Using Ontologies for Managing User Profiles in Personalised Mobile Service Delivery

Sensor-based human activity recognition system with a multilayered model using time series shapelets

“…For the transformation approach, we represent a user's activity as a set of time series, each one from a particular measured sensor attribute. A shapelet is a representative of a class of time series [44]; that is to say, an activity 80 can be represented by a shapelet. For instance, to represent activities in a game of basketball, "walking" can be represented as a set of time series collected from accelerometer sensors on a user's leg.…”

“…Current approaches for multilayered activity recognition are typically semantic-based because of their ability to model complex relationships in dif-1090 ferent levels of activities (e.g. ontology-based approaches) [80]. However, logical rules or formulae in these semantic-based approaches need to be carefully handcrafted to a specific application [15].…”