Searching for Temporal Patterns in AmI Sensor Data

Romain Tavenard (1) , Albert Ali Salah (2) , Eric Pauwels (3)

1 UR - Université de Rennes
2 UvA - Universiteit van Amsterdam
3 CWI - Centrum voor Wiskunde en Informatica

Data streams emanating from sensors in AmI scenarios often exhibit interesting temporal patterns that point to semantically meaningful events. In this paper we explore two methodologies to uncover this hidden temporal structure: compression-based pattern extractors utilize temporal ordering whereas T-patterns attempt to include estimates for the inter-event times. We suggest improvements to simplify the statistical underpinnings of T-patterns and discuss the relative merits of both methodologies in the light of a number of simulation experiments.

Mots clés: Data mining; Ambient Intelligence

Domaines: Sciences de l'information et de la communication

Romain Tavenard, Albert Ali Salah, Eric Pauwels. Searching for Temporal Patterns in AmI Sensor Data. Constructing Ambient Intelligence, Nov 2007, Darmstadt, Germany. pp.53-62, ⟨10.1007/978-3-540-85379-4_7⟩. ⟨halshs-01138512⟩