# Semantic Scholar MCP > Stanford Semantic Scholar provides AI-powered access to the world's largest academic knowledge graph. Use this MCP to search millions of papers, track citation chains, analyze author impact metrics like the h-index, and discover foundational research related to your topic. ## Overview - **Category:** education - **Price:** Free - **Tags:** semantic-scholar, academic-papers, citations, research, literature-review, bibliography, ai-recommendations ## Description Need deep context on a scientific or technical subject? This MCP connects you directly to Semantic Scholar’s massive academic database. It lets your agent go beyond simple keyword searches to understand the actual *context* of published work. You can trace how an idea evolved by finding every paper that cited a key source, or conversely, see what foundational papers influenced a modern breakthrough. Need to review dozens of authors? Use this MCP to quickly pull author metrics, seeing their total citation count and h-index without leaving your agent client. When you connect this through Vinkius, your AI can handle the entire literature review process—from identifying key seminal works to building out bibliometric reports on demand. It’s how you get deep academic insight into your workflow. ## Tools ### batch_get_authors Retrieves multiple author profiles, providing their names, affiliations, paper counts, citation counts, and h-indices at once. ### batch_get_papers Accepts lists of IDs (DOIs, ArXiv, PubMed) to retrieve full metadata for multiple papers in a single call. ### bulk_search_papers Searches for very large result sets of academic papers and returns continuation tokens so you can process all results systematically. ### get_author Pulls a definitive profile for one author, detailing their affiliations, total paper count, citation count, and h-index. ### get_author_papers Retrieves every paper by a specific author, listing titles, years, venues, and whether the work is open access. ### get_multi_recommendations Generates focused literature suggestions by finding papers similar to a set of positive sources but unlike a set of negative ones. ### get_paper Fetches all details for a single paper using multiple identifiers, including DOI, ArXiv ID, or PubMed ID. ### get_paper_authors Identifies the contributing authors of a specific article and provides their individual metrics like h-index. ### get_paper_citations Finds all follow-up work by listing metadata for papers that cite a given source, showing how an idea was used later. ### get_paper_references Determines the intellectual roots of a paper by listing the foundational works it cited when it was written. ### get_recommendations Uses content similarity and citation patterns to suggest the most relevant papers you should read next, based on one seed article. ### match_paper_title Finds the correct paper metadata when you only have a slightly misspelled or generalized title string. ### search_authors Searches across the academic graph to locate researchers by name, providing their full profiles and metrics. ### search_by_field Filters available papers to only include those that fall within a specific discipline like Medicine or Computer Science. ### search_by_venue Narrows down the search results to publications from specific, high-impact conferences like Nature or NeurIPS. ### search_papers Performs a broad keyword search across 200 million papers, allowing filtering by year, field, and journal. ## Prompt Examples **Prompt:** ``` Find the most cited papers on transformer architectures published since 2020 ``` **Response:** ``` I've searched Semantic Scholar for "transformer architecture" papers from 2020-2026. The top results include "Attention Is All You Need" (the foundational paper), Vision Transformer (ViT), BERT, GPT-3, and their derivatives, sorted by citation count. ``` **Prompt:** ``` What is Geoffrey Hinton's h-index and how many papers has he published? ``` **Response:** ``` I've found Geoffrey Hinton's profile on Semantic Scholar. He has published over 400 papers with a combined citation count exceeding 500,000, giving him one of the highest h-indices in computer science. ``` **Prompt:** ``` Recommend papers similar to "Attention Is All You Need" ``` **Response:** ``` Using the AI recommendation engine with "Attention Is All You Need" as the seed paper, I've found highly relevant papers including BERT, GPT-2, the Universal Transformer, Transformer-XL, and other key works that built upon the attention mechanism paradigm. ``` ## Capabilities ### Search papers by criteria Find relevant articles using keywords and filtering the results by specific fields, years, or top-tier journals. ### Analyze author impact metrics Retrieve detailed professional profiles for researchers, including their total publication count and h-index score. ### Trace citation history Map the intellectual lineage of a paper by finding both its citing works (forward citations) and the papers it references (backward citations). ### Get AI-powered recommendations Discover highly relevant, yet unfamiliar, research using algorithms that analyze content similarity across multiple source papers. ### Process batches of metadata Handle large lists of papers or authors by pulling all necessary metrics in one single request for efficient analysis. ## Use Cases ### Proving the scope of a research problem An R&D team needs to prove that their new AI model addresses an existing gap. They use get_paper_references on key papers in the field and then run search_by_field (Computer Science) combined with searching by year (2015-2020) to define exactly what research was done before them. ### Assessing a collaborator's standing A PI needs to evaluate a potential co-author. They use get_author and search_authors on the candidate, immediately seeing their total paper count, h-index, and key affiliations before committing to collaboration. ### Building a comprehensive literature review A student needs to write a survey of transformer architectures. They start with 'Attention Is All You Need' using get_paper_citations to find all subsequent work, and then use get_recommendations to discover the next key papers they must read. ### Comparing multiple related works A data scientist needs to compare three different models (e.g., BERT, GPT-2, ViT). They feed all three unique identifiers into batch_get_papers to pull and analyze the full metadata simultaneously. ## Benefits - Stop guessing research gaps. Use get_multi_recommendations to find highly relevant, non-obvious papers that build on your existing literature set. - Build robust bibliometric reports instantly. The batch_get_authors tool lets you analyze multiple researcher profiles—including h-index and citation counts—in one shot. - Understand the full life cycle of an idea. Use get_paper_references to find the original foundational work, or use get_paper_citations to see how it influenced later research. - Speed up systematic reviews. The bulk_search_papers tool handles massive result sets with continuation tokens, ensuring you never miss a paper in your review set. - Pinpoint credibility fast. Search by venue lets you filter results only down to top-tier conferences like Nature or Science, guaranteeing high quality. ## How It Works The bottom line is you get deep academic graph analysis right inside your chat client without needing an API key or running local scripts. 1. Connect your preferred AI client to this MCP via Vinkius. 2. Direct your agent to use a search tool, providing the paper title, author name, or specific ID (like a DOI). 3. Your agent executes the query and returns structured data: full abstracts, metrics, citation counts, and links to related work. ## Frequently Asked Questions **How do I find related papers using Semantic Scholar MCP?** Use get_recommendations or get_multi_recommendations. You feed the tool one or more seed papers, and it analyzes content similarity to suggest relevant literature you might not know exists. **Can I search for papers by a specific journal using Semantic Scholar MCP?** Yes, use search_by_venue. You simply name the conference or journal (like Nature or CVPR) and filter all searches to only include articles published there. **What if I don't have a DOI for a paper?** No problem. Try match_paper_title first; it uses fuzzy logic to find the correct metadata even if your title is slightly off or incomplete. You can then use get_paper with the found ID. **How do I compare multiple authors' work?** The batch_get_authors tool is designed for this. Give it a list of names, and you get all their key metrics (h-index, citations) in one request for easy comparison. **Is Semantic Scholar MCP limited to Computer Science research?** Not at all. The tool supports searching across major fields like Medicine, Biology, Physics, and Economics, giving you a massive scope of academic knowledge.