Reproduce “MG-ShopDial: A Multi-Goal Conversational Dataset for e-Commerce” 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
Nolwenn Bernard, Krisztian Balog
MG-ShopDial offers a human-human conversational dataset for e-commerce, comprising 64 conversations with 2,196 annotated utterances that naturally blend search, recommendation, and question answering goals. This resource supports the development of conversational AI capable of managing complex, evolving user information needs during product discovery.
Key insights
- Multi-goal conversations consistently exhibit a phased structure, typically beginning with recommendation, followed by information seeking (search/QA), and sometimes a secondary recommendation phase.
- Meta-communication plays a continuous role throughout conversations, facilitating fluidity and structure rather than specific goal completion.
- The intents associated with QA and Search goals show significant overlap, suggesting a potential blurriness between these goals that requires further refinement in conversational AI modeling.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.