This topic describes the core concepts of vector databases and vector search.
OceanBase AI Database supports dense FLOAT vectors with up to 16,000 dimensions, as well as sparse vectors. It supports various types of vector distance calculations, including Manhattan distance, Euclidean distance, inner product, and cosine distance. It also supports the creation of HNSW/IVF-based vector indexes, as well as incremental updates and deletions, with these operations having no impact on recall.
Core features
OceanBase AI Database provides the capability to store, index, and search for Embedding vector data. Specifically:
Core features |
Description |
|---|---|
| Vector data types |
|
| Vector index |
|
| Vector Search SQL Operators | Supports basic operators such as addition, subtraction, multiplication, comparison, and aggregation for vectors. |
Limitations and considerations:
- OceanBase AI Database uses the NULL first comparison mode by default. Therefore, NULL values are placed at the beginning when sorting. It is recommended to add a
NOT NULLcondition in queries.
Key concepts
Unstructured data
Unstructured data refers to data that does not have a well-defined format or organizational structure. It typically includes text, images, audio, video, as well as social media content, emails, and log files. Due to the complexity and diversity of unstructured data, specific tools and technologies are required to process it, such as natural language processing, image recognition, and machine learning.
Vector
A vector is essentially a projection of an object into high-dimensional space. Mathematically, a vector is a floating-point number array with the following characteristics:
Each element in the array represents a dimension of the vector, and each element is a floating-point number.
The size (number of elements) of the vector array determines the dimensionality of the entire vector space.
Vector embedding
Vector Embedding is the process of using deep learning neural networks to extract content and semantics from unstructured data, transforming images, videos, and other media into feature vectors. Embedding technology maps raw data from a high-dimensional space to a low-dimensional space, converting multimodal data rich in features into multi-dimensional vector data.
Vector similarity search
In today's information age, users often need to quickly search for information from vast amounts of data. Online literature databases, e-commerce platform product catalogs, and growing multimedia content libraries all require efficient search systems to rapidly locate content of interest. As data volumes continue to surge, traditional keyword-based search methods can no longer meet user demands for search accuracy and speed, leading to the emergence of vector search technology. Vector similarity search uses feature extraction and vectorization techniques to convert different types of unstructured data—such as text, images, and audio—into vectors. It then employs similarity metrics to compare these vectors, capturing the deeper semantic information of the data and providing more accurate and efficient search results.