Definition
A high-dimensional mathematical manifold where data is represented as continuous vectors, enabling the measurement of semantic similarity through geometric distance. In RAG, it is the environment where text chunks and user queries are compared, though increasing dimensionality improves precision at the cost of higher latency and storage overhead.
Geometric semantic mapping vs. keyword-based indexing.
"A 3D star map where similar concepts form dense constellations, while unrelated ideas are separated by vast voids of empty space."
- Cosine Similarity(Prerequisite)
- Vector Database(Component)
- Dimensionality Reduction(Optimization)
- Semantic Search(Functional Outcome)
Conceptual Overview
A high-dimensional mathematical manifold where data is represented as continuous vectors, enabling the measurement of semantic similarity through geometric distance. In RAG, it is the environment where text chunks and user queries are compared, though increasing dimensionality improves precision at the cost of higher latency and storage overhead.
Disambiguation
Geometric semantic mapping vs. keyword-based indexing.
Visual Analog
A 3D star map where similar concepts form dense constellations, while unrelated ideas are separated by vast voids of empty space.