Ubiquitous Data

Ubiquitous knowledge discovery systems must be captured from many different perspectives. In earlier chapters, aspects like machine learning, underlying network technologies etc. were described. An essential component, which we shall discuss now, is still missing: Ubiquitous Data. While data themselves are a central part of the knowledge discovery process, in a ubiquitous setting new challenges arise. In this context, the emergence of data itself plays a large role, therefore we label this part of KDubiq systems ubiquitous data. It clarifies the KDubiq challenges related to the multitude of available data and what we must do before we can tap into this rich information source. First, we discuss key characteristics of ubiquitous data. Then we provide selected application cases which may seem distant at first, but after further analysis display a set of clear commonalities. The first example comes from Web 2.0 and includes network mining and social networks. Later, we look at sensor networks and wireless sensor networks in particular. These examples provide a broad view of the types of ubiquitous data that exist. They also emphasize the difficult nature of ubiquitous data from an analysis/knowledge discovery point of view, such as overlapping or contradicting data. Finally, we provide a vision how to cope with current and future challenges of ubiquitous data in KDubiq.