Incremental generation of spatial referring expressions in situated dialog

Abstract

This paper presents an approach to incrementally generating locative expressions. It addresses the issue of combinatorial explosion inherent in the construction of relational context models by: (a) contextually defining the set of objects in the context that may function as a landmark, and (b) sequencing the order in which spatial relations are considered using a cognitively motivated hierarchy of relations, and visual and discourse salience.

Generating effective referring expressions using charts

“…one which fits only one object in the domain and not any of the others. This perspective has made it possible to apply search-based (Kelleher and Kruijff, 2006), logicbased (Areces et al, 2008) and graph-based (Krahmer et al, 2003) methods to the problem, and overall has been one of the success stories of NLG.…”

Recent Advances in Natural Language Generation: A Survey and Classification of the Empirical Literature

“…Another extension of incremental algorithm is introduced by Kelleher and Kruijff [48] where algorithm is used to generate locative expressions. In this attempt of research, more focus is placed on the context model which signals whether the generated referring expression accurately distinguishes the element need to be referred.…”

Referring expressions as formulas of description logic

“…In this way, it is reminiscent of the Incremental Algorithm, which in fact can be seen as the special case of the EL algorithm for the non-relational case. However, unlike Dale and Haddock (1991) and its successors, such as Kelleher and Kruijff (2006), we do not have to take special precautions to avoid infinite regress. While Dale and Haddock's algorithm attempts to generate a RE for a single individual for successive individuals in the model, our algorithms consider all individuals in parallel and are guaranteed to always terminate.…”

“…However, the classical algorithm by Dale and Haddock (1991) was recently shown to be unable to generate satisfying REs in practice (Viethen and Dale, 2006). Furthermore, the Dale and Haddock algorithm and most of its successors (Krahmer et al, 2003;Kelleher and Kruijff, 2006) are vulnerable to the problem of "infinite regress", where the algorithm jumps back and forth between generating descriptions for two related individuals infinitely, as in "the book on the table which supports a book on the table . .…”

“…(Kelleher and Kruijff, 2006)), where it seems particularly natural to use expressions such as "the book on the table". However, the classical algorithm by Dale and Haddock (1991) was recently shown to be unable to generate satisfying REs in practice (Viethen and Dale, 2006).…”