A. Blocher and J. R. J. Schirra. Optional Deep Case Filling and Focus Control with Mental Images: ANTLIMA-KOREF. In Proc. of the 14th IJCAI, Montreal, Canada, 1995. to appear.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... does not assign simple truth values to spatial predications, but instead introduces a measure of degrees of applicability that expresses the extent to which a spatial relation is applicable [3] The degree of applicability may also be understood as a measure of typicality for a spatial proposition [10], 37] On the one hand, more exact scene descriptions are possible since the degree of applicability can be expressed by linguistic hedges [26] such as direkt, directly or in etwa, more or less . One the other hand, the degree of applicability can be used to select the most appropriate ....

A. Blocher and J. R. J. Schirra. Optional Deep Case Filling and Focus Control with Mental Images: ANTLIMA-KOREF. In Proc. of the 14th IJCAI, Montreal, Canada, 1995. to appear.

....handle the selection, linearization and encoding of propositions (see [Andr e et al. 88] Herzog Wazinski 94] To meet the requirements of simultaneous scene description, an incremental generation strategy is employed in our approach. The listener model provides an imagination component (see [Blocher Schirra 95] Schirra Stopp 93] in order to anticipate the listener s visual conceptualizations of the described scene. Since we do not assign simple truth values to spatial predications, but instead have introduced a measure of degrees of applicability that expresses the extent to which a spatial ....

A. Blocher und J. R. J. Schirra. Optional Deep Case Filling and Focus Control with Mental Images: ANTLIMA-KOREF. In: Proc. of the 14th IJCAI, Montreal, Canada, 1995. to appear.

.... Wahlster, 1982; Wahlster Kobsa, 1989) The types of user response that have been anticipated in AFLs include the following, among others: the correct interpretation of elliptical utterances (Jameson Wahlster, 1982) the accurate visualization of scene descriptions (Novak, 1987; Schirra, 1995; Blocher Schirra, 1995); the drawing of correct inferences (see, e.g. Joshi, Webber, Weischedel, 1984; Zukerman, 1990) and the pragmatic interpretation of utterances (Jameson, 1989) This research is being supported by the German Science Foundation in its Special Collaborative Research Program on Artificial ....

....the results of the anticipations r 1 : r i Gamma1 . Further variants will be mentioned below. Global Anticipation Feedback To date, almost all implemented systems that have employed anticipation feedback have used a limited part of the system to realize a local AFL. ANTLIMA (Schirra, 1995; Blocher Schirra, 1995) is a rare example of a system that uses a global AFL (Wahlster Kobsa, 1989, pp. 22 26) A large part of the system s own understanding capabilities is used to anticipate the user s responses in ANTLIMA, the way the user will visualize verbal descriptions of events perceived by the system. ....

Blocher, A., and Schirra, J. R. J. 1995. Optional deep case filling and focus control with mental images: ANTLIMAKOREF.

....The named relations can be used both for the generation of spatial descriptions, i.e. from an image which represents a spatial configuration to suitable spatial propositions which represent a spatial expression, and for the analysis of spatial propositions. Following [Schirra and Stopp, 1993] and [Blocher and Schirra, 1995], the latter can be used in an anticipation feedback loop in order to check if a listener will be able to understand the planned utterance in the context. Starting with the spatial propositions corresponding to the utterance, a so called mental image is constructed using a hillclimbing algorithm ....

....one spatial expression because more information is needed to enable the user to identify an object or a position. The understandability of a locative description could, e.g. be checked in an anticipation feedback loop using a listener model with mental images, see [Schirra and Stopp, 1993] and [Blocher and Schirra, 1995]. The answer to a where question, then, can be improved by combining some applicable propositions to get a better spatial description. The number of propositions that can be combined in a German natural language utterance is highly restricted to at most three elementary propositions, e.g. left ....

A. Blocher and J. Schirra. Optional Deep Case Filling and Focus Control with Mental Images: ANTLIMAKOREF. In: Proc. of the 14th IJCAI, pp. 417-- 423, Montreal, Canada. 1995.

....utterance, taking into account the coherence of the whole report. This core system has been extended by two components: Replai [ Retz Schmidt, 1992 ] which recognizes the plans and intentions of the observed agents, and Antlima Koref [ Schirra and Stopp, 1993; Blocher, 1994; Schirra, 1994; Blocher and Schirra, 1995 ] a listener model. 3 Spatial Relations in VITRA Following [ Herskovits, 1986 ] we distinguish between the basic meaning of a preposition itself and an instance of it with respect to some arguments: an object to be localized (LO) and one or more reference objects (RO or REFO) 1 This ....

....and efficient to decode for a listener. 9 In order to produce a description which fulfills these properties, we must consider the listener s knowledge about the context. For this task, we use Antlima, a listener model with mental images [ Blocher et al. 1992; Blocher, 1994; Schirra, 1994; Blocher and Schirra, 1995 ] instances of spatial relations are verified with respect to the typicality of their appearance in the context. Since the same methods are used as for the generation of spatial relations the so called typicality value (t value) corresponds to the degree of applicability. Goal 7 (Left Right) ....

A. Blocher and J. R. J. Schirra. Optional deep case filling and focus control with mental images: ANTLIMA-KOREF. In Proc. of the 14th IJCAI, Montreal, 1995. To Appear.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC