Bradley S, Giustini D. GoPubMed versus PubReMiner for analyzing PubMed search results: a head to head comparison of two free web ‘data mining’ tools. 2011 CHLA/ABSC Conference, Calgary, Alberta.

From HLWIKI Canada
Jump to: navigation, search
Are you interested in contributing to HLWIKI International? contact:

To browse other articles on a range of HSL topics, see the A-Z index.


See also GoPubMed | PubMed Alternative Interfaces | PubReMiner


GoPubMed versus PubReMiner for analyzing PubMed search results: a head to head comparison - SlideShare

  • Sue M. Bradley. MLIS. School of Library, Archival and Information Studies, Irving K, Barber Learning Centre, 470-1961 East Mall, University of British Columbia, Vancouver BC V6T 1Z1, Canada. E-mail: Phone: 604-733-0214
  • Dean Giustini. UBC Biomedical Branch Librarian, Diamond Health Care Centre, 2775 Laurel Street, Vancouver BC Canada. E-mail: Phone: 604-875-4505

Introduction: When health librarians perform PubMed searches for the systematic review or search for answers regarding clinically-focussed questions, the analyses of natural language terms, index terms (medical subject headings), acronyms, etc., of relevant articles can be an invaluable step in improving search strategies. GoPubMed and PubReMiner are two free websites that perform statistical analyses of PubMed citations’ language usage when PMIDs are entered. In head to head comparisons, however, data discrepancies arose between GoPubMed and PubReMiner. This study examines whether these free web data-mining tools provide reliable suggestions for searchers, and to what extent they are useful in finding additional textwords and MeSH terms.

Methods: PubMed searches were conducted for English language articles from 2001 to 2010 and indexed with MeSH terms “borderline personality disorder”[mh] AND “therapy”[mh]. To allow statistical analysis by hand, the search was developed to yield a reasonably small number of records (n=257). PubMed IDs (PMIDs) were entered into both GoPubMed and PubReMiner and statistical results were compared to those obtained by hand. Discrepancies were examined.

Discussion: GoPubMed and PubReMiner provide statistical analyses of some of the same fields (e.g. publication years, MeSH terms, author names), and some different fields (e.g. PubReMiner includes chemical substance names but GoPubMed does not). Variances in results were found for publication names, author names and countries. Problems were found with both tools, and discrepancies will be discussed.

Conclusion: These two data-mining web tools are a useful supplement for search planning but should be used with caution. Examining PubMed records and using the ‘related articles’ feature are still advised for those developing search strategies.
Personal tools