Predictive Text for Enhanced User Experience

Around the World: - People are spending more time on mobile devices for activities like social networking, banking, and emails.

The Problem: - Typing on mobile devices can be frustrating and time-consuming.

The Solution: - Predictive text models, like those used by SwiftKey, make typing faster and easier. For example: - When someone types: “I went to the” - Predictions might include: gym, store, restaurant.

Our Mission: - Build a smart predictive text tool to enhance user experience on any device.

Our predictive text model uses n-gram probabilities and a backoff strategy to suggest the next word based on user input.

1. Analyze Input: The model splits the input text into individual words (tokens).
2. Check for Patterns: It searches for matching patterns in:
   • 4-grams: Uses the last three words to predict the next.
   • 3-grams: If no match, uses the last two words.
   • 2-grams: If still no match, uses the last word.
   • 1-gram: Always includes the most frequent single words as a fallback.
3. Combine Predictions: Results from all levels are ranked by probability, and the top three suggestions are shown.

1. Enter a phrase: Type a partial sentence in the input box.
2. Click ‘Predict Next Word’: The app will show you the top 3 predictions.

https://dacia.shinyapps.io/capstone_project_app/

This approach ensures fast, accurate predictions while always offering a suggestion, even with minimal input.

• Fast and efficient: Saves time by predicting the next word in seconds.
• Easy to use: Simply type a phrase and select a prediction.