Previous Issues


eTBLAST and Déjà vu: a Text Similarity Search Engine and a Database of Highly Similar Citations

Do you need help finding potential reviewers for your grant or paper? Are you looking for the right journal for submission of your manuscript? Are you are curious about publication activity on your topic of interest or concerned that your published biomedical work was plagiarized?

Thanks to a group from the University of Texas Southwestern Medical Center, these questions can be answered using their Web server, eTBLAST, and affiliated database, Déjà vu.

eTBLASTeTBLAST is a free tool designed to search databases, such as MEDLINE (for literature) and CRISP (for grants), and identify documents similar to your search query. First, users enter entire portions of text (as opposed to just keywords), such as their own manuscripts, the discussion section from a previously published work by colleagues, or the specific aims from a grant proposal. Next, eTBLAST uses natural language processing, keyword weighting, and sentence alignment to output a list of articles or grants ranked by relevance to the original text. The results can be organized to:

DejavuExpanding on the functionality of eTBLAST, Déjà vu is a database of highly similar citations. eTBLAST was calibrated to identify articles from MEDLINE exhibiting similar if not identical text and then store them in Déjà vu for subsequent manual inspection to verify the possibility of plagiarism, redundant publications, or translated articles.

Publications in Déjà vu are organized by a large, flexible classification scheme that delineates between appropriate and inappropriate duplication types: distinct, duplicate, erratum, sanctioned, no abstract, and unverified.

To learn more, read the original articles about eTBLAST and Déjà vu.

~ Carrie Iwema

Posted in the 2009 October 2009 Issue