OntologyLink

OntologyLink - Decentralized Knowledge Discovery for Researchers

• Github repository

OntologyLink transforms how researchers discover collaborators and build knowledge networks by creating semantic profiles from personal research notes. Our intelligent analyzer processes researchers' Obsidian vaults, extracting deep knowledge patterns across various domains. The data is structured following specific schemas, creating omprehensive ontological profiles that map not just what you know, but how deeply you know it and how your knowledge connects across disciplines. Finally structured knowledge assets are published to OriginTrail's Decentralized Knowledge Graph (DKG), ensuring permanent, verifiable, and discoverable research profiles.

Main Focus

• Configuring Obsidian and OriginTrail MCP servers
• Defining data structure for ontology profile 
• Optimising prompt to adhere to specified ontology/schema
• Leveraging Claude Desktop as an MPC client

Challenges we faced

• we had to understand how to integrate Obsidian and OriginTrail MCP servers
• we spent a lot of time optimising the prompt for an agent. Agents didn't follow our defined ontology/schema at first, so we had to modify prompts and reiterate in order to achieve well structured data that can be published on DKG.
• we faced http looping issue when trying to publish data to DKG, which we managed to overcome by switching the DKG public node.
  • [originTrail] [info] Message from server: {"jsonrpc":"2.0","id":6,"result":{"content":[{"type":"text","text":"Error creating knowledge asset: Request failed: HTTPSConnectionPool(host='v6-pegasus-node-02.origin-trail.network', port=8900): Max retries exceeded with url: /v1/publish/351b6c76-18ef-4e84-99d3-9e2f82ca8d2e (Caused by NewConnectionError(': Failed to establish a new connection: [WinError 10061] No connection could be made because the target machine actively refused it'))"}],"isError":false}}

Future work

1. robust server architecture to improve data structuring and analysis

2. intelligent matching engine that runs on top of DKG and connects researchers with the common interests

• this would require running of DKG Edge node that connects to a custom server running locally
• the server would run a matching engine logic

3. advanced privacy framework which allows selective data sharing

• edge node allows for private data storage and sharing

Published DKG assets

• https://dkg-testnet.origintrail.io/explore?ual=did:dkg:base:84532/0xd5550173b0f7b8766ab2770e4ba86caf714a5af5/194218
• https://dkg-testnet.origintrail.io/explore?ual=did:dkg:base:84532/0xd5550173b0f7b8766ab2770e4ba86caf714a5af5/194108