• Predicts the next word from a user-entered phrase.
• Uses an N-gram language model.
• Built using the SwiftKey dataset.
• Deployed as a Shiny application.

• Sampled blogs, news, and Twitter datasets.
• Converted text to lowercase.
• Removed punctuation, numbers, URLs, and extra spaces.
• Built unigram, bigram, and trigram models.

• Uses trigram lookup first.
• Falls back to bigram lookup.
• Falls back to the most common unigram.
• Returns one predicted next word.

Features

• User enters a phrase.
• Click Predict.
• Displays the predicted next word instantly.

Application URL:

https://mayuriyadav.shinyapps.io/nextwordprediction/

• Support four-gram and five-gram models.
• Improve prediction accuracy.
• Add spelling correction.
• Optimize memory usage.