Skip to content

Latest commit

 

History

History
44 lines (32 loc) · 2.79 KB

File metadata and controls

44 lines (32 loc) · 2.79 KB

Content Search

Content Search is a core multimodal service designed for smart classroom environments. It enables AI-driven video summarization, document text extraction, and semantic search capabilities using advanced RAG (Retrieval-Augmented Generation) workflows.

Quick Start

Automatic Dependency Installation

We provide a unified installation script that automates the setup of the databases, Python virtual environment, and core dependencies.

Note: Open PowerShell as Administrator before running the script.

# Run the automation script from the content search root
.\install.ps1

Launching Services

Once the environment is configured, activate the virtual environment and start the orchestration service:

# Activate the virtual environment
.\venv_content_search\Scripts\Activate.ps1

# Start all microservices
python .\start_services.py

API Endpoints

Endpoint Method Pattern Description
/api/v1/task/query/{task_id} GET SYNC Task Status Inspection: Retrieves real-time metadata for a specific job, including current lifecycle state (PENDING, PROCESSING, COMPLETED, FAILED), and error logs if applicable.
/api/v1/task/list GET SYNC Batch Task Retrieval: Queries task records. Supports filtering via query parameters (e.g., ?status=PROCESSING) for monitoring system load and pipeline efficiency.
/api/v1/object/ingest-text POST ASYNC Text-Specific Ingestion: Primarily processes raw text strings passed in the request body for semantic indexing. It also supports fetching content from existing text-based objects in MinIO.
/api/v1/object/upload-ingest POST ASYNC Atomic Upload & Ingestion: A unified workflow that first saves the file to MinIO and then immediately initiates the ingestion pipeline. Features full content indexing and AI-driven Video Summarization for supported video formats.
/api/v1/object/search POST SYNC Semantic Content Retrieval: Executes a similarity search across vector collections using either natural language queries or base64-encoded images. Returns ranked results with associated metadata and MinIO object references.
/api/v1/object/download POST STREAM Original File Download: Securely fetches the raw source file directly from MinIO storage. Utilizes stream-bridging to pipe binary data to the client.

For detailed descriptions and examples of each endpoint, please refer to the: Content Search API reference

Components API reference

Ingest and Retrieve

Video Preprocess

VLM OV Serving