Keeping track of users' activity is a common scenario in nowadays Internet. Large volumes of usage data are collected and stored on Web servers, proxies and clients throughout the network on daily basis. Properly interpreted, these data are a valuable resource for determining effective marketing strategies, automatic generation of personalized user profiles, and network performance optimization. Information and data correlation needed to perform these tasks are extracted from collected data through data mining process.This paper presents usage tracking performed by the public information system mediator'. Mediator provides a designtime support for development of distributed services and a run-time support for their execution. One of the aspects of the service run-time support is collecting service's usage. Usage collecting mechanism introduced by the mediator is adaptable through service specific filters. Collected usage data are grouped into general and service specific usage data. Based on these two groups of usage data, two types of data mining processes are being proposed. General data mining uses general usage data to extract the usage patterns and optimize performance of the mediator. Service specific data mining is done by the service vendor through semantic interpretation of service specific usage data and is used for extending service offerings and optimizing service performance.
Proxycaches in classical server-client architecture may decrease the amount of data transfer and reduce network congestion, but they also impose a cache-coherence problem. Objects replicated in proxy caches must be kept coherent. Although various protocols for maintaining cache coherence in the World Wide Web (WWW) have been introduced, there does not exist a highly efJicient solution to the coherence problem.We have designed a new hybrid self-adaptive protocol (hSAiTL) for maintaining cache coherence. The protocol dynamically switches between three basic cache coherence mechanisms: server-oriented writeinvalidate, server-oriented write-update, and proxyoriented time-to-live (TTL). We describe the implementation of the hSATTL protocol and compare it to other currently used protocols. In order to perfarm measurements, we also develop a special software package. As a performance measures, we use the level of coherence maintained by the protocols and the amount of network traffic generated by the protocols in order to maintain cache coherence.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.