Semantic & Similarity Search in Oracle Database Using ONNX Models | Oracle 26ai database

In this video, we explore similarity search and semantic search using ONNX models inside Oracle Database.
You’ll see how Oracle 26 AI allows embedding generation and vector search without external vector databases.
This is a complete hands-on walkthrough with real setup steps, errors, and fixes.

This video explains how similarity search (semantic search / vector search) works and how it can be implemented directly inside Oracle Database using ONNX embedding models.

We start with a clear explanation of:
✔️ What ONNX models are
✔️ What similarity and semantic search actually mean
✔️ Why embeddings are required
✔️ How Oracle Database can execute ONNX models natively

Then we move into the practical setup:
✔️ Creating Oracle directory objects
✔️ Creating physical directories on the database server
✔️ Downloading and storing ONNX models
✔️ Understanding how Oracle accesses these models
✔️ Using Claude Desktop with MCP to generate vector tables
✔️ Creating a vector column and sample dataset
✔️ Handling real-world database issues like:
Tablespace quota errors
Auto-extend disabled issues
✔️ Successfully generating vectors using ONNX models inside Oracle Database

This demo focuses on understanding the internal flow rather than just running commands, making it ideal for developers who want clarity, not black-box solutions.

If you are working with Oracle Database, Oracle APEX, PL/SQL, or exploring AI capabilities within Oracle, this video will give you a solid foundation.


Blog's Link: https://intotheoracleverse.blogspot.c...

My Blog has all codes that I have executed (including Tablespace and Data file issues).

Chapters:
00:00 Channel Introduction
00:09 Video Introduction
01:03 ONNX Model Explanation
01:29 Semantic & Similarity Search Explanation
02:40 Benefits of ONNX Models Inside Oracle Database
03:17 Setup Begins – Create Directory Object
04:50 Create Physical Directory in DB File System
05:42 Download ONNX Model into Directory
06:25 Setup Summary
06:46 Using Claude Desktop (MCP) to Create Vector Table
07:46 Access Schema Using MCP
08:55 Prompt Explanation – Vector Column & Sample Data
10:41 Error – Tablespace Quota Issue
10:56 Increasing Tablespace Quota
11:17 Retrying Vector Generation
11:28 Error – Auto Extend Disabled
11:48 Enabling Datafile Auto Extend
12:04 Successful Vector Generation Using ONNX Model
13:17 Conclusion


👍 Like the video if this helped you understand semantic search in Oracle
📌 Subscribe to Into The Oracle Verse for more Oracle & AI content
💬 Comment if you want a deeper dive into queries or Oracle APEX integration


Keywords:
oracle semantic search
oracle similarity search
oracle vector search
onnx oracle database
oracle 26 ai
oracle ai database
onnx embedding model
oracle database machine learning
oracle apex ai
plsql ai