Knowledge-Graph Media Recommendation for Diverse Relevant Results
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing media recommendation systems often provide monotonous suggestions by recommending media similar to what has already been watched, lacking diversity and failing to consider broader contextual entities related to the media content.
Innovation Solution
A method involving extracting media and text representation vectors, performing knowledge retrieval in a knowledge graph to obtain an entity sub-graph, and fusing these vectors to generate a knowledge augmented vector for recommending target media data that aligns with user interests, incorporating entity information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If media recommendation is based on similar watched content, then recommendation accuracy is improved, but recommendation diversity deteriorates
Solution Approach 1:
The patent transitions from single-dimension content similarity to multi-dimensional recommendation by integrating knowledge graph entities (actors, directors, genres, themes) as additional recommendation dimensions. This allows the system to recommend media based on entity associations beyond mere content similarity, thereby improving diversity while maintaining accuracy.
Solution Approach 2:
The knowledge graph serves as an intermediary structure that connects media content with contextual entities. By retrieving entity sub-graphs and fusing entity representation vectors with media representation vectors, the system mediates between watched content and recommended content through entity relationships, enabling diverse yet accurate recommendations.
2Adaptability or versatility
If knowledge graph retrieval is performed, then recommendation diversity is improved, but system complexity increases
Solution Approach 1:
The patent segments the recommendation system into distinct modules: media representation extraction, knowledge graph retrieval, entity sub-graph extraction, entity representation vector fusion, and recommendation generation. This modular segmentation manages system complexity by organizing functions into independent, manageable components.
Solution Approach 2:
The knowledge graph structure serves multiple functions: storing media information, providing entity relationships, enabling contextual retrieval, and facilitating vector fusion. This multi-functionality reduces overall system complexity by using a single unified structure for multiple purposes rather than separate systems for each function.
3Measurement precision
If entity representation vectors are fused with media vectors, then recommendation relevance is improved, but computational requirements increase
Solution Approach 1:
The patent retrieves entity sub-graphs (partial knowledge graph) rather than the entire knowledge graph, and fuses only relevant entity representation vectors with media vectors. This partial action approach improves recommendation relevance through selective vector fusion while reducing computational requirements by processing only necessary subsets of data.
Data Source
AI summary
In a method of media data recommendation, a media representation vector is extracted from media data and a text representation vector is extracted from a description text of the media data. A knowledge retrieval is performed in a knowledge graph according to the media representation vector, to obtain an entity sub-graph of the media data. An entity representation vector of the entity sub-graph is determined. A feature fusion processing is performed on the media representation vector, the text representation vector, and the entity representation vector, to obtain a knowledge augmented vector. Target media data is obtained based on the knowledge augmented vector that is a fused vector of the media representation vector, the text representation vector, and the entity representation vector. The target media data is recommended to a target object. Apparatus and non-transitory computer-readable storage medium counterpart embodiments are also contemplated.