Skip to Main Content
This paper deals with the problem of mining acronyms and their definitions from biomedical text. We propose an effective text mining system by using pattern matching method. Different stages of the design have been explained with pseudo code. We used space reduction heuristic constraints (D. Nadeau and P. Turney, 2005) which will increase the precision by reducing the number of candidate definitions and will include most of the true cases. The pattern matching method does not require training data to run as in the case of learning techniques. This will make the process simple and fast. Evaluation has been done by using three metrics - recall (measure of how much relevant information the system has extracted from text), precision (measure of how much information returned by the system is actually correct) and f-factor (combined value of recall and precision). Experimental results achieved 92% recall and 97.2% precision.