Samuelogeno/sarcasm-detection

Language evolves rapidly. In Kenya, Gen Z and Gen Alpha have largely moved from "Habari yako" to "Niaje" or "Wozaa." For a computer to be useful in the real world, it must move beyond static dictionaries and understand contextual evolution. # Sarcasm Detection & The Pragmatics of "Habari Yako" 🇰🇪 🤖 This project explores the intersection of **Linguistics Theory** and **Modern NLP** using a viral moment from Kenya as a case study. Inspired by the book *'Natural Language Processing in the Real World'* by Jyotika Singh, this repository demonstrates why "getting the joke" is the hardest challenge in AI. ## 📖 The Context: "Habari Yako" When YouTuber IShowSpeed visited Kenya, he repeatedly used the textbook greeting *"Habari yako."* A local kid's sarcastic reply—*"I don't even know what it means"*—perfectly illustrated the concept of **Pragmatics**. While the **Form** and **Semantics** were clear, the **Intent** and **Context** shifted the meaning entirely. ## 🧠 Linguistic Theory Based on NLP fundamentals, language is understood in layers: 1. **Form:** Phonology (sounds), Morphology (units of meaning), and Syntax (grammar). 2. **Semantics:** The literal dictionary meaning. 3. **Pragmatics 🎯:** The social context and intent. This is where sarcasm lives. ## 🚀 The Implementation Using the **Hugging Face Transformers** library, this project utilizes a pre-trained RoBERTa model fine-tuned for irony and sarcasm detection. ## Social Media Post (LinkedIN) linkedin.com