skills.rest
yunqiqiliang
yunqiqiliangCOMMUNITY
Explore all skills by this author →
GitHub

clickzetta-ai-vector-search

Implement vector storage, indexing, and ANN retrieval in ClickZetta Lakehouse.

8|3|Updated Mar 19, 2026
Search Tags:#vector#ann#hnsw#embedding#semantic-search#vector-search
One-click install
npx skills add https://github.com/yunqiqiliang/clickzetta-skills --skill clickzetta-ai-vector-search
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: clickzetta-ai-vector-search
Source: https://github.com/yunqiqiliang/clickzetta-skills/tree/main/clickzetta-ai-vector-search
Command: npx skills add https://github.com/yunqiqiliang/clickzetta-skills --skill clickzetta-ai-vector-search

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

在 ClickZetta Lakehouse 中实现向量存储、向量索引（HNSW）和向量检索，构建 RAG、语义搜索、图像检索等 AI 应用。覆盖 VECTOR 数据类型定义、向量索引创建（cosine/l2/hamming 距离）、向量数据插入与转换、 ANN 近似最近邻检索、向量+倒排索引融合检索等完整工作流。 当用户说"向量检索"、"向量索引"、"语义搜索"、"embedding 存储"、"RAG"、"ANN 搜索"、"HNSW"、"cosine_distance"、"l2_distance"、"VECTOR 类型"、"向量数据库"、"相似度搜索"、"向量 + 标量融合检索"、"文本向量化"时触发。

Core Features & Use Cases

  • 向量数据类型定义与索引创建（cosine/l2/hamming）的完整工作流，支持向量化检索与混合模式。
  • 构建 RAG、语义检索、图像检索等 AI 应用，结合向量和标量信息进行高效查询。
  • 使用场景示例：在 Lakehouse 中存储文档嵌入并对查询向量进行最近邻检索以返回最相关文档。

Quick Start

创建一个包含 VECTOR 向量列的表并为 vec 列建立向量索引，即可开启向量检索工作流。

Frequently Asked Questions about clickzetta-ai-vector-search

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build RAG and semantic search applications in a Lakehouse environment?▼

RAG and semantic search in a Lakehouse require vector storage, indexing, and retrieval. This implements VECTOR data types and HNSW indexing to store embeddings and perform nearest neighbor searches for AI workflows.

How do I create a vector index for ANN search using cosine and L2 distance?▼

ANN search requires vector index creation with cosine, L2, or Hamming distance metrics. You define a VECTOR data type column and build an HNSW index to enable approximate nearest neighbor retrieval.

Can I perform hybrid retrieval combining vector search with scalar filtering?▼

Hybrid retrieval combines vector and inverted indexes. This allows executing vector plus scalar fusion queries, filtering embeddings by metadata attributes alongside semantic similarity searches.

What is the best way to store and retrieve embeddings for image retrieval workflows?▼

Image retrieval stores generated embeddings in a VECTOR column. After vector insertion and transformation, an HNSW index enables fast similarity searches to return matching images.

Does this vector search approach require a separate vector database?▼

No separate vector database is needed. This implements vector storage, indexing, and ANN search directly within the Lakehouse using native VECTOR data types, avoiding external dependencies.

When should I use HNSW indexing for embedding storage and similarity search?▼

Use HNSW indexing for embedding storage when similarity search requires high-performance approximate nearest neighbor retrieval. It is optimal for large-scale RAG, semantic, and image retrieval workflows.

Related Skills

Explore More Data & Analytics Skills→
zepdbzepdb
43

zeppelin

Index and search text with Zeppelin vector store and BM25 full-text indexing.

Official
Advanced
jhl-labsjhl-labs
50

RAG & Vector Search

Retrieve relevant documents and augment LLM prompts with vector search.

Official
Advanced
kivo360kivo360
1

moai-essentials-refactor

Transform domain knowledge into vector embeddings for semantic retrieval.

Community
Intermediate
winsorllcwinsorllc

Embedding Search

Provides semantic search across documents with Tunrsity and Tunrsity-like features.

Community
Intermediate