Beyond causality: Representing event relations in knowledge graphs

Rebboud, Youssra; Lisena, Pasquale; Troncy, Raphaël
EKAW 2022, 23rd International Conference on Knowledge Engineering and Knowledge Management, 26-29 September 2022, Bolzano, Italy

Dynamic environments can be modeled as a series of events and facts that interact with each other, these interactions being characterised by different relations including temporal and causal ones. These have largely been studied in knowledge management, information retrieval or natural language processing, leading to several strategies aiming at extracting these relationships in textual documents. However, more relation types exist between events, which are insufficiently covered by existing data models and datasets if one needs to train a model to recognise them. In this paper, we use semantic web technologies to design FARO, an ontology for representing event and fact relations. FARO allows representing up to 25 distinct relationships (including logical constraints), making it a possible bridge between (otherwise incompatible) datasets. We describe the modeling decision of this ontology resource. In addition, we have re-annotated two already existing datasets with some of the FARO properties.

DOI
HAL
Type:
Conférence
City:
Bolzano
Date:
2022-09-26
Department:
Data Science
Eurecom Ref:
6977
Copyright:
© Springer. Personal use of this material is permitted. The definitive version of this paper was published in EKAW 2022, 23rd International Conference on Knowledge Engineering and Knowledge Management, 26-29 September 2022, Bolzano, Italy and is available at : https://doi.org/10.1007/978-3-031-17105-5_9

PERMALINK : https://www.eurecom.fr/publication/6977