Long Short Term Memory model-based automatic next word generation for text-based applications In contrast to the N-gram model

Main Article Content

P. Niharika, S. John Justin Thangaraj

Abstract

Aim: The proposed study aim is to implement automatic next word generation for text based apps using novel long short term memory model and improving the accuracy using recurrent neural network in comparison with n-gram approach. Materials and Methods: Novel long short term memory is applied on data i.e. text file that consists of a sequence of words. Novel long short term memory model for suggestion accuracy of the next word which compares a n-gram model. LSTM has been proposed and developed in this study. The sample size was measured as 5046 per group with the G power value 0.8. Results: The accuracy was maximum in predicting the next word for text based using long short term memory model 48% with minimum mean error when compared with n-gram model 39% for the same dataset with the p value of 0.02 (p<0.05). Conclusion: The study proves that  novel long short term memory exhibits better accuracy than n-gram in suggesting the next word for text based apps. 

Article Details

Section
Articles