Capstone Project:

Next-Word Predictor App

Coursera Data Science Specialization

John Hopkins University

The app takes a word or (multiple words) from the user. It is an exercise of using a training data from a preprocessed corpus, and some Natural Language Processing algorithms. As output, a predicted word is displayed back to the user. Created using RStudio as a Shiny app

User Instructions:

• Enter a few words in the text area box
• Use the clear button to reset and clear the Textarea box

Probability of an upcoming word:

• P(W) = P(w4|w1,w2,w3)
• P(Wi|Wi-1) = count(Wi,Wi-1)/count(Wi-1)
• P(“its water is so transparent”) = P(its) x P(water|its) x P(is|its water) x P(so|its water is) x P(transparent|its water is so)

Maximum Likelihood Estimation (MLE)

Intead of using probabilities, Kneser-Ney Smoothing and N-grams are used to rank possible next word. The ranking is based on a training data from the Corpus or Dataset using