Modeling Collaborative Semantics with a Geographic Recommender

Abstract

In the Semantic Web paradigm, geo-ontologies are closely related to geospatial information communities. Each community comes with its own ontology which is modeled, most frequently, within the framework of description logics. The paper questions a central assumption underlying this approach, namely that communities (and ontologies) are defined by crisp semantic boundaries. The idea of a semantic boundary contrasts sharply with the notion of a community of data producers/consumers that characterizes Web 2.0 applications. Well-known examples are GPS-trail libraries for hikers and bikers or image libraries of places of touristic interest. In these applications, conceptualizations are created as folksonomies by voluntary contributors who associate georeferenced objects (e.g. trails, images) with semantic tags. We argue that the resulting folksonomy can not be considered an ontology in the sense of Semantic Web technology. However, we propose a novel approach for modeling the collaborative semantics of geographic folksonomies. This approach is based on multi-object tagging, that is, the analysis of tags that users assign to composite objects, e.g. a group of photographs.