Introduction

• This Shiny app predicts the next word based on user input.
• It uses an n-gram language model with backoff smoothing.
• Hosted on shinyapps.io: https://tbros2000.shinyapps.io/predictorofwordsfinal2025/

Prediction Algorithm

• The model is trained on large text corpora using n-grams.
• Backoff model workflow:
  • Check for a matching trigram.
  • If no match, check bigrams.
  • If still no match, default to the most common unigram.
• Data is preprocessed and stored in .RData files for efficient retrieval.

How the App Works

1. User enters a phrase in the text box.
2. The app processes the phrase and applies the backoff model.
3. The predicted next word is displayed.
4. Designed for fast and lightweight word prediction.

User Experience

• Simple and intuitive interface.
• Responsive predictions with minimal delay.
• Suitable for applications like mobile text prediction or chatbot input assistance.

Why This Matters

• Demonstrates efficient NLP in R with Shiny.
• Could be expanded into predictive typing tools.
• Shows ability to develop, deploy, and present a functional data product.
• Would you hire me for your data science startup?