Situating Spatial Language and the Role of Ontology: Issues and Outlook

Abstract

Although the analysis and formalisation of the meaning of spatial expressions represents an extremely active area of research at this time, there remain significant problems in relating spatial expressions to their intended interpretations in context. In this article we motivate a spatial semantics architecture that draws a formal separation of linguistic semantics and non‐linguistic spatial descriptions. Methods drawn from ontological engineering are suggested to be relevant both for the representations of these areas and for specifying the interrelationships between them. In addition, a characterisation of current ontologies involving space is given, clarifying why spatial language is particularly difficult to interpret. The article concludes with our own proposal for a general architecture for natural language spatial semantics capable of being used across diverse applications. Extensive bibliographical references to the state of the art in spatial language and spatial ontologies are provided.

Language and Space: a two-level semantic approach based on principles of ontological engineering

“…Despite the fact that the analysis and formalization of the meaning of spatial expressions is an extremely active area of research at this time (for extensive literature reviews, see: Carlson and van der Zee 2005;Tenbrink 2007;Coventry et al 2009;Bateman 2010b), there remain significant problems in formulating mechanisms by which this interpretation might be achieved. In particular, the very notion of the 'meaning' of spatial expressions and terms in natural languages is still approached from a variety of diverging perspectives that are less than optimal for supporting interpretation in context.…”

Learning to interpret spatial natural language in terms of qualitative spatial relations*

“…Our approach is motivated by the potential cognitive adequacy of its components as well as by technical benefits of the component distinctions. The diversity and variety of spatial language usage (for more detailed examples see Denis and Fernandez; Taylor and Brunyé; Seifert and Tenbrink, this volume) requires a flexible mechanism that allows a situation-based interpretation (Bateman, 2010a) resulting in a combination of different spatial components. Here, we concentrate on one component for representing spatial language in an adequate way and another component for representing qualitative spatial information, and we map the first to the second (Kordjamshidi et al, 2011a).…”

“…Such models reflect, however, typically not spatial language semantics but specific meanings that are determined by the model's axioms. The diverse use of language requires yet more complex and flexible logical approaches (Bateman, 2010a). These approaches have to take into account contributing aspects for spatial language interpretations, such as context or world knowledge, in order to achieve a mapping from spatial language to qualitative representations (Hois and Kutz, 2008b;Kordjamshidi et al, 2010).…”

Modeling Spatial Knowledge for Generating Verbal and Visual Route Directions

“…Approaches following [6] propose to link underspecified linguistic and rich non-linguistic levels of representation [8, inter alia]. In contrast, we follow Bateman [1] who argues that linguistic and non-linguistic information needs to be kept in separate ontological modules and needs to relate individual concepts as required by the context.…”

“…A new line of approaches distinguishes the inherent semantics of linguistic expressions and their spatial interpretation in a given situation [1]. This is necessary because the interpretation involves a non-trivial mapping between ontologies and concepts from spatial language.…”