Author Description

This thesis explores the problem of recognizing composite events. Our work extends the event specification language proposed in Vasiliki Kantere's M. Sc thesis titled "A Rule Mechanism for Peer-to-Peer Data Management", and proposes an efficient mechanism, the Recognizer, for composite event recognition. Among other features, the Recognizer supports different kinds of event consumption policies and unrestricted context, so it can be easily applied to different applications. The Recognizer has been implemented and evaluated under different workload assumptions. Peer-to-Peer (P2P) computing has become a popular topic in Computer Science because it offers a new paradigm for data sharing and service provision. Each peer in a P2P network is independent and autonomous, and (in our work) is assumed to own a relational database. Peers can establish (and cancel) acquaintances with other peers, and share data and services with their acquaintances. They can also coordinate their databases through Event-Condition-Action (ECA) rules.