What is Reor
Reor is an open-source, localized AI personal knowledge management application. The application supports automatic connection of related notes, semantic search, and Q&A features; users edit notes in a Markdown editor similar to Obsidian. Reor is built on technologies such as Ollama, Transformers.js, and LanceDB to enable local operation of LLMs and embedding models, ensuring data privacy. Reor provides a local writing assistant to help users efficiently organize and retrieve knowledge. Reor supports multiple platforms, including Mac, Linux, and Windows, so users can easily download and install it.
Reor's Main Features
- Automatically Connect Related Notes: Reor can automatically identify and connect notes with similar topics or content, without requiring users to manually create links.
- Semantic Search: Supports semantic search; users do not need to remember the exact wording in their notes, but only need to enter relevant keywords, and Reor can return the most relevant search results.
- Q&A Feature: Built-in large language model (LLM); users ask it questions, and the LLM provides answers based on the content of all the user's notes.
- Local-First: All Reor features run and are stored entirely locally, ensuring data privacy.
- WYSIWYG Markdown Editing: Supports Markdown editing; users use Markdown syntax to format notes, and the editor provides a WYSIWYG experience.
- Local Writing Assistant: Provides a local writing assistant so users can get suggestions and prompts from related notes at any time while writing, helping users write more smoothly.
- Supports Running Local Models: Integrates with Ollama, allowing users to download and run local models directly in Reor, or connect to an OpenAI-compatible API.
- Multi-Platform Support: Supports multiple operating systems such as Mac, Linux, and Windows, so users can use Reor on different devices.
Reor's Technical Principles
- Vector Database: Uses an internal vector database to chunk each note and embed it into the database. Based on vector similarity calculation, it automatically connects related notes.
- LLM-Driven Q&A: Q&A is based on an LLM (large language model), using retrieval-augmented generation (RAG) technology to retrieve relevant information from the note corpus and assist in generating answers.
- Local Model Operation: Reor supports running LLMs and embedding models locally, based on technologies such as Llama.cpp, Transformers.js, and LanceDB, ensuring data privacy and performance optimization.
- Markdown Parsing and Editing: Reor uses Markdown as the note format and supports WYSIWYG editing, making it convenient for users to quickly write and organize notes.
- Semantic Search Technology: Based on embedding models, text is converted into vectors, and vector similarity is used for semantic search, improving search accuracy and efficiency.
Reor's Official Website Address
- Project Official Website：https://www.reorproject.org/
- GitHub Repository：https://github.com/reorproject/reor
Reor's Application Scenarios
- Personal Knowledge Management: Helps users organize notes, quickly find related content based on automatic linking and semantic search, and improve knowledge retrieval efficiency.
- Academic Research: Makes it convenient for researchers to organize literature notes and experimental records, quickly associate related research content, assist in paper writing, integrate and cite existing results, and improve research efficiency.
- Team Collaboration: Help team members share knowledge, organize project documents, quickly find relevant experience and best practices within the team, improve collaboration efficiency, and support real-time document collaboration.
- Enterprise Knowledge Management: Enterprises manage internal documents and policy files, build project knowledge bases, make it easy for employees to find and learn, improve overall efficiency, and support employee training material management.
- Creative Writing: Creative workers manage inspiration records and creative drafts, quickly associate them to spark more inspiration, assist in writing novels, scripts, etc., and organize character backgrounds and plot threads.