@yrzhe_top: Tried Claude Science, it feels great for early-stage research. It breaks down a research question into multiple retrieval facets, connects to literature databases like OpenAlex and arXiv, fetches real literature data, then automatically deduplicates, filters, clusters, and visualizes, finally generating a report with topic maps, growth trends, and representative papers.

Tried Claude Science, it feels great for early-stage research. It breaks down a research question into multiple retrieval facets, connects to literature databases like OpenAlex and arXiv, fetches real literature data, then automatically deduplicates, filters, clusters, and visualizes, finally generating a research report with topic maps, growth trends, and representative papers. I asked it to analyze human-AI collaboration. It first fetched 2,301 papers, cleaned them down to 1,711, automatically divided them into 10 topic communities, and produced a complete research map. For someone new to a field who wants to quickly know "what are the directions, who is working on them, which papers are important, and what are the trends," this is very helpful. Of course, it cannot replace a rigorous systematic review, but as a tool for preliminary research and topic exploration, it is highly efficient.