Mermaid Diagram
flowchart TB subgraph Input D[Document Collection] Q[User Query] end subgraph "Graph-based Text Indexing" direction TB E["Entity & Relation Extraction (LLM-based)"] P["LLM Profiling Generate key-value pairs"] DD["Deduplication Merge identical entities"] E --> P P --> DD end subgraph "Knowledge Graph Construction" direction TB N["Nodes (Entities)"] R["Edges (Relations)"] KV["Key-Value Store"] N --- R N --> KV R --> KV end subgraph "Dual-level Retrieval" direction TB LL["Low-level Retrieval Entity-specific info"] HL["High-level Retrieval Global themes"] KE["Keyword Extraction (LLM-based)"] VM["Vector Matching"] KE --> LL & HL LL & HL --> VM end subgraph "Answer Generation" direction TB C["Context Integration"] G["LLM Generation"] A["Final Answer"] C --> G G --> A end D --> E DD --> N DD --> R Q --> KE VM --> C classDef process fill:#f9f,stroke:#333,stroke-width:2px classDef storage fill:#bbf,stroke:#333,stroke-width:2px classDef input fill:#bfb,stroke:#333,stroke-width:2px classDef output fill:#fbb,stroke:#333,stroke-width:2px class D,Q input class N,R,KV storage class E,P,DD,KE,VM process class A output
Footer
⚠ Switch to EXCALIDRAW VIEW in the MORE OPTIONS menu of this document. ⚠ You can decompress Drawing data with the command palette: ‘Decompress current Excalidraw file’. For more info check in plugin settings under ‘Saving’
Excalidraw Data
Text Elements
Suggestions?
- Blocked due to hardware limitations
Bry’s Personalized & Open Source AI
Research Sources
AVI-0-1-2 PUNTED
AVI-0-1-1 RELEASED
How it works
AVI-Coming-Next
- Insert Markdown, PDF, or Paste
- Visualize Graph
- Chat with your Knowledge
Open Roadmap
AVI-Coming-Later
LitRag Roadmap
Element Links
DDE242mL: https://lightrag.streamlit.app/
FZt5QKrR: https://www.youtube.com/watch?v=PF5NCnBtZsA
hpEsBDTw: https://www.youtube.com/watch?v=oageL-1I0GE&t=1s
5qryyaxA: https://www.youtube.com/watch?v=g21royNJ4fw
dQb23H3b: https://arxiv.org/pdf/2306.08302
fJTKrAOQ: https://arxiv.org/pdf/2410.05779
7fU7BZis: Mermaid Diagram
OL2znQrU: https://www.anthropic.com/news/contextual-retrieval
NbH3BZCQ: LitRAG
Embedded Files
562b48423af391592510ca9dcac43f175410790a: [[CleanShot 2024-10-22 at 09.53.04@2x.png]]
be0f88aa235f8d296b231391c8357493b1738710: [[CleanShot 2024-10-25 at 12.29.13@2x.png]]
683315319fe68930949dfb3e99873c8e7476bac5: [[CleanShot 2024-10-25 at 12.34.31@2x.png]]