Artificial Neural Network With M5P Tree Method Based Typical Error Detection in English Language

In recent times, the growth of network technique and spreading if digital documents has made technique for automatic correction of English text are significant. In English language process, identifying and fix mistakes in word meaning is interest and significant task. In this article, the Artificial Neural Network (ANN) with M5P tree method to detect the typical error in English language. The ANN captures the complex non-linear relationship in data and M5P tree method is utilized for refining the predictions through employing the linear regression to result of ANN. The M5P tree method is included in the traditional ANN method to enhance the detection performance of ANN effectively with high detection accuracy. The ANN with M5P tree method attained the 95.46% accuracy, $\mathbf{93.98\%}$ specificity (True Negative Rate), $\mathbf{93.17\%}$ sensitivity (True Positive Rate), $\mathbf{5.6\%}$ miss-rate (False Negative Rate), 5.448% fall-out (False Positive Rate), 93.7% precision (Positive Predictive Value) and $\mathbf{90.5\%}$ Negative Predictive value (NPV) on training phase. The ANN with M5P tree method showed the effective performed while compared with conventional techniques.