Geo-tagging non-spatial concepts

Abstract

Concept Geo-tagging is the process of assigning a textual identifier that describes a real-world entity to a physical geographic location. A concept can either be a spatial concept where it possesses a spatial presence or be a non-spatial concept where it has no explicit spatial presence. Geo-tagging locations with non-spatial concepts that have no direct relation is a very useful and important operation but is also very challenging. The reason is that, being a non-spatial concept, e.g., crime, makes it hard to geo-tag it. This paper proposes using the semantic information associated with concepts and locations such as the type as a mean for identifying these relations. The co-occurrence of spatial and non-spatial concepts within the same textual resources, e.g., in the web, can be an indicator of a relationship between these spatial and non-spatial concepts. Techniques are presented for learning and modeling relations among spatial and non-spatial concepts from web textual resources. Co-occurring concepts are extracted and modeled as a graph of relations. This graph is used to infer the location types related to a concept. A location type can be a hospital, restaurant, an educational facility and so forth. Due to the immense number of relations that are generated from the extraction process, a semantically-guided query processing algorithm is introduced to prune the graph to the most relevant set of related concepts. For each concept, a set of most relevant types are matched against the location types. Experiments evaluate the proposed algorithm based on its filtering efficiency and the relevance of the discovered relationships. Performance results illustrate how semantically-guided query processing can outperform the baseline in terms of efficiency and relevancy. The proposed approach achieves an average precision of 74% across three different datasets.