Media Recommendation Content Generation From Interactive Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating recommended content for media content is difficult and time-consuming, often resulting in poor quality due to reliance on user-generated text and images.
Innovation Solution
A method and apparatus that automatically generate recommended content based on basic information and interactive content associated with media content, including a recommended image and text, using pre-constructed image and text generation models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If user manually writes recommended text and determines recommended picture, then recommended content can be generated, but the process is difficult and time-consuming with poor quality
Solution Approach 1:
The system automatically generates recommended text and pictures by extracting information from the media content itself and using pre-trained generation models, eliminating the need for manual user input. The media content serves its own recommendation needs through automated processing of its basic information and interactive content.
Solution Approach 2:
The manual mechanical process of users writing text and selecting pictures is replaced by automated computational systems including information extraction modules and AI generation models, transforming a manual creative process into an automated computational one.
2Reliability
If user manually creates recommended content, then some recommendation functionality is achieved, but the quality is poor
Solution Approach 1:
Pre-trained generation models act as intermediaries between the extracted media information and the final recommended content output. These models translate structured information into high-quality natural language text and images, ensuring consistent and reliable content generation quality.
Solution Approach 2:
The system changes the parameters of content generation by using AI models with adjustable parameters to optimize text and image generation quality. The generation models process information with optimized parameters to produce high-quality recommended content that highlights media characteristics.
3Productivity
If automated generation models are used, then content generation difficulty is reduced and quality is improved, but system complexity increases
Solution Approach 1:
Generation models are pre-trained in advance on large datasets, so that during actual content generation, only inference is needed rather than training. This preliminary preparation reduces the complexity of the operational system while maintaining high generation capabilities.
Solution Approach 2:
The content generation system is divided into separate functional modules: information extraction module, text generation module, and picture generation module. Each module handles a specific task independently, reducing overall system complexity while improving productivity through specialized processing.
Data Source
AI summary
Embodiments of the present disclosure provide a recommended content generation method, an apparatus, a device, a readable storage medium, and a product. The method includes: determining at least one piece of media content to be recommended; obtaining, for each piece of media content, basic information corresponding to the piece of media content, and obtaining at least one type of interactive content generated based on the piece of media content; generating, based on the basic information corresponding to each piece of media content and the at least one type of interactive content, recommended content corresponding to the piece of media content, wherein the recommended content includes a recommended image and a recommended text; and generating target recommended content based on the recommended content corresponding to each piece of media content, and posting the target recommended content.