Skip to main content

browse_search

The browse_search tool searches for academic papers across multiple platforms simultaneously. It supports querying 19+ academic databases with flexible filtering options.

Basic Usage

Search a single platform:

browse_search([
{"searcher": "arxiv", "query": "machine learning", "max_results": 5}
])

Search multiple platforms simultaneously:

browse_search([
{"searcher": "arxiv", "query": "deep learning", "max_results": 5},
{"searcher": "pubmed", "query": "cancer immunotherapy", "max_results": 3},
{"searcher": "semantic", "query": "climate change", "max_results": 4}
])

Search all enabled platforms:

browse_search([
{"query": "quantum computing", "max_results": 10}
])

Parameters

ParameterTypeRequiredDefaultDescription
querystringYes-Search query (1-500 characters)
searcherstringNoallSearch platform (omit to search all platforms)
max_resultsintegerNo10Number of results (1-100)
yearstringNo-Year filter (Semantic Scholar only)
fetch_detailsbooleanNotrueFetch paper details (IACR only)
kwargsobjectNo-Additional parameters (CrossRef only)

Available Searchers

Free Data Sources (No API Key Required)

SearcherDescription
arxivPhysics, math, computer science preprints
pubmedMEDLINE biomedical literature
pmcPubMed Central full-text repository
biorxivBiology preprint server
medrxivHealth sciences preprint server
semanticSemantic Scholar AI-powered search
crossrefCrossRef DOI metadata
google_scholarGoogle Scholar search
coreCORE open access aggregator
iacrIACR cryptography preprints
SearcherAPI Key Variable
ieeeIEEE_API_KEY
scopusSCOPUS_API_KEY
springerSPRINGER_API_KEY
sciencedirectSCIENCEDIRECT_API_KEY

Search Examples

# Search for machine learning papers on arXiv
browse_search([
{"searcher": "arxiv", "query": "machine learning", "max_results": 5}
])

# Search for biomedical papers on PubMed Central
browse_search([
{"searcher": "pmc", "query": "cancer treatment", "max_results": 5}
])

# Search for open access papers on CORE
browse_search([
{"searcher": "core", "query": "climate change", "max_results": 5}
])
# Search multiple platforms simultaneously
browse_search([
{"searcher": "arxiv", "query": "deep learning", "max_results": 5},
{"searcher": "pubmed", "query": "cancer immunotherapy", "max_results": 3},
{"searcher": "pmc", "query": "diabetes treatment", "max_results": 3}
])

Platform-Specific Parameters

Semantic Scholar with Year Filter:

browse_search([
{"searcher": "semantic", "query": "climate change", "max_results": 4, "year": "2020-2023"}
])

Year filter formats:

  • Single year: "2019"
  • Year range: "2016-2020"
  • From year onwards: "2010-"
  • Up to year: "-2015"

CrossRef with Additional Filters:

browse_search([
{
"searcher": "crossref",
"query": "deep learning",
"max_results": 5,
"kwargs": {
"filter": "from-pub-date:2020,has-full-text:true",
"sort": "relevance",
"order": "desc"
}
}
])

IACR without Fetching Details:

browse_search([
{"searcher": "iacr", "query": "cryptography", "max_results": 10, "fetch_details": false}
])
# Search IEEE Xplore (requires IEEE_API_KEY)
browse_search([
{"searcher": "ieee", "query": "neural networks", "max_results": 5}
])

# Search Springer Link (requires SPRINGER_API_KEY)
browse_search([
{"searcher": "springer", "query": "quantum computing", "max_results": 5}
])

# Search Scopus (requires SCOPUS_API_KEY)
browse_search([
{"searcher": "scopus", "query": "artificial intelligence", "max_results": 5}
])

Response Format

Results are returned as formatted text:

Source: 'arxiv'
Paper ID: '2303.08774'
Title: GPT-4 Technical Report
Authors: OpenAI
Abstract: We report the development of GPT-4, a large-scale...
Published Date: 2023-03-15
URL: https://arxiv.org/abs/2303.08774
DOI: 10.48550/arXiv.2303.08774
Categories: cs.CL; cs.AI

Input Validation

The tool validates input before searching:

  • query: Must be 1-500 characters, cannot be empty or whitespace only
  • max_results: Must be between 1 and 100
  • searcher: Must be one of the enabled data sources (if specified)
  • year: Must match format YYYY, YYYY-YYYY, YYYY-, or -YYYY

:::tip Best Practice Specify the searcher parameter to target specific platforms. Omitting it will search all enabled platforms, which may be slower and return more results than needed. :::

Error Handling

If a search fails, the tool continues with other searches and returns partial results. Common errors:

  • Invalid searcher: Returns list of available searchers
  • Empty query: Returns validation error
  • API rate limit: Returns error for that data source, continues with others
  • Network timeout: Returns error for that data source, continues with others

Next Steps