2026-06-15

Problem Statement

  • Typing on mobile devices can be slow.
  • Predicting the next word improves typing speed.
  • The application suggests the most likely next word based on previous words.
  • Inspired by predictive keyboards such as SwiftKey.

Data and Method

  • Data sources: Blogs, News and Twitter text.
  • A sample of 3,000 lines was used.
  • Text was cleaned and tokenized.
  • Bigram model was created using RWeka.
  • The most frequent word pair is used for prediction.

Application Demo

Input Prediction
of the
in the
to the

The application accepts a phrase and predicts the next word instantly.

Conclusion

  • Simple and lightweight prediction model.
  • Easy to use through a Shiny interface.
  • Demonstrates Natural Language Processing concepts.
  • Can be enhanced using trigrams and larger datasets.