Connected Data London 2025 Masterclass:
Combining Data from Structured and Unstructured Sources to create High-Quality Knowledge Graphs
We'll be reviewing demo of https://github.com/DerwenAI/strwythura and walking through the code.
The strwythura component is being re-written to make it simpler to
understand the workflow steps and also enhance the components needed
for a more ideal implementation:
- use domain-specific context to constrain relation resolution
- resolve natural language co-reference within the source text to enhance the textgraph algorithm used to build a lexical graph
- use domain-specific context to priorize relations followed during semantic random walks, probably using RL
- enhance inference beyond LLM "reasoning"
- show how to incorporate use of multimodal embeddings
Stay tuned for more soon --
same bat-time, same bat-channel!