Improving of Automatic Term Recognition in Biotagging Systems

Abstract

Elaborated information technologies are crucial for effective data acquisition and integration from growing body of the biomedical literature. Successful term recognition is the key to getting access to the stored literature information, as it is the terms that convoy knowledge across scientific articles. Due to complexities of dynamically changing biomedical terminology, term recognition has been recognized as the current bottleneck in text mining, and as consequence has become an important research topic both in natural language processing and bioscience communities. Exact match algorithms is often the method used in extracting information from biomedical documents. However exact string match algorithm approach have main problem, it missing term location (low recall) due spelling variations, this thesis tackle this problem by using approximate string match. Experimental results using Genia Tagger revealed that using approximate string match improved Recall and F-score.