Event

October 7-9, 2013 Bethesda, Maryland

BioCreative IV Challenge and Workshop

BioCreative: Critical Assessment of Information Extraction in Biology

Part of the workflow NamesforLife uses for constructing controlled vocabularies.

George Garrity presented an overview of the text mining approaches employed by NamesforLife during the DOE Panel on October 8th, 2013.

How might one maintain quality, consistency and usability of stored observational data over time, knowing that both the information and the underlying data are fluid and often inconsistent or even contradictory?

While text mining, natural language processing and machine reasoning are all thought of as computational problems, our experience teaches that the human element, provided by Subject Matter Experts and data curators is crucial if one is to obtain useable and meaningful results. Subject Language Terminologies (SLTs) are dynamic and may contain terms that have many nuanced meanings.

We have developed a generalized process to mitigate these challenges that includes a flexible data model, document analysis methods, and a workflow.

George Garrity, Text Mining Approaches at NamesforLife

Download Presentation (9MB PDF) Download Abstract (86kB PDF)

[permalink] Posted October 7, 2013.