Typing speed and accuracy are critical in digital communication.
This app predicts the next word in a typed phrase using language data, helping users type faster — especially in messaging, writing, and search.

• Model: N-Gram Language Model with backoff:
  Trigram → Bigram → Unigram fallback
• Data: Blogs, news, Twitter (English corpora)
  
• Steps:
  • Lowercase conversion
    
  • Removed punctuation/profanity
    
  • Created unigram, bigram, trigram datasets
    
• Tools: R, tidyverse, Shiny

• Input: User types a phrase
  
• Output: Predicts most likely next word

Prediction Flow:
1. Match last 2 words → trigram
2. If not found → bigram
3. Still not found → unigram

Examples:
- “i love” → you
- “thank you” → for
- “based on” → the

Interface & Features - Minimal Shiny UI
- Textbox + one-click prediction
- Uses .rds files → fast response
- Works on browser or local machine

• Can power:
  • Mobile keyboards
    
  • Search assistants
    
  • Writing tools
    
• Future Ideas:
  • Add deep learning (Transformers)
    
  • Grammar + spell correction
    
  • Personalization based on user

Thank you!