Abstract:

We suggest a machine learning approach to classify English sentences as correct or incorrect. The basic idea is to investigate and compare several machine learning techniques to perform such a classification. For this paper, three machine learning tools that recognize syntactically correct English language sentences were used. The performances of the three tools - the C4.5 algorithm, Artificial Neural Network and the Lazy Clustering algorithm - are investigated with regard to a basic training set. The Lazy Clustering algorithm and the trained Artificial Neural Network were found to give a significantly higher result compared with C4.5 algorithm. Results also indicate that these two tools complement each other. This is an encouragement for further investigation to develop a hybrid solution to the problem.