AAL Domain Ontology for Event-based Human Activity Recognition

The design of an Ambient Assisted Living (AAL) aims to create better living conditions for the elderly, especially those who choose to live in their own houses, as long as possible. To this objective, AAL systems must mainly monitor the health status of the elderly through the analysis of data gathered via technologies based on sensor devices. Sensors networks produce collections of data of fine-grained nature, regarding general information such as device name, data type, data value, timestamp, but also specific one. The data analysis, due to its granularity and heterogeneity, makes very difficult to infer a clear overall view of the status of the elderly, it demands automatic tools for selecting meaningful data and mapping them in a common conceptual schema. In the last decade, ontologies became widely used tool to describe application domains and to enrich data with its meaning. In this paper, we propose an ontology-based methodology to perform semantic queries on a data repository, where records originated from networks of heterogeneous sources are stored. A semantic query is a pattern matching process that supports the recognition of specific temporal sequences of events that can be extracted from fine-grained data. In our framework a domain ontology are exploited at different levels of abstraction and the reasoning techniques are used to pre-process data for the final temporal analysis. The proposed approach is a deliverable of the ongoing AALISABETH project funded by Region Marche Government; while the software component is integrated into the AALISABETH framework.