Querying Temporal Drifts at Multiple Granularities

Sofia Kleisarchaki (1) , Sihem Amer-Yahia (2) , Ahlame Douzal-Chouakria (3) , Vassilis Christophides (4, 5)

1 HADAS - Heterogeneous and Adaptive distributed DAta management Systems
2 SLIDE - ScaLable Information Discovery and Exploitation [Saint Martin d’Hères]
3 LIG - Laboratoire d'Informatique de Grenoble
4 CSD-UOC - Computer Science Department [Crete]
5 MUSE - Measuring networks for enhancing USer Experience

There exists a large body of work on online drift detec- tion with the goal of dynamically finding and maintaining changes in data streams. In this paper, we adopt a query- based approach to drift detection. Our approach relies on a drift index, a structure that captures drift at different time granularities and enables flexible drift queries. We formalize different drift queries that represent real-world scenarios and develop query evaluation algorithms that use different mate- rializations of the drift index as well as strategies for online index maintenance. We describe a thorough study of the performance of our algorithms on real-world and synthetic datasets with varying change rates.

Mots clés: temporal drift detection; Quering; index trees; Multiple granularities

Domaines: Machine Learning [stat.ML]; Intelligence artificielle [cs.AI]

Sofia Kleisarchaki, Sihem Amer-Yahia, Ahlame Douzal-Chouakria, Vassilis Christophides. Querying Temporal Drifts at Multiple Granularities. CIKM 2015, Oct 2015, Melbourne, Australia. ⟨hal-01208397⟩