Brian Ward Thomson

Supporting smart interactions with predictive analytics

Smart interactions, where web services are configured and integrated across multiple servers in order to better address the needs of the user, will be much more user-centric and responsive to user needs than current interactions. However, Smart interactions associated with decision-making tasks will specifically have to provide enhanced information or guidance linked to that task. In this paper we examine how predictive analytics can be used to provide cognitive support for smart interactions and outline a method consistent with the smart internet user model to facilitate the creation of predictive analytics components or services to support smart interactions for decision-making tasks.

Intelligence for the personal web

The traditional paradigm for Web interactions, where the interactions are server-driven rather than user-driven, has limitations that are becoming increasingly apparent. The Personal Web proposes to provide intelligent services that support a more user-centric interaction paradigm in order to allow the user to more easily assemble and aggregate web elements to accomplish specific tasks. In this paper we examine the role predictive analytics can play in intelligent services supporting decision-making tasks and describe the Predictive Analytics in Smart Interactions Framework (PASIF), which is a framework for incorporating predictive analytics into intelligent services. PASIF achieves effective levels of support in the dynamic real-time environment of the Personal Web by incorporating ensemble models and techniques to detect and adapt to concept drift in the data sources.