• The Next Word Prediction app was developed using the R package Shiny

• Using a set of cleaned data and transformed into a Corpus, a predictive text model was created as part of the Data Science Capstone Project.

• The outputs of the model included Unigram, Bigram, Trigram and Quadgram (four words) files that were generated from NGram frequencies.

The image below shows the application interface with appropriate labels. The next slide will provide the user instructions.

• The Next Word Prediction App can be accessed by clicking Here!

• In the field provided, enter a valid English word or phrase.

• The value inputted will appear under the ‘Your Input’ heading.

• The algorithm will then automatically output a predicted word under the ‘The next word predicted is:’ heading.

• A user-friendly message will appear below to either confirm the model used to predict the word or prompt the user to try a different word or phrase

Note:

• One valid English word inputted will be calculated using ‘Bigram’
  
• Two valid English words inputted will be calculated using ‘Trigram’
  
• Three valid English words inputted will be calculated using ‘Quadgram’
  

• The Next Word Prediction app demonstrates a working prototype and can be accessed by clicking Here!

• A one word input returned a prediction based on Bigram frequency

• Two words input returned a prediction based on Trigram frequency

• Three words input returned a prediction based on Quadgram frequency

• Non English or incorrectly spelled words returned a friendly message to try a different word or phrase

End of Presentation

Thank you!