This app predicts the next word based on user input using n-gram language models. Inspired by predictive keyboards like SwiftKey.

• Trained on 5% sample from blogs, news, and Twitter (English corpus)
• Uses unigram, bigram, trigram frequency
• Algorithm: Stupid Backoff
• Built with tidytext, dplyr, shiny

Prediction logic:

1. Try trigram
   
2. Backoff to bigram
   
3. Backoff to unigram

• Model Size: 3 .rds files (~30MB)
• Speed: Instant response in Shiny app (<1 sec)
• Accuracy: Most predictions match context in common phrases
• Example input: “I want to” → Predicts: “go”

Optimized for lightweight deployment.

• URL: https://cv6m1k-hanifah-gladis.shinyapps.io/textpredictor/

Usage:

1. Type a phrase: I want to
2. Click Predict
3. Get next word suggestion instantly

Try various inputs and see how model responds!

• Demonstrates core NLP + R skills: data wrangling, modeling, and deployment.
• Reusable framework for SMS prediction, chatbot systems, etc.
• Lightweight, fast, and open-source.

Thank you for visiting my app!