Quote Prediction with LSTM & Greedy Search Decoder

Today, human beings want to express themselves smarter, better, faster and without mistakes. Text predictions increase the speed of typing significantly and also decrease the chance of typos. Auto-text completion property becomes a smart companion for users that helps them to respond to emails quickly, to think ahead for expressing thoughts and to compile documents faster. In this paper, the aim is to generate quotes using mood personalization which involves generating positive, negative or neutral codes on the basis of the input given by the user. The proposed approach deals with the generation of Quotes based on some training on Quotes given by some famous personalities using Long Short-Term Memories and the greedy search decoder approach. A spell checker library is used to replace meaningless words and to find closest in-vocabulary meaningful words depending on Levenshtein distance.