I have written the code for generating each unigram and bigrams and then smoothing is used to increase the performance
I have used the perplexity and entropy measures to evaluate our model
It will generate a tweet based on the key-word provided and performs well