Overlay: The Basic Operation for Discourse Processing

Linkage of Heterogeneous Knowledge Resources within In-Store Dialogue Interaction

“…A sweet spot between the two extremes is to constrain natural language in order to create a formal, user-friendly query language [5] or a controlled language for posing questions [6]. There are diverse examples for current dialogue systems, for instance SmartKom -a multimodal dialogue system that combines speech, gesture and mimics input [7] as well as DELFOS, an dialogue manager system that enables the integration of OWL ontologies as external knowledge resources for dialogue systems [8]. The combination of NLP and ontologies facilitates the development of novel dialogue systems that use ontologies as a core knowledge component regarding linguistic and non-linguistic knowledge representations.…”

“…Presentation of the answer asked to rate questionnaire items with regard to the theoretical constructs described above. Consistent with prior research, we adopted 7-point Likert scales that range from strongly disagree (1) to strongly agree (7).…”

2008 International workshop on content-based multimedia indexing

“…remove ambiguities and contradictory outputs and produce a final semantic interpretation of the multimedia content. Techniques presented in the literature for multimodal fusion, not necessarily for the purpose of knowledge-assisted analysis, include probabilistic approaches [22] and methods that treat information fusion as a structure fusion problem [23], [24]. Among the advantages of techniques belonging to the latter category, such as overlay, is that they rely on structures that represent the semantics of the information to be fused, to identify which portions of information are competing or contradictory.…”

Mind: A Context-Based Multimodal Interpretation Framework in Conversational Systems

“…Currently, MIND applies an operation, called covering, to draw inferences from the conversation context [Chai, 2002a]. Although the mechanism of our covering operation is similar to the overlay operation described in [Alexandersson and Becker, 2001], not only can our covering infer the focus of attention (as overlay does), but it can also infer the intention. What makes this operation possible is our underlying consistent representation of intention and attention at both the discourse and the input levels.…”

“…There are also more sophisticated systems that combine multimodal inputs and outputs [Cassell et al, 1999], and those that work in a mobile environment [Johnston et al, 2002;Oviatt, 2000]. Recently, we have seen a new generation of systems that not only support multimodal user inputs, but can also engage users in an intelligent conversation [Alexandersson and Becker, 2001;Gustafson et al, 2000;Johnston et al, 2002]. To function effectively, each of these systems must be able to adequately interpret multimodal user inputs.…”