Introducing VelesQL: A SQL-like Query Language for AI Applications
VelesQL is a new query language that extends SQL with vector similarity search, graph pattern matching, and full-text ranking, allowing developers to perform multi-model queries in a single language.
Why it matters
VelesQL's unified query interface can simplify the development of AI applications that require diverse search capabilities.
Key Points
- 1VelesQL keeps the familiar SQL syntax and adds vector similarity search, graph pattern matching, and full-text ranking
- 2It replaces the need for separate vector databases, graph databases, and text search engines, providing a unified query interface
- 3VelesDB is a lightweight, serverless database that runs as a Python library, making it easy to set up and use
Details
Typical AI applications require three types of search: vector search to find documents similar to an embedding, graph traversal to explore relationships between entities, and full-text search to rank documents by keyword relevance. Today, developers often use separate systems and query languages for these tasks, leading to complexity and fragmentation. VelesQL aims to solve this by providing a single query language that supports all three search capabilities. It keeps the familiar SQL syntax while adding new clauses like NEAR for vector similarity and MATCH for graph pattern matching. VelesDB is the underlying database that powers VelesQL, running as a lightweight Python library without the need for a separate server process. This makes it easy to set up and integrate into AI applications.
No comments yet
Be the first to comment