Automatic and Effective Mining of Coevolving Online Activities

Given a large collection of time-evolving online user activities, such as Google Search queries for multiple keywords of various categories (celebrities, events, diseases, etc.), which consist of \(d\) keywords/activities, for \(l\) countries/locations of duration \(n\), how can we find patterns and rules? How do we go about capturing non-linear evolutions of local activities and forecasting future patterns? We also aim to achieve good monitoring of the data sequences statistically, and detection of the patterns immediately. In this paper, we present \({\varDelta \hbox {-}\textsc {SPOT}} \), a unifying analytical non-linear model for analysing large scale web search data, which is sense-making, automatic, scalable and free of parameters. \({\varDelta \hbox {-}\textsc {SPOT}} \) can also forecast long-range future dynamics of the keywords/queries. Besides, we also provide an efficient and effective fitting algorithm, which leads to novel discoveries and sense-making features, and contribute to the need of monitoring multiple co-evolving data sequences.