As part of the Coursera Data Science Specialization, this text prediction application was developed using Natural Language Processing (NLP) techniques.

The process is illustrated below:

The Text Analysis was made using quanteda in R. N-Grams were created starting with 1-grams and up to 5-grams.

For Smoothing, that is, to rebalance probability for unseen n-grams, I used Interpolation, defined, for a 3-gram example, as:

\[ \begin{aligned} P_{interp}(w_i | w_{i-2}w_{i-1}) = \lambda_1P(w_i | w_{i-2}w_{i-1}) +\\ \lambda_2P(w_i | w_{i-1}) +\\ \lambda_3P(w_i) \end{aligned} \] Where: \[ \lambda_1 + \lambda_2 + \lambda_3 = 1 \]

• Main: to introduce text and get prediction for the next words.

• Stats: Show some useful stats about the usage of the app.

• Try the App at https://adanlp.shinyapps.io/enus/