Using vector embeddings to classify items into categories without training data for those specific categories. Leverages semantic similarity between text and category descriptions for instant classification.
Loading more......
Zero-shot classification uses embeddings to categorize items without category-specific training examples. Works by computing similarity between item and category descriptions.
categories = [
"product complaint",
"shipping inquiry",
"billing question"
]
# Embed categories
cat_embeddings = [model.encode(c) for c in categories]
# Classify new message
message = "Where is my package?"
msg_embedding = model.encode(message)
# Find closest category
similarities = [cosine_similarity(msg_embedding, c) for c in cat_embeddings]
category = categories[np.argmax(similarities)] # "shipping inquiry"
Depends on embedding model used.