Modeling preference dependencies improves narrative driven recommendations
Modeling Preference Dependencies with Directed Acyclic Graphs for Narrative-Driven Recommendation
Information RetrievalComputation and Language
Summary
People often share their needs through stories or long descriptions that include many linked preferences. The authors found that existing recommendation systems miss important details because they treat user requests like simple lists of keywords. They created a new method that breaks down these complex preferences into a graph showing how they depend on each other. This helps the system find better information from the web and make more accurate recommendations. Their tests in multiple languages and areas showed this approach works better than many other methods.
What this means in practice
- •For software engineers: Build recommendation features that interpret complex user preferences expressed in stories for more accurate suggestions in apps and websites.
- •For product managers: Design user-driven recommendation systems that improve content or product suggestions by modeling how user preferences depend on each other.
Authors
Mingzhe Lu, Jian Qin, Ruize Qin, Jinhao Liu, Yunpeng Li, Hang Su
Abstract
Narrative-driven recommendation allows users to express multifaceted needs through free-form text where stated preferences naturally qualify and condition one another. Existing methods typically treat narratives as flat keyword combinations or unconstrained generative queries, failing to capture how complex preferences hierarchically refine and contextualize each other. We propose the Preference Dependency Graph Framework (PDG-F), a narrative-driven recommendation framework centered on explicit preference dependencies. Specifically, PDG-A (PDG-Agent) first parses unstructured requests to organize user preferences and their mutual qualifications into a directed acyclic Preference Dependency Graph (PDG). Through this graph, candidate evidence from the open Web is acquired via directed path retrieval, and final recommendations are determined by topological path ranking. Comprehensive experiments across three diverse domains spanning English and Chinese contexts confirm that PDG-F consistently achieves superior recommendation performance over competitive baselines, open-domain retrieval agents, and frontier language models. Further diagnostic analyses show that our dependency-guided retrieval markedly raises the attainable candidate-pool ceiling, demonstrating the fundamental value of modeling preference dependencies in narrative-driven recommendation.