Recommendation
Suggesting items a user is likely to want, from behaviour and content.
Overview
Recommendation systems rank items for a user using past behaviour, item similarity or both. It solves the problem that a catalogue large enough to be useful is too large to browse. It exists because the ranking signal directly changes engagement, and modern systems increasingly combine collaborative signals from interactions with embedding similarity over item content, which is what makes them useful when interaction data is sparse.
Technology Graph
How Recommendation connects to the rest of the stack.
Uses and depends on
- Neural Network concept Algorithms Recommendation uses Neural Network High ✓ source
- Embedding concept Techniques Recommendation based on Embedding High ✓ source