SIRUP was an interactive playground for exploring search-based book recommendation methods that leverage readers’ written reviews. From large amounts of review text, the system built concise textual user profiles and used them to retrieve relevant books under different situative contexts.
The demo let visitors try multiple profile-construction methods and see recommendations when focusing on:
Through these interactions, SIRUP illustrated how natural-language signals from reader communities can drive personalized, context-aware book discovery across large datasets.