Search Mailing List Archives

Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[Gofriends] Extracting annotation evidence from free text

Lu MD PhD, Xinghua lux at
Thu Jul 22 07:01:47 PDT 2010

Dear GO friends,

We are text mining group interested in developing statistical methods for automatically identifying textual evidences from free text for GO annotation.  We need help to create a training set of 400 MEDLINE abstract to validate the results.

Using annotation data from GO, we find that many PubMed records are associated with more than one GO annotations; we have developed a statistical model to identify different text regions supporting the observed GO terms, aka, extracting textual evidences for distinct annotations.  We are working at the MEDLINE abstract level first, and planning to test the method on full text later.   We understand that MEDLINE abstracts may not contain sufficient information for annotation but it is a good start.  We have developed a web application allowing ones to associate text regions, i.e., sentences within an abstract, to the observed GO terms.

We already performed some association-annotations and are making our annotations publicly available for anyone interested in such a data set.  I am writing this email to ask for help from expert annotators to help validate our associations so that the data set will contain associations by us (amateurs) and professional annotators.  We would appreciate your help if you can login to our website and annotate a few abstracts so that the data set becomes a resource for the text mining community.  Thanks in advance for your help.

Best regards,

Xinghua Lu, MD, PhD
Associate Professor
Director, Bioinformatics Program
Department of Biochemistry
Medical University of South Carolina
135 Cannon Street, Suite 303
Charleston, SC 29425
Tel: (843)876-1111
FAX: (843)876-1126

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the go-friends mailing list