IndexKind Class
Definition
Important
Some information relates to prerelease product that may be substantially modified before it’s released. Microsoft makes no warranties, express or implied, with respect to the information provided here.
Defines a list of well known index types that can be used to index vectors.
public static class IndexKindtype IndexKind = classPublic Class IndexKind- Inheritance
-
IndexKind
Remarks
Not all Vector Store connectors support all index types and some connectors may support additional index types that are not defined here. See the documentation for each connector for more information on what is supported.
Fields
| DiskAnn |
Disk-based Approximate Nearest Neighbor algorithm designed for efficiently searching for approximate nearest neighbors (ANN) in high-dimensional spaces. The primary focus of DiskANN is to handle large-scale datasets that cannot fit entirely into memory, leveraging disk storage to store the data while maintaining fast search times. |
| Dynamic |
Dynamic index allows to automatically switch from Flat to Hnsw indexes. |
| Flat |
Does a brute force search to find the nearest neighbors. Calculates the distances between all pairs of data points, so has a linear time complexity, that grows directly proportional to the number of points. Also referred to as exhaustive k nearest neighbor in some databases. |
| Hnsw |
Hierarchical Navigable Small World, which performs an approximate nearest neighbour (ANN) search. |
| IvfFlat |
Inverted File with Flat Compression. Designed to enhance search efficiency by narrowing the search area through the use of neighbor partitions or clusters. Also referred to as approximate nearest neighbor (ANN) search. |
| QuantizedFlat |
Index that compresses vectors using DiskANN-based quantization methods for better efficiency in the kNN search. |