Next Word Prediction Application

Final application submission for the Johns Hopkins Coursera Data Science Capstone

• Next Word Prediction Application is created using Swiftkey datasets of twitter, news, and blog natural language text
• This application uses the N-gram model of natural language processing of a sentence and will try to “guess” the next word
• The prediction model is based on the Pentagram Stupid Backoff model
• There are limitations, I was only able to use 10% of the data
• Shows the top 8 possibilities

The application is straightfoward, the user only need to input a string of text and the application will try to predict the next word.

• N-gram model with “Stupid Backoff” (Brants et al 2007)
• Checks if highest-order (this case n=5) N-gram has been seen. If not it will degrades to a lower model (n=4, n=3 and n=2)
  

• JHU DS Capstone Swiftkey Dataset
• List of Dirty, Naughty, Obscene, and Otherwise Bad Words
• Large Language Models in Machine Translation

• Language Modeling

• Try the Next Word Prediction Application