Evaluating the state of the art in coreference resolution for electronic medical records by Ozlem Uzuner, Andreea Bodnari, Shuying Shen, Tyler Forbush, John Pestian, and Brett R South. (J Am Med Inform Assoc 2012; 19:786-791 doi:10.1136/amiajnl-2011-000784)
Background The fifth i2b2/VA Workshop on Natural Language Processing Challenges for Clinical Records conducted a systematic review on resolution of noun phrase coreference in medical records. Informatics for Integrating Biology and the Bedside (i2b2) and the Veterans Affair (VA) Consortium for Healthcare Informatics Research (CHIR) partnered to organize the coreference challenge. They provided the research community with two corpora of medical records for the development and evaluation of the coreference resolution systems. These corpora contained various record types (ie, discharge summaries, pathology reports) from multiple institutions.
Methods The coreference challenge provided the community with two annotated ground truth corpora and evaluated systems on coreference resolution in two ways: first, it evaluated systems for their ability to identify mentions of concepts and to link together those mentions. Second, it evaluated the ability of the systems to link together ground truth mentions that refer to the same entity. Twenty teams representing 29 organizations and nine countries participated in the coreference challenge.
Results The teams’ system submissions showed that machine-learning and rule-based approaches worked best when augmented with external knowledge sources and coreference clues extracted from document structure. The systems performed better in coreference resolution when provided with ground truth mentions. Overall, the systems struggled in solving coreference resolution for cases that required domain knowledge.
That systems “struggled in solving coreference resolution for cases that required domain knowledge” isn’t surprising.
But, as we saw in > 4,000 Ways to say “You’re OK” [Breast Cancer Diagnosis], for any given diagnosis, there is a finite number of ways to say it.
Usually far fewer than 4,000. If we capture the ways as they are encountered, our systems don’t need “domain knowledge.”
As the lead character in O Brother, Where Art Thou? says, our applications can be as “dumb as a bag of hammers.”
PS: Apologies but I could not find an accessible version of this article. Will run down the details on the coreference workshop tomorrow and hopefully some accessible materials on it.