PaperFlow: Profiling, Recommending, and Adapting Across Daily Paper Streams

arXiv·medium signal

PaperFlow reframes scientific paper recommendation from static ranking over a fixed candidate set to a dynamic daily-stream problem that profiles a reader, recommends, and adapts over time. For practitioners drowning in arXiv volume, it models reading as it actually happens — an evolving daily feed rather than a one-shot ranked list — making it directly relevant to anyone building research-monitoring or agentic literature workflows.