Modeling anticipatory event transitions

Authors:
Ridzwan Aminuddin;Ridzwan Suri;Kuiyu Chang;Zaki Zainudin;Qi He;Ee-Peng Lim
Affiliations:
School of Computer Engineering, Nanyang Technological University, Singapore;School of Computer Engineering, Nanyang Technological University, Singapore;School of Computer Engineering, Nanyang Technological University, Singapore;School of Computer Engineering, Nanyang Technological University, Singapore;School of Computer Engineering, Nanyang Technological University, Singapore;School of Computer Engineering, Nanyang Technological University, Singapore
Venue:
PAKDD'07 Proceedings of the 11th Pacific-Asia conference on Advances in knowledge discovery and data mining
Year:
2007

Citing 2
Cited 0

Bursty and hierarchical structure in streams

Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining
A model for anticipatory event detection

ER'06 Proceedings of the 25th international conference on Conceptual Modeling

Quantified Score

Hi-index	0.00

Visualization

Abstract

Anticipatory Event Detection (AED) attempts to monitor user anticipated events, called Anticipatory Events (AE) that have yet to occur. Central to AED is the Event Transition Graph (ETG), which defines the pre and post states of a user specified AE. A classification model can be trained on documents in the pre and post states to learn to detect an AE. However, this simplistic classification model does not make use of discriminatory keywords between the two states. We propose a simple but effective feature selection method to identify important bursty features that highly discriminate between the pre and post states of an AE. Bursty features are first computed using Kleinberg's Algorithm, then various combination of features in both states are selected. Experimental results show that bursty features can significantly improve the accuracy of AED.