Abstract: In Informatics, an ontology is the specification of a system of categories accounting for a certain view of the world. It usually includes categories for the things that are in time, which are commonly called continuants and include things such as a person, a piece of rock, or a machine. An ontology may also include categories for the things that hap pen in time as a transition through successive situations (i.e., instantaneous snapshots of part of the world). Those entities are usually called events or processes and include, e.g., a meeting, the erosion of a mountain, or the manufacturing of a good. Despite the usual priority given to continuants, in practical terms, a good model of events can support several ontology-based reasoning activities, such as pre- and post-condition inference or inference of temporal relations. Accordingly, current ontologies offer powerful modeling constructs that allow us to rep resent a rich variety of types of events. In contrast, they provide much weaker constraints over the possible models that can be constructed. In special, there are several shortcom ings in the current criteria to determine what sequence of situations suitably characterizes the unfolding of a given event and which continuants participate in the event at each of these situations. This lack of clear restrictions on how to model events compromises the ability of such ontologies to guide the modeling process and allows a higher degree of ambiguity in the resulting models. Hence, stricter constraints over the notion of events can be useful to empower modelers to convey the intention behind their models more ef fectively. Besides that, they can help us to uncover novel relations between events and types of events to account for relevant modeling scenarios. In view of that, this work presents a theory for ontological analysis and modeling of events based on the notion of systems as the invariant element that delimits an event. Under this perspective, an event would be a transition through instantaneous snapshots of an invariant system. We argue that such a constraint captures the observed cohesion among the situations that compose the course of an event. Furthermore, it renders a clearer criterion to decide which objects can be said to participate in an event at each instant as well as which succession of situations can adequately trace out the unfolding of an event. Thus, in this work, we introduce the notion of system-invariant events as a type of event whose instances are delimited by systems and derive sub-types according to the type of the system that delimits their instances. Following, we propose an ontological account for the notion of auxiliary events, i.e., events that interfere with other events (e.g., by causing the entry/exit of participants into/from other events, by affecting the dynamics of other events), and derive a taxonomy of auxiliary events based on the type of effect they have on other events. Finally, based on the referred taxonomies and on the principle of onto logical conservation we propose some general guidelines for the modeling of events. We demonstrate this approach with a case study in the domain of Geology (namely, the case of turbidity currents and associated processes such as erosion and deposition).

Keywords: Ontology, Conceptual Modeling, Events, Processes, Occurrents, Auxiliary events, Systems.

Author: Fabrício Henrique Rodrigues

Advisor: Prof. Dr. Mara Abel

Thesis/Dissertation: PPGC, UFRGS, Porto Alegre. Doctorate.

Publication: https://lume.ufrgs.br/handle/10183/263999