Refining Temporal Representations using OntoElect​

Vadim Ermolayev, Natalya Keberle, Andreas Harth

The progress in understanding the World and its data in their dynamics is based on having an adequately expressive model of time and therefore pushes forward the refinement of time models. The developments in Philosophy, Artificial Intelligence, Databases, Distributed Systems, etc. in the last two decades have brought to life several prominent theoretical frameworks dealing with temporal aspects. Some parts of these theories gave boost to research in logics – yielding a family of temporal logics comprising temporal description logics. Based on this foundation, knowledge representation languages have received their capability to represent time, and several ontologies of time have been implemented by the Semantic Web community. It is however important to find out if this plenty is enough to meet the demand in Semantic Data Management.

To assess the sufficiency of this coverage, the consensual set of the features of time has to be extracted and appropriately structured. A way to do that is to analyze the document corpus produced by the appropriately chosen professional community and extract the requirements for the features from there ­ the TIME community ( in this case. The document corpus for temporal feature extraction has been formed of the proceedings papers of the TIME Symposia series 1994­2013. The OntoElect methodology has been used for the extraction. Using OntoElect also allowed receiving the evidence that the extracted set of features represents a consensual opinion of the community about the requirements regarding the formalizations of time. The extracted feature set was cleaned and structured in the taxonomy of temporal features. This taxonomy was regarded as a requirements set for assessing existing temporal formalisms. The tutorial will present how OntoElect was used for requirements elicitation.

The available theories and ontologies of time have been evaluated against these requirements. The results revealed that several substantial gaps persist in the available ontologies. For example the following temporal features are not satisfactorily addressed: (i) density; (ii) relaxed linearity; (iii) scale factors; (iv) proper and periodic subintervals; (v) temporal measures and clocks. The tutorial will present in detail the results of this evaluation.

Based on this requirement analysis the development of the Syndicated Time Ontology (SOT) has been started to close the revealed gaps. As the solutions, the tutorial will present the key SOT structural contexts of a TimeInstant, TimeInterval, and TimeLine. Special attention will be paid to keeping the output conformant to the W3C standards of OWL 2 DL and SWRL.

Finally, the tutorial will outline how could the temporal features, chosen and developed for SOT, be compared to the other temporal representations, like Memento or other relevant approaches for Linked Open Data (LOD).