The Problem

Typing is slow. Predictive text speeds it up.

Our solution: - Uses natural language patterns from social media - Predicts the next word based on context - Helps users write faster and more naturally

Use cases: - Mobile keyboards
- Chatbots
- NLP research tools

The Algorithm

I built a trigram model trained on 2M+ tweets.

Steps: - Clean and tokenize text
- Extract trigrams (3-word sequences)
- Count frequencies
- Match last two words to predict the third

Simple, fast, and surprisingly accurate.
Includes backoff logic and real-time analytics.

The App

Built with R + Shiny

How it works: - Enter a phrase (e.g., “cant wait till”)
- Click “Predict”
- App returns the most likely next word

Features: - Styled UI with clickable examples
- Responsive layout and theme
- Real-time usage analytics

Try it live: https://emajor.shinyapps.io/WordPredXApp/

Why It Matters

User Experience:
- Clean design
- Instant feedback
- Feels intuitive and intelligent

Innovation:
- Real-world language modeling
- Lightweight deployment
- Easily extendable to other corpora

Thank you!