Sci-Surf: Navigating Scientific Literature Discovery through Human Feedback and Intelligent Summarizatio

2026-08-12

AI Summary

This work addresses the limitations of existing academic platforms in accurately capturing fine-grained user intent and supporting deep paper comprehension. The authors propose an intent-driven scientific knowledge discovery system that introduces an interpretable user profiling mechanism and leverages large language models to deliver personalized recommendations and generate multimodal, structured summaries—integrating both textual content and figures in a blog-style format. The system continuously refines its intent modeling and information fusion through human feedback. Online user evaluations demonstrate that, within one month, the approach improves alignment between recommended results and users’ true preferences by an average of 10.4%, significantly enhancing both literature discovery and in-depth understanding.

Abstract

The rapid growth of scientific publications makes it increasingly difficult for researchers to identify relevant new studies and effectively comprehend them. Existing academic discovery platforms typically rely on static topic subscriptions or embedding-based similarity and provide only abstracts or short summaries, offering limited support for nuanced intent modeling and in-depth paper summarization. We present Sci-Surf, an intent-centric knowledge discovery system that integrates feedback-driven personalized recommendation with multi-modal blog-style paper digestion. Our approach refines user intent representations through LLM-based user profiling, while generating structured summaries that synthesize textual and visual information from full papers. The demo presents an end-to-end academic discovery pipeline and demonstrates measurable improvements in both recommendation quality and digestion quality through real-user evaluations. Specifically, the integration of verbalized profiles led to a 10.4% average improvement in predictive alignment with real-world user preferences throughout a month-long online evaluation.

Problem

• scientific literature discovery
• user intent modeling
• paper summarization
• academic recommendation
• information overload

Innovation

• intent-centric discovery
• feedback-driven recommendation
• multi-modal summarization
• LLM-based user profiling
• scientific literature digestion