Temporal Aspects-based Replacement in Media Object Caches

Caching is an appropriate and well-known approach for reducing data transmissions in distributed information systems by creating and maintaining redundant data. As cache memory is limited and cached data might get outdated, it is impossible to store everything forever. If a cache is full, a replacement strategy decides on the cache entry that needs to be replaced by new data. There exist various strategies utilising different indicators for making this decision. Almost all of them do not take the content and the context of the systems’ users into account. In this paper, we present two novel cache replacement strategies called TA and aTA that utilize temporal aspects included in media objects such as websites, specified by the user, or learnt from her or his behaviour. Our evaluation results show that, in the used application scenario, TA and aTA outperform classical replacement schemes like LRU.