Reproduce “FolkScope: Intention Knowledge Graph Construction for E-commerce Commonsense Discovery” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Changlong Yu, Weiqi Wang, Xin Liu, Jiaxin Bai, Yangqiu Song, Zheng Li, Yifan Gao, Tianyu Cao and 1 more
FolkScope introduces an intention knowledge graph for e-commerce, built via a human-in-the-loop framework that integrates large language model generation with human validation. This structured asset captures implicit commonsense about user purchasing motivations, demonstrably enhancing recommendation system performance, such as achieving an RMSE of 1.081 for Clothing compared to 1.104 for the Wide&Deep baseline.
Key insights
- Large Language Models can effectively generate implicit, domain-specific commonsense knowledge regarding user intentions, going beyond explicit product descriptions.
- Rigorous human evaluation and modeling of both plausibility and typicality are essential for ensuring the quality and causal relevance of generated intentional knowledge, as these two dimensions are distinct.
- A hybrid approach combining LLM generation with human-in-the-loop validation provides a scalable yet high-quality method for constructing complex, implicit knowledge graphs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.