A Generalized Model for Characterizing Content Modification Dynamics of Web Objects

Abstract

In this paper, we would like to investigate one of the most fundamental questions behind dynamic web caching, the pattern and modeling for the content modification dynamics of web objects. Since content providers cannot guarantee 100% accuracy about the life-span of a web object, a single-valued expire time is found to be too simple to describe its content modification dynamics for risk analysis to reuse copies of the data in the local cache. To capture the statistical features of the content modification dynamics of a web object, a novel mathematical quantity matrix is proposed. The three statistical metrics of measurement are: dynamic degree, predictability index, and safety bound for prediction. All our arguments and models are verified through active monitoring of the content change of real web objects.