Text Mining Science

LEAP-FS: Literature Enhanced Automated Prediction of Functional Sites

This project has its own project website.  Please see: LEAP-FS Project Site

LEAP-FS is an approach that integrates high-throughput structure-based and text-based functional site predictions. Functional site identification plays an important role in targeted drug design. The approach is described in our recent PLoS One publication.  We have since explored finer-grained detection of catalytic sites from the biomedical literature in a 2013 Pacific Symposium on Biocomputing paper.

The project is a collaboration between Karin Verspoor, and Michael Wall and Judith Cohn at Los Alamos National Laboratory. It was initially funded by an NIH National Library of Medicine award and now is supported by NICTA and Los Alamos National Laboratory.

Standards for Linguistic Annotation on the Semantic Web

In order to facilitate both the use of biological knowledge (information about biological entities and events) in our text mining systems, and to share the results of our text mining work with the broader community, we are participating in efforts to develop annotation standards for the semantic web. Specifically, we are collaborating with the Open Annotation Collaboration through a grant from the Andrew W. Mellon Foundation. We are also engaging with the W3C Health Care and Life Sciences interest group, as they have related objectives.

I previously worked on an OASIS technical standard for the Unstructured Information Management Architecture (UIMA). UIMA originated at IBM but is now an Apache Open Source project. This provides a standard, modular architecture of facilitate reuse of text analytics. One aspect of our current project is to enable our UIMA pipelines to produce Open Annotation-compatible RDF output.