Mining the Biomedical Literature in the Genomic Era: An Overview

Abstract

The past decade has seen a tremendous growth in the amount of experimental and computational biomedical data, specifically in the areas of genomics and proteomics. This growth is accompanied by an accelerated increase in the number of biomedical publications discussing the findings. In the last few years, there has been a lot of interest within the scientific community in literature-mining tools to help sort through this abundance of literature and find the nuggets of information most relevant and useful for specific analysis tasks. This paper provides a road map to the various literature-mining methods, both in general and within bioinformatics. It surveys the disciplines involved in unstructured-text analysis, categorizes current work in biomedical literature mining with respect to these disciplines, and provides examples of text analysis methods applied towards meeting some of the current challenges in bioinformatics.

Text mining biomedical literature for constructing gene regulatory networks

“…Precision and recall are the most common parameters when evaluating the IE system. Precision is to evaluate whether the system can only extract correct information, recall is to evaluate whether the system can recognize all useful information (Shatkay et al, 2003). The definition of precision and recall is shown as follows: Precision = number of correctly extracted entities / number of total extracted entities Recall = number of correctly extracted entities / number of all correct entities…”

“…(Shatkay, et al, 2003;Cohen, et al, 2004;Goodman, et al, 2004;Woods, et al, 2004). The experimental data is embedded in tens of thousands of published literature and it is difficult for the individual researcher to extract a comprehensive view of the gene function and regulation in different genomes.…”

Gleaner: Creating ensembles of first-order clauses to improve recall-precision curves

“…Recently, biomedical journal articles have been a major source of interest in the IE community for a number of reasons: the amount of data available is enormous; the objects, proteins and genes, do not have standard naming conventions; and there is interest from biomedical practitioners to quickly find relevant information (Blaschke et al, 2002;Shatkay and Feldman, 2003;Ray and Craven, 2001;Bunescu et al, 2005). We have focused on learning multi-slot protein localization from Medline 1 abstracts, where the task is to identify links between phrases which correspond to a protein and the location of that particular protein in a cell.…”

Evaluating and integrating treebank parsers on a biomedical corpus

“…However, the newspaper English which makes up the bulk of the PTB is only one of many distinct genres of writing in the Anglophone world, and certainly not the only domain where potential natural-language processing (NLP) applications exist that would benefit from robust and reliable syntactic analysis. Due to the massive glut of published literature, the biomedical sciences in general, and molecular biology in particular, constitute one such domain, and indeed much attention has been focused recently on NLP in this area (Shatkay and Feldman, 2003;Cohen and Hunter, 2004).…”