Hello, language hackers! 👋 Have you ever wondered how a computer understands that ‘coffee’ and “tea” are almost the same thing, but ‘tractor’ is from another planet altogether? ☕️🚜
The secret lies in vector embeddings! Applied linguistics turns every word into a long list of numbers (coordinates). Imagine a giant space where: 🌌 Every word is a star. 📏 Words with similar meanings fly together (like constellations). 📐 And mathematics (cosine distance) helps to calculate how ‘close’ they are.
For example, if we take the vector of the word ‘King’, subtract ‘Man’ from it and add “Woman”, we get… the vector of the word ‘Queen’! ✨👑
This is not magic, it is linguistic modelling, thanks to which ChatGPT, Google Translate and smart search engines work. 🧠💻
Want to learn how to build your own universes from words? Come to our ICS department! 🎓🚀