Empirical Observations regarding Predictability in User Access-Behavior in a Distributed Digital Library System

• Document archives are today geographicallydistributed but often not replicated. This canpotentially result in a low quality of service interms of reduced availability and long user-perceivedaccess times, especially during peakhours. Indiscriminate replication is not feasible dueto the sheer size of the database and itsadministration. In an ongoing project, the goal is tostudy the effectiveness of caching techniques likeprefetching and selective preloading to improvequality of service of digital library systems.In this paper, we analyze whether user accessbehavior is predictable enough to use it to guesswhat articles to prefetch or to preload based on useraccess logs from DADS, a digital library systemdeveloped at the Technical Knowledge Center ofDenmark, DTV. We have found that once a literaturesearch has been narrowed down to less than tenarticles, there is a high likelihood that some ofthem will be eventually downloaded. This suggeststhat prefetching can be used to hide the articletransfer latency. We have also found that as many as80% of the article downloads are confined to lessthan 20% of the journals. This suggests thatpreloading a small fraction of the digital librarydatabase can significantly shorten the access latencyas well as improving the availability.

Ämnesord

NATURVETENSKAP  -- Data- och informationsvetenskap -- Datorteknik (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Computer Engineering (hsv//eng)

NATURVETENSKAP  -- Data- och informationsvetenskap -- Annan data- och informationsvetenskap (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Other Computer and Information Science (hsv//eng)

Nyckelord

Digital Library

Caching

Publikations- och innehållstyp

kon (ämneskategori)

ref (ämneskategori)