Support Vector Machine and Hidden Markov Model in Name Entity Recognition of Natural Language Processing

Abstract

This paper illustrates a comparison between the Hidden Markov Model and the Support Vector Machine, two important methodologies and tools, used in Natural Language Processing. Breaking down the model formulations of each, this paper first describes the mathematical motivations behind their applications in NLP. The mathematical motivations will be discussed through formulas, ideas, and examples. Then, this paper applies two real pre-established algorithms, one for each model, as examples to further rationalize their unique characteristics, similarities, and differences. These aspects will be broken down further into algorithmic efficiency, effectiveness, and other factors. Based on their performances analyzed through each factor, specific toolkits will be proposed, explained, and tested to optimize the test results, as the improving method. Some examples of toolkits include YamCha, TinySVM, etc. Overall, Name Entity Recognition involves different methodologies, and SVM and HMM, which represent two leading areas of NLP research, can best describe future trends and current situations.