Sentiment Analysis
Neutral
Negative
Neutral
Positive
Word Importance:
Analyze text to see which words contribute most to the sentiment...
Tokens:
Click "Show Tokens" to see tokenization...
Word Embedding Space (t-SNE visualization) - Similar words cluster together
Understanding the Components:
- Tokenization: Breaking "Hello world!" into ["Hello", "world", "!"]
- Word Embeddings: Each word becomes a vector (e.g., "king" = [0.2, 0.5, -0.1, ...])
- Sentiment Analysis: Classify text as positive, negative, or neutral
- Word Importance: Which words most influence the sentiment (highlighted intensity)
- Context Matters: "not good" is negative, even though "good" is positive!
🤖 Chatbots
Customer service, virtual assistants
🌍 Translation
Google Translate, DeepL
📊 Sentiment Analysis
Social media monitoring, reviews
⚡ Fun Fact: Word Analogies
Word embeddings capture semantic relationships! The famous example: king - man + woman ≈ queen. This works because word vectors encode meaning in their direction and magnitude. Words used in similar contexts end up close together in vector space. Modern models like GPT and BERT use these embeddings to understand and generate human-like text!