Big Data, Linked Data, Smart Dust, Digital Earth, and e-Science are just some of the names for research trends that surfaced over the last years. While all of them address different visions and needs, they share a common theme: How do we manage massive amounts of heterogeneous data, derive knowledge out of them instead of drowning in information, and how do we make our findings reproducible and reusable by others? In a network of knowledge, topics span across scientific disciplines and the idea

more »

... of domain ontologies as common agreements seems like an illusion. In this work, we argue that these trends require a radical paradigm shift in ontology engineering away from a small number of authoritative, global ontologies developed top-down, to a high number of local ontologies that are driven by application needs and developed bottom-up out of observation data. Similarly as the early Web was replaced by a social Web in which volunteers produce data instead of purely consuming it, the next generation of knowledge infrastructures has to enable users to become knowledge engineers themselves. Surprisingly, existing ontology engineering frameworks are not well suited for this new perspective. Hence, we propose an observation-driven ontology engineering framework, show how its layers can be realized using specific methodologies, and relate the framework to existing work on geo-ontologies.