Next Word

This is the final project of Coursera Data Science Specialization Capstone course. The course was taught by professors of the Johns Hopkins University with cooperation of Swiftkey. The goal of the project is to build a Shiny application that is able to predict the next word from a line of text.

All text mining was and natural language processing was done with the usage of a variety of well known R packages.

After creating a data sample from the Swiftkey Corpa data, the sample was “cleaned”. The cleaning consisted of converting the data to lower-case, removing punctuation, links, whitespace, numbers, and special characters. The sample was then tokenized into n-grams. N-grams are a continuous sequence of n items from a given sequence of text or speech.

The n-grams were aggregated into data frames for one, two, three, four, and five n-grams. The data frames were used to predict the next word of text to be entered by a user of the app by matching the text entered to the data frames.

The user interface was designed similar to using a text message application on a mobile device. The user begins typing text into a text box. Under the text box, there are three action buttons with potential next words. The user can continue to type or click one of the buttons. Once a button is clicked, the word from the box appends the end of the entered text and new words fill the action buttons.

The app is located at https://dlanncpa.shinyapps.io/Find_Next_Word/.

The supporting documentation is located at https://github.com/dlanncpa/Data_Science_Capstone_Predictive_Model.