Helping NIH Find Reviewers for Proposal

Kareem Darwish

The KEVRIC Company


UMIACS Computational Linguistics Colloquium

September 24, 2003,
11:00am, AVW Room 2120


 

At NIH, incoming grant proposals are evaluated by Study Sections, which are panels of 18 to 20 experts, who are chosen based on their research interests or scientific disciplines.  Each study section is headed by Scientific Review Administrator (SRA) who receives the proposals assigned to his or her study section and is tasked with appropriately identifying members of his study sections or external reviewers who will be fit to review an incoming proposal.  In doing so, they perform one or more of six distinct tasks, namely to identify reviewers or to identify existing publications and their authors based on the text of an incoming proposal, a profile of a reviewer or author, or a manually constructed user query.

The presentation will focus on three aspects of the problem:

1.      Adapting an existing dataset to test the different functions performed by SRA's. This includes constructing reviewer or author profiles.

2.      Examine the effect of using controlled vocabularies and manually entered metadata to enhance search effectiveness.

3.      The effect of document length normalization on retrieval effectiveness.


For the colloquium series schedule, see the UMD Computational Linguistics Colloquium Series web page at http://umiacs.umd.edu/~resnik/cl_colloquium/. If you are interested in meeting with the speaker, please contact Doug Oard (oard@umiacs.umd.edu).