From d598467f0fc0ce4cc6db39e0df0f037f18da63d3 Mon Sep 17 00:00:00 2001 From: Munyoki Kilyungi Date: Wed, 24 Jul 2024 11:45:32 +0300 Subject: Add meeting notes for Shelby+Munyoki. Signed-off-by: Munyoki Kilyungi --- topics/meetings/jnduli_bmunyoki.gmi | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'topics/meetings/jnduli_bmunyoki.gmi') diff --git a/topics/meetings/jnduli_bmunyoki.gmi b/topics/meetings/jnduli_bmunyoki.gmi index e5f2023..df5998d 100644 --- a/topics/meetings/jnduli_bmunyoki.gmi +++ b/topics/meetings/jnduli_bmunyoki.gmi @@ -1,4 +1,18 @@ # Meeting Notes +## 2024-07-23 +### LLM Meeting (@shelby+@bmunyoki) +* There's no clear way of ingesting human-readable data with context into the RAG Graph from RDF. +* What specific graph should we ingest into the RAG Graph from RDF? @bmunyoki suggested RIF, PubMed Metadata. We'll figure this out. +* @bonfacem recommended: Much better to work with SPARQL than directly with TTL files. +* We've uploaded rdf triples, yet they loose their strength as the RAG system is not undergirded with a knowledge graph. @bonfacem should read the following for more context and should reach out to @shelby on how to move forward with SPARQL more concretely: + +=> https://r2r-docs.sciphi.ai/cookbooks/knowledge-graph#r2r-knowledge-graph-configuration + +* We need to test the knowledge graph backend of R2R to see how feasible it is to use with the existing data (RDF). +* Fahamu just stored the object and lost the subject+predicate +* Loop in Alex. + + ## 2024-07-19 Plan for this week: -- cgit v1.2.3