Semantic Modeling and Inference with Episodic Organization for Managing Personal Digital Traces - (Short Paper).

Many individuals generate a flood of personal digital traces (e.g., emails, social media posts, web searches, calendars) as a byproduct of their daily activities. To facilitate querying and to support natural retrospective and prospective memory of these, a key problem is to integrate them in some sensible manner. For this purpose, based on research in the cognitive sciences, we propose a conceptual modeling language whose novel features include (i) the super-properties “who, what, when, where, why, how” applied uniformly to both documents and autobiographic events; and (ii) the ability to describe prototypical plans (“scripts”) for common everyday events, which in fact generate personal digital documents as traces. The scripts and wh-questions support the hierarchical organization and abstraction of the original data, thus helping end-users query it. We illustrate the use of our language through examples, provide formal semantics, and present an algorithm to recognize script instances.