Towards a Situation Awareness Framework Based on Primitive Relations

BeAware!—Situation awareness, the ontology-driven way

“…The essential difference of our approach is the incorporation of spatio-temporal relation types, which are introduced in the following section. In particular, related approaches, such as the ones by Matheus et al [9] and Kokar et al [10], provide only limited support for time and space [11] in the form of time instants on objects (but not pre-defined relation types), and locations (but neither spatial reference frame, nor pre-defined spatial relation types, as proposed in our previous work [12]). Instead, these works require the users of their ontologies to define relation types.…”

On Optimization of Predictions in Ontology-Driven Situation Awareness

“…For describing such relations between objects, we discern families of spatial and temporal relations, with each family modeling a certain real-world aspect [3]. These comprise mereotopological reasoning about regions, describing, e. g., whether a traffic jam occurs in a tunnel, positional and orientational reasoning about points, expressing, e. g., that an accident happened in front of a traffic jam, as well as size and distance.…”

“…In situation awareness, however, such critical situations often endanger life and are, besides that, not observable in sufficient quantity to obtain meaningful training data for machine learning (e. g., a wrong-way driver rushing into a traffic jam). In our previous work [3], [4], [5], [6], we therefore proposed-in accordance with Llinas et al [7]-to pursue a different approach using domain-independent ontologies describing qualitative facts for achieving situation awareness, as well as techniques thereupon predicting future situations without relying on historic data. For this, we described situation evolution in terms of transitions of the relations contributing to a situation.…”

“Same, Same but Different” A Survey on Duplicate Detection Methods for Situation Awareness

“…Data Nature. As motivated in the example, duplicate detection methods have to support quantitative data, e. g., spatial positions or timestamps according to global reference systems like WGS-84 [17] or UTC [18], as well as qualitative data, like road names or spatial and temporal relations taking into account mereotopology and orientation [19]. Therefore, we distinguish duplicate detection methods with respect to the supported nature of data, which can be quantitative and/or qualitative.…”