Personalized explanations for hybrid recommender systems
summary
System extends HyPER and uses Probabilistic Soft Logic to translate grounded logical rules into real-time hybrid explanations for Last.fm items; crawlers take ~10 minutes and inference ~1 minute. A crowdsourced study with 198 filtered users finds item-based explanations more persuasive than social or popularity cues, users prefer about three to four explanation styles (mean 4.3 after filtering, 40% ranked none), textual formats beat visuals, personality traits significantly correlate with persuasiveness, and accuracy is good but novelty remains low.
AI-written summary · about
chapters (7)
- 0:00Start
- 0:47The Research Question
- 2:10Challenges
- 2:45Which Explanation Styles Are Preferred by Users
- 7:13Example of a Hybrid Explanation
- 13:16User Personality Traits
- 15:30The Job Recommendation Domain
related talks
09mVisual, textual or hybrid: the effect of user expertise on different explanations
10mCritiquing for Music Exploration in Conversational Recommender Systems
16mMaking recommendations explainable
09mAre Explanations Helpful? A Comparative Study of the Effects of Explanations in AI-Assisted ...
22mAgentic Hybrid Retrieval: Orchestrating LLM Agents, Hybrid Search & Explainability in OpenSearch
33m"Are You Sure?": blending product comparisons and recommendations with A.I.
29mBuilding Generative Content Understanding Systems
01h 14mBuilding a Recommendation System with a Hybrid Search From Scratch