Resumen de Adaptive fragment designs to reduce the latency of web caching in content aggregation systems
Web cache performance has been reduced in Web 2.0 because of the customization levels the increase in content updates. This results in low re-usability among the responses of the user requests. This problem can be solved by reducing the granularity of the contents stored and managed in the web cache. The objective of this thesis is to adapt the content fragments of the web pages in order to reduce the user-perceived latency. The adaptation is based on the values of the content characterization parameters of the web pages (structure and size) and in the user behaviour (update and request rates). We have proposed a solution based on decision trees. These decision trees have been obtained from a knowledge discovery process. In this process, performance and characterization data have been mined. This data mining process is done in an off-line training phase in order to reduce the resource usage.
