Named Entity Recognition in Biomedical Literature: A Comparison of Support Vector Machines and Conditional Random Fields

In this paper, we propose two named entity recognition systems for biomedical literature, System1 using support vector machines and System2 using conditional random fields. Through employing several sets of experiments, we make a comprehensive comparison between these two systems. The final results reflect that System2 can achieve higher accuracy than System1, because System2 can catch more essential properties by handling the richer set of features, i.e., adding not only the individual and dynamic features as System1 does but also the combinational features, which can improve the performance further. Furthermore, with carefully designed features, System2 can recognize named entities in biomedical literature with fairly high accuracy, which can achieve the precision of 89.43%, recall of 83.32% and balanced F-beta=1 score of 86.28%.