From 09884ed3c13e1525c6419e8fc5ab1a75eb0bcebc Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Lu=C3=ADs=20de=20Sousa?= Date: Tue, 20 Feb 2024 11:57:35 +0100 Subject: [PATCH] Expands text on the knowledge graph --- tech/docs/kmc/knowledge-graph.md | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) diff --git a/tech/docs/kmc/knowledge-graph.md b/tech/docs/kmc/knowledge-graph.md index 92bd5db4..1364e8fb 100644 --- a/tech/docs/kmc/knowledge-graph.md +++ b/tech/docs/kmc/knowledge-graph.md @@ -2,7 +2,21 @@ The knowledge graph is meant to add a formal semantics layer to the meta-data collected at the SWR. It mirrors the XML-based meta-data harnessed in the Catalogue Server, but using Semantic Web standards such as DCAT, Dublin Core, VCard or PROV. This meta-data is augmented with links to domain web ontologies, in particular GloSIS. This semantically augmented meta-data is the main pilar of knowledge extraction activities and components. -The Large Language Model foreseen in this project will be trained on this knowledge graph, thus forming the basis for the Chatbot component of the user interface. The knowledge graph will further feed the facilites for machine-based access to the SWR: a knowledge extration API and a SPARQL end-point. +Besides meta-data the knowledge graph is also expected to host the results of +knowledge extration activities. This assumes that knowledge to be semantically +laden, i.e. linking to relevant domain ontologies. The identification of +appropriate ontologies, and ontology mappinds thus becomes an essential aspect +of this project, bridging together various activities and assets. + +It is important to recognise the knowledge graph as an immaterial asset that +cannot exist by itself. In order to be usable the knowledge graph must be +stored in a triple store, thus highlighting the role of that component in the +architecture. In its turn the triple store provides another important +architectural component, the SPARQL end-point. That will be the main access +gateway to the knowledge graph, particularly by other techonological components +and software. + +The Large Language Model foreseen in this project will be trained on the knowledge graph, thus forming the basis for the Chatbot component of the user interface. The knowledge graph will further feed the facilites for machine-based access to the SWR: a knowledge extration API and a SPARQL end-point. - metadata storage - metadata linking @@ -11,4 +25,4 @@ The Large Language Model foreseen in this project will be trained on this knowle - connections with: APIs, presentation, processing, harvester, metadata scheme, storage & structure - technologies used: DCAT, Dublin Core, VCard, PROV, GloSIS, ... - responsible person: -- participating: Luís de Sousa \ No newline at end of file +- participating: Luís de Sousa