When 💀 means laughter rather than death: how AI is learning to understand our slang 🎭🤖

To a human, it’s obvious: if someone writes ‘You’re such a genius 🤡’ in a chat, it’s clearly not a compliment. But for a computer algorithm, it’s a real brainteaser!

How do sentiment analysis specialists teach neural networks to distinguish between sincerity and sarcasm?

Token and context analysis:
The model evaluates not the individual word ‘genius’, but its neighbouring words, the exclamation mark and the presence of ‘ironic’ emojis.

Vectorisation of emotions:
Each word is assigned a numerical coefficient (ranging from -1.0 for negative to +1.0 for positive).

Slang datasets:
Computational linguists are constantly updating dictionaries so that AI can understand modern memes and the context of social media.

It is thanks to this that major brands can instantly see whether people like their new product, simply by analysing millions of online comments!

Share to: