Guess the Word

EugLee
08 Sep 2018

Introduction

This mini application is created to the model Natural Language Processing for the Data Science Capstone Capstone Final Project by John Hopkins University, Coursera.

The Model

Dataset is provided by SwiftKey. The n-gram model is used with n = c(2:4). Expanding from Markov rule, n-values greater than 4 may be of limited value to the model.

For each n-gram, the probability of n-th word given (n-1) words is computed. The model will return the top-10 words for the n-th position.

The model searches in descending order from 4-gram to 2-gram. If no match is found, the model returns NA.

The App

The app is very simplified. User will input a string of words in the input box on the left. The model will then display a table of best guesses and associated probability on the right on a close-to-realtime basis.

https://elyminator91.shinyapps.io/Guess/

Guess the Word

The End