🤖 The Vision
In CWFrame v0.2, we don't want search to be just a rigid string match. Our goal is to achieve semantic search, allowing users to find knowledge points through natural language.
！！！==> Real-world Example:
Imagine a user searches for i++. A traditional search would look for that exact string. In CWFrame, our AI-enhanced search should understand the intent and suggest nodes like "Programming Language Core Syntax" or "Arithmetic Operators", even if the node title doesn't contain the characters i++.
🎯 Key Objectives
AI-Enhanced Search: Introduce lightweight models to achieve semantic understanding (e.g., searching for "data structure" will locate "linked list" or "binary tree" even if those two words are not explicitly stated in the node).
Client-side Processing: To protect privacy and ensure responsiveness, all AI logic should run on the front end (Edge AI).
Local Embedding: Convert knowledge graph nodes into vectors (Embeddings) to enable local vector retrieval in the browser.
💡 Tech Stack Ideas We are currently exploring the following path:
Transformers.js: Running the Hugging Face model directly in the browser.
🤖 The Vision
In CWFrame v0.2, we don't want search to be just a rigid string match. Our goal is to achieve semantic search, allowing users to find knowledge points through natural language.
！！！==> Real-world Example:
Imagine a user searches for i++. A traditional search would look for that exact string. In CWFrame, our AI-enhanced search should understand the intent and suggest nodes like "Programming Language Core Syntax" or "Arithmetic Operators", even if the node title doesn't contain the characters i++.
🎯 Key Objectives
AI-Enhanced Search: Introduce lightweight models to achieve semantic understanding (e.g., searching for "data structure" will locate "linked list" or "binary tree" even if those two words are not explicitly stated in the node).
Client-side Processing: To protect privacy and ensure responsiveness, all AI logic should run on the front end (Edge AI).
Local Embedding: Convert knowledge graph nodes into vectors (Embeddings) to enable local vector retrieval in the browser.
💡 Tech Stack Ideas We are currently exploring the following path:
Transformers.js: Running the Hugging Face model directly in the browser.