Personalized Advertisement Push Using User Interest Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing advertisement push methods on video platforms have a low association degree with content and fail to meet personalized user interests, resulting in poor advertising effectiveness.
Innovation Solution
A method and system for personalized advertisement push based on user interest learning, utilizing multitask sorting learning to obtain user interest models, extracting objects of interest in videos, and retrieving related advertising information from a database based on visual features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If predefined advertisements are pushed using traditional methods (time domain insertion, peripheral display, or partial overlap), then the advertisement delivery is simple and coverage is wide, but the association degree with video content is low and personalized user interests are not met
Solution Approach 1:
The system performs preliminary actions by pre-learning and establishing user interest models before advertisement delivery. The multitask sorting learning algorithm pre-processes user behavior data to create personalized interest profiles, which are then used to guide advertisement selection and positioning, ensuring high association with both user interests and video content before the actual advertisement push occurs.
Solution Approach 2:
The system implements dynamics by making the advertisement push process adaptive and flexible. The advertisement positioning is dynamically adjusted based on real-time analysis of video content and user interest models. The system can dynamically select from multiple push methods (time domain, peripheral, or overlapped) and dynamically position advertisements within video frames based on detected objects of interest, rather than using fixed predetermined positions.
2Ease of operation
If advertisements are displayed on the periphery of the video player, then user viewing experience is less disturbed, but the advertisement is often ignored as background and effectiveness is reduced
Solution Approach 1:
The system applies local quality by differentiating the treatment of different spatial regions within the video frame. Instead of uniformly displaying advertisements on the periphery, the system identifies specific local regions containing objects of interest that match user interests and places advertisements in those targeted locations. This allows the advertisement to be positioned in semantically meaningful local areas rather than generic peripheral zones, making it more noticeable while maintaining contextual relevance.
Solution Approach 2:
The system uses objects of interest in the video as intermediaries between the video content and the advertisement. By detecting and analyzing objects within the video frame, the system creates a semantic bridge that connects the advertisement to relevant content elements. The advertisement is positioned relative to these intermediary objects, which serve as focal points that naturally draw user attention and create meaningful associations between the video content and the advertisement.
3Reliability
If a small advertisement is overlapped on a part of the video content, then the advertisement is more visible, but the normal viewing experience of the user is affected to a certain extent
Solution Approach 1:
The system applies local quality by selectively overlapping advertisements only on specific local regions of the video frame that contain objects of interest, rather than uniformly overlapping across the entire video or non-critical areas. This targeted approach ensures that advertisements are placed in semantically relevant locations that naturally attract user attention, thereby improving visibility without unnecessarily disturbing the viewing experience in other important regions of the video.
4Adaptability or versatility
If multiple user interest models are obtained through multitask sorting learning and visual features are extracted for personalized advertisement retrieval, then personalized advertisement push is achieved, but the computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary action by pre-learning and establishing user interest models using multitask sorting learning algorithms before actual advertisement delivery. This offline pre-processing stage creates reusable user profiles that capture individual preferences and interests. During runtime, the system only needs to perform lightweight matching between these pre-established models and current video content, significantly reducing the computational burden and processing time during actual advertisement push operations.
Solution Approach 2:
The system applies segmentation by dividing the complex advertisement push process into distinct modular stages: user interest model learning, video content analysis, object of interest detection, and advertisement retrieval. Each stage operates independently and can be optimized separately. The visual feature extraction is segmented into specific feature types (color, texture, shape) that can be processed in parallel, reducing overall processing time while maintaining comprehensive analysis for personalized advertisement selection.
Data Source
AI summary
Embodiments of the present invention relate to a method and a system for personalized advertisement push based on user interest learning. The method may include: obtaining multiple user interest models through multitask sorting learning; extracting an object of interest in a video according to the user interest models; and extracting multiple visual features of the object of interest, and according to the visual features, retrieving related advertising information in an advertisement database. Through the method and the system provided in embodiments of the present invention, a push advertisement may be closely relevant to the content of the video, thereby meeting personalized requirements of a user to a certain extent and achieving personalized advertisement push.