Learning the Grammar of Distant Change in the World-Wide Web

One problem many Web users encounter is to keep track of changes of distant Web sources. Push services, informing clients about data changes, are frequently not provided by Web servers. Therefore it is necessary to apply intelligent pull strategies, optimizing reload requests by observation of data sources. In this article an adaptive pull strategy is presented that optimizes reload requests with respect to the ’age’ of data and lost data. The method is applicable if the remote change pattern may approximately be described by a piecewise deterministic behavior which is frequently the case if data sources are updated automatically. Emphasis is laid on an autonomous estimation where only a minimal number of parameters has to be provided manually.