2026-01-01
Problem Statement
- Typing on mobile devices is slow and error-prone
- Predicting the next word improves speed and user experience
- Goal: build a simple and fast next-word prediction app
Data & Model
- Based on N-gram language models
- Uses:
- Unigrams (single words)
- Bigrams (two-word phrases)
- Trigrams (three-word phrases)
- Higher-order n-grams are given priority
Prediction Algorithm
- User enters a phrase
- Algorithm checks:
- Trigram matches
- Bigram matches
- Most frequent unigram fallback
- Designed for speed and simplicity
Shiny Application
- Interactive web interface using Shiny
- Real-time next-word prediction
- Easy-to-use text input
- Lightweight and responsive
Conclusion
- Demonstrated a working prediction model
- Successfully deployed using Shiny
- Can be extended using larger datasets
- Suitable for real-world text prediction systems