Abstract
This paper describes a technique for proactively suggesting contextually relevant collaborative opportunities within a short-form video ecosystem. The system employs a graph-based architecture to model the video environment as a massive, interconnected network where individual videos are represented as nodes. Relationships between these nodes are established through multi-modal edges that represent visual content similarity, audio characteristics, and user engagement patterns. By utilizing a graph neural network (GNN) to aggregate features from a video’s neighborhood, the system generates rich, context-aware embeddings. These embeddings enable real-time identification of stylistically and contextually compatible content through efficient search mechanisms. When paired with a large language model (LLM), the technique allows for conversational queries and intelligent matchmaking to foster a dynamic and collaborative creator community.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Shin, D, "Context-Aware Short-Form Video Collaboration via Graph Neural Networks and Large Language Models", Technical Disclosure Commons, (February 12, 2026)
https://www.tdcommons.org/dpubs_series/9327