Article Preview
TopThe situated nature of human cognition has been intensively studied (Robbins & Aydede, 2009) with an attempt to represent context in a form available for automatic processing (e.g. McCarthy & Buvac, 1997). Representing situations for automatic processing has implications for the development of future information technology; from building friendly human-computer interfaces to natural language understanding and situation awareness. In this context, Situation Semantics aims “to construct a mathematically rigorous theory of meaning” (Akman, 2009, pp. 402) by “placing situational context (context of situation) at the center of all discussions of meaning” (Akman, 2009, pp. 402). This venture has been mainly pursued by Barwise & Perry (1983)Barwise & Seligman (1997) and Devlin (2006), who have been trying to formulate a mathematically rigorous theory of situation. Despite the rich formalism developed by these authors and others, the attempt to construct a “mathematically rigorous” theory of meaning/situation is in sharp contrast with the accommodating and up to date knowledge on human cognition emphasizing its probabilistic (Griths, Chater, Kemp, Perfors, & Tenenbaum, 2010) rather than logical mathematical form, as well as its heuristic rather than algorithmic nature (Gigerenzer & Gaissmaier, 2011). For instance, let us take the situation of DINNER. What are the objects constituting the Dinner's “Ontology”? A dinner may take place at home or at the restaurant. It may include guests or just family members, and may be used to celebrate thanksgiving or a birthday. While situation theory takes into account the existence of “parameters” as an inherent part of the situation, it does not provide any methodology for identifying the situation's components and relations in a bottom-up fashion. It should be emphasized that while later developments of situation theory seriously consider the linguistic aspect of situations (Devlin, 2006), it does not consider language as a resource for representing a situation in a “soft” and adaptive manner. In this context, it must be noticed that there is an emerging recognition that language mediates a variety of cognitive processes and functions as context for cognition (Barrett, Lindquist, & Gendron, 2007; Neuman, Turney, & Cohen, 2012). Therefore, it seems important to consider ways of fusing cognitive and linguistic resources for better approaching tasks of computational semiotics. In this paper, we adopt the dual-space model of Turney (2012.) proposing (1) domain and (2) function similarity of words as two central dimensions for various tasks of computational semantics. More specifically, we suggest that a meta-form can be approached, as a relational structure comprised of objects and their domain and function similarity.