Word Sense Disambiguation of Clinical Abbreviations with Hyperdimensional Computing

AMIA 2013 Annual Symposium

Sungrim Moon ; PhD1 ; Bjoern-Toby Berster2 ; MS

PID  :  122617

来源: CEUR

【 摘 要 】

Automated Word Sense Disambiguation in clinical documents is a prerequisite to accurate extraction of medicalinformation. Emerging methods utilizing hyperdimensional computing present new approaches to this problem. Inthis paper, we evaluate one such approach, the Binary Spatter Code Word Sense Disambiguation algorithm, on 50ambiguous abbreviation sets derived from clinical notes. This algorithm uses reversible vector transformations toencode ambiguous terms and their contextspecific senses into vectors representing surrounding terms. The sensefor a new context is then inferred from vectors representing the terms it contains. Onetoone BSCWSD achievesaverage accuracy of 94.55% when considering the orientation and distance of neighboring terms relative to thetarget abbreviation, outperforming Support Vector Machine and Nave Bayes classifiers. Furthermore, it ispractical to deal with all 50 abbreviations in an identical manner using a single onetomany BSCWSD model with