Semantic Scholar

Search 200M+ papers and get one-sentence TLDRs

Some setup needed Web

research assistant #paper-search#citation-graph#research-discovery

About

Search across 200M+ papers and skim one-sentence TLDRs instead of long abstracts. Researchers use it to find related work, save citations to a Library, and follow AI-curated Research Feeds to stay current. Compared with keyword search, its semantic ranking and TLDRs (beta on ~60M papers) help shortlist relevant papers faster.

Editor's Take

Worth trying if you need to shortlist literature quickly or keep project-specific feeds for ongoing discovery; best suited for researchers who can organize papers into Library folders and tolerate partial TLDR coverage.

Key Features

Type a research query → get semantically ranked results across 200M+ publications
Open a paper with TLDRs → read a one-sentence summary to judge relevance quickly (beta on ~60M papers)
Follow Research Feeds on chosen topics → receive AI-recommended new papers similar to items in your Library
Save papers to your Library → track and export citations from organized folders

Use Cases

A PhD student preparing a literature review on reinforcement learning failure modes for a thesis chapter
A lab PI tracking weekly updates in multiple subfields via Research Feeds tied to project folders
A university librarian teaching students how to run semantic searches and manage citations for term papers

Try It Like This

1
Quickly shortlist papers for a lit review
Sign up and enter a focused semantic query (e.g., “failure modes of RLHF”) → scan semantically ranked results and read one-sentence TLDRs to reject irrelevant papers quickly → save chosen papers to a Library folder to build an exportable reading list.
2
Track new work in a niche subfield
Create a Library folder for the project and add representative seed papers → enable a Research Feed on that folder so the system recommends similar new papers → review the feed weekly and save immediately relevant items to the same folder.
3
Find methods and datasets used by a paper
Open a target paper from search results and use the citations/linked-papers view to see related work → inspect TLDRs or the Semantic Reader (when available) to surface mentions of datasets and methods → follow or save the most relevant methodological papers for deeper reading.
4
Prepare citations for a thesis chapter
Collect papers into a Library organized by chapter or subsection → use the export function to download citations (small batches if needed) → import exported citations into Zotero or another reference manager and paste summaries into your draft.
5
Discover cross-disciplinary connections
Run a broad semantic query that mixes terms from two fields (e.g., “robotics + causality”) → use the ranked results and citation graph to spot papers that bridge both areas → save cross-cutting papers to a project folder and set a Research Feed to capture future interdisciplinary work.

Pros & Cons

Pros

Searches are semantic and return results ranked by meaning rather than strict keyword matches, which helps find relevant papers without enumerating synonyms.
TLDR one-sentence summaries (beta on roughly 60M papers) let researchers judge relevance at a glance and shorten screening time.
Research Feeds let users follow topics tied to Library folders, surfacing AI-recommended new papers similar to items already saved.

Cons

TLDRs are still in beta and cover only part of the corpus (~60M papers), so many results lack the one-sentence summary.
Citation export requires saving items to a Library and is limited to small batches, which can complicate large-scale export workflows.
Some search edge cases reported (e.g., queries with the word “and”) can produce unexpected results, so query wording may need iteration.

Getting Started

1 Go to semanticscholar.org and sign in or create an account
2 Enter a topic or question and apply filters to narrow results
3 Open a result with a TLDR and add key papers to your Library to start a Research Feed