Authors: M. Dikaiakos and D. Zeinalipour-Yazti

Title: A Distributed Middleware Infrastructure for Personalized Services

Conference: Computer Communications Journal (Elsevier)

Year: 2004

Abstract: In this paper we present an overview of eRACE, a modular and distributed intermediary infrastructure that collects information from heterogeneous Internet sources according to registered profiles or end-user requests. Collected information is stored for filtering, transformation, aggregation, and subsequent personalized or wide-area dissemination on the wireline or wireless Internet. We study the architecture and implementation of the main module of eRACE, an HTTP proxy named WebRACE. WebRACE consists of a high-performance, distributed and multithreaded Web crawler, a multithreaded filtering processor and an object cache. We discuss the implementation of WebRACE in Java, describe a number of performance optimizations, and present its performance assessment.