Which kind of relations between events it is interesting to represent?
This question is the focus of our paper Beyond Causality: Representing Event Relations in Knowledge Graphs (Youssra Rebboud, Pasquale Lisena & Raphael Troncy), which has been accepted at EKAW 2022, which will be held in Bolzano (Italy) next week.
In this work, we made an extensive literature review, looking at different data models, ontologies and dataset, comparing the type of event relationship they use. If temporal relations are quite commonly represented (almost in a standard way, thanks to the effort of the TimeML initiative and causality has grown in interest in last years (see for example these workshop at ICLR and NeurIPS), very few works include other possibility such as preventing or enabling an event to happen.
For this reason, we developed the Facts and Events Relationship Ontology (FARO), which can be seen as an harmonisation effort for the event relation types from the literature. With FARO, you can represent in a Knowledge Graph the interconnection of events and condition from the relality. For example, the following sentence:
A tight monetary program caused a temporary downturn but prevented a monetary meltdown
can be represented with this graph.
The paper include also the description of a first effort to produce a dataset and a Knowledge Graph according to these different relations. If you want to know more, read our paper and check our slides.