AtomsMasher: Personal Reactive Automation for the Web

The rise of "Web 2.0" has seen an explosion of web sites for the social sharing of personal information. To enable users to make valuable use of the rich yet fragmented sea of public, social, and personal information, data mashups emerged to provide a means for combining and filtering such information into coherent feeds and visualizations. In this paper we present AtomsMasher (AM), a new framework which extends data mashups into the realm of context-aware reactive behaviors. Reactive scripts in AM can be made to trigger automatically in response to changes in its world model derived from multiple web-based data feeds. By exposing a simple state-model abstraction and query language abstractions of data derived from heterogeneous web feeds through a simulation-based interactive script debugging environment, AM greatly simplifies the process of creating such automation in a way that is flexible, predictable, scalable and within the reach of everyday Web programmers.