class: center, middle, inverse, title-slide .title[ # Next Word Prediction App Pitch ] .author[ ### Bao Khang ] --- # Slide 1: Introduction ## Next Word Prediction App - Built using advanced **N-gram language models** (unigram, bigram, trigram) - Trained on large datasets: Blogs, News, Twitter - Utilizes statistical NLP techniques to predict the next word in real-time --- # Slide 2: Problem & Opportunity ## Why Predictive Text? - Speeds up typing & improves user experience - Valuable for keyboards, chatbots, and voice assistants - Huge market in mobile apps and AI-powered communication tools --- # Slide 3: Our Approach - Sampled & cleaned 3 large corpora (>10 million words) - Used **quanteda** package for efficient tokenization & n-gram creation - Created frequency tables for unigrams, bigrams, trigrams - Developed a Shiny app that predicts the next word given user input --- # Slide 4: App Demo & Features - Enter any phrase to get the most likely next word prediction - Model fallback: uses higher-order n-grams first, then backs off gracefully - Clean UI with real-time prediction display - Works offline with preloaded frequency data --- # Slide 5: Business Impact & Next Steps - Potential integration into keyboard apps and messaging platforms - Opportunity to monetize via API or app marketplace - Next: improve model with smoothing techniques & deep learning - Expand to multiple languages and dialects --- # Thank You!