Figure 1

eGRAB module for retrieving Medline abstracts about a gene. Gene names and aliases are gathered from EntrezGene to retrieve Medline abstracts. eGRAB filters out abstracts that mention an ambiguous gene name in some other context. This is done by creating bigram language models for each sense of an ambiguous name and picking the language model that best fits an abstract. DPP is used here to illustrate the ambiguity of gene names (DPP stands for two different genes - Decapentaplegic and Dentin Phosphoprotein - as well as a technique - Differential polarography, among multiple other senses).