Definition
A mathematical operation that calculates the sum of the products of corresponding elements in two vectors, serving as a similarity metric in RAG to determine the relevance of document chunks to a user query. In vector search, it measures both the directional alignment and the magnitude of the embeddings.
Distinguish from Cosine Similarity: Dot Product is magnitude-sensitive, meaning longer or more 'weighted' vectors yield higher scores unless embeddings are normalized.
"A flashlight beam hitting a target: the score increases not just by how centered the beam is (angle), but also by how bright the bulb is (magnitude)."
- Cosine Similarity(Alternative Metric)
- Embeddings(Prerequisite)
- Vector Database(Execution Environment)
- L2 Normalization(Preprocessing Step)
Conceptual Overview
A mathematical operation that calculates the sum of the products of corresponding elements in two vectors, serving as a similarity metric in RAG to determine the relevance of document chunks to a user query. In vector search, it measures both the directional alignment and the magnitude of the embeddings.
Disambiguation
Distinguish from Cosine Similarity: Dot Product is magnitude-sensitive, meaning longer or more 'weighted' vectors yield higher scores unless embeddings are normalized.
Visual Analog
A flashlight beam hitting a target: the score increases not just by how centered the beam is (angle), but also by how bright the bulb is (magnitude).