and content feature information of content that is to be recommended in response to a content recommendation request of the target object, the content feedback information comprising explicit feedback information and implicit feedback information and object portrait information of the target object; performing feature interaction according to the explicit feedback information and the implicit feedback information in the content feedback information, and obtaining behavior preference information; performing feature extraction based on the behavior preference information, the content feedback information, and the content feature information, and obtaining a predicted click-through rate (CTR); and determining, according to the predicted CTR, recommended content from the pieces of content that is to be recommended, and transmitting the recommended content to a terminal device.