Comparative study of machine learning techniques for boundary determination of explanation knowledge from text

摘要

This research aim to determine the explanation knowledge boundary for improvement of basic diagnosis. This paper compares different machine learning techniques including Maximum Entropy, Bayesian Networks, and Naive Bayes for solving the boundary determination problems of the discourse marker's connection problem, usage of several discourse markers within the boundary, and implicit discourse marker. The results have shown an improvement through using machine learning techniques comparing with Centering Theory used in the previous work.