Project Overview

Objective - Build a Shiny app to predict the next word - Apply NLP using n-gram models

Deliverables - Deployed Shiny app - RPubs presentation - Example predictions

Data & Preprocessing

Data Source - SwiftKey corpus (Blogs, News, Twitter)

Preprocessing - Lowercasing - Remove punctuation and numbers - Remove extra whitespace - Tokenization

Prediction Model

Backoff N-gram - Trigram → Bigram → Unigram - Fast, efficient, handles unseen phrases

Shiny App & Examples

Features - Text input for phrases - Predicts next word - Simple UI

Examples

Input Phrase Prediction
The weather today is very sunny
I love watching movies

Conclusion & Future Work

Summary - Built a next-word prediction Shiny app - Used real-world text data - Implemented backoff n-gram model

Future Work - RNN / LSTM models - Improved accuracy - Mobile-friendly UI