George Ferguson, Knowledge Representation and Reasoning for Mixed-Initiative Planning, PhD thesis, University of Rochester, Rochester, NY, 14627, 1995, Technical Report 562.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....problem solving, we include a wide range of applications, including interactive planning and scheduling, task related information retrieval, and command and control applications. Specifically, we have been focusing on mixed initiative planning about transportation scheduling [Allen et al. 1994; Ferguson, 1995] This domain is used as a vehicle for studying two capabilities that we believe are inextricably linked: human machine dialogue and mixed initiative planning. First, we believe it makes little sense to study human machine dialogue independently of a class of tasks that needs to be accomplished. ....

George Ferguson, Knowledge Representation and Reasoning for Mixed-Initiative Planning, PhD thesis, University of Rochester, Rochester, NY, 14627, 1995, Technical Report 562.

.... generally not reactive , in the sense that, given a goal and initial state, they just produce a plan with no more interaction from a user or environment (though most plan execution systems would at least be reactive to the environment in trying to apply plans) Other planning frameworks, e.g. Ferguson, 1995, Lochbaum, 1994 ] could be described as having reactivity and some communicative or collaborative competence, since they can elegantly represent interleaved planning and plan recognition, but might lack the kind of autonomy or negotiative ability to decide when to put forth an initiative rather ....

George Ferguson. Knowledge Representation and Reasoning for Mixed-Initiative Planning. PhD thesis, University of Rochester, 1995. Also available as TR 562, Department of Computer Science, University of Rochester.

....work [6] where the revised model is obtained through the use of heuristics to obtain the minimal model. On the other hand, the update procedures allow us to reason about past and present attitudes and to build an intentional structure of the interaction. This an advantage over Ferguson s approach [5] where his framework does not allow these kind of introspective reasoning. The use of extended logic programming with well founded semantics also avoids the complexity problems with the modal logic approach of Sadek [10] in his definition of rational agents. In the next section, the logic ....

George Ferguson. Knowledge Representation and Reasoning for Mixed-Initiative Planning. PhD thesis, University of Rochester, 1995.

.... ability to specify arbitrary structural constraints on the acceptable solutions, e.g. allowing intermediate and maintenance goals to be specified; cast the planner in the role of a decision support system involving interactive and incremental processing, e.g. to support mixed initiative planning [5]. The dichotomy of the planning logic providing semantics for the actions and the data structure description of plans is one big obstacle to reach the required plan representation flexibility using classical planning approaches. Now, the paper focuses on two aspects. First, a kind of paradigm. ....

G. Ferguson, Knowledge Representation and Reasoning for MixedInitiative Planning, Ph.D. dissertation, University of Rochester, 1995.

....numbers of operators. 6 Discussion In contrast to the action logic of Traverso and Spalazzi [ Traverso and Spalazzi, 1995 ] the logic presented in this paper allows us to deal explicitly with concurrent events. Another work with similar aims to ours is Ferguson s interval based temporal logic [ Ferguson, 1995 ] for use in planning, which supports the representation of concurrent events and action failure. Work which has features in common with our approach to events is that of Drakengren [ Drakengren, 1996 ] who also emphasizes compositionality. As this work was brought to our attention only ....

G. M. Ferguson. Knowledge Representation and Reasoning for Mixed-Initiative Planning. PhD thesis, University of Rochester, Rochester, New York, 1995.

....number of parts. These include: the goals of the plan, the actions which are to be performed in executing the plan, objects used in the plan, and constraints on the execution of the plan. The detailed structure of TRAINS 93 domain plans and the view of plans as arguments are described in detail in [Ferguson, 1995] . Views of domain plans are represented as follows. The shared modality will include aspects of plans assumed to be shared plans jointly intended by the system and the user. The mutually believed proposal modalities include plans proposed by one or the other party but which have not yet been ....

George Ferguson. Knowledge Representation and Reasoning for Mixed-Initiative Planning. PhD thesis, University of Rochester, 1995. Also available as TR 562, Department of Computer Science, University of Rochester.

.... details left unspecified by the user but necessary to connect the current utterance to the plan) and during planning (for example, persistence assumptions) Since these assumptions often drive the dialogue, we have investigated an explicit representation of plans as arguments based on assumptions [Ferguson, 1995]. The trains plan reasoner supports interleaved planning and recognition (as necessary in processing dialogue) and exports a variety of functions to other modules of the system, in particular the dialogue manager. The dialogue manager uses the results of plan reasoning to disambiguate speech act ....

....terms of the structure of the plan graph. Then there are evaluation predicates. These include absolute predicates such as Plausible or Impossible and relative predicates that allow plans to be compared, for example, according to resource usage or time constraints. Further details are available in [Ferguson, 1995]. 4.4 Tyro In order to do the reasoning necessary for executing and monitoring plans and statistical reasoning about the effects of actions, we have developed a language, called Tyro. Like ebtl, Tyro is implemented as an extension of rhet. Tyro allows representation of statistics about the ....

[Article contains additional citation context not shown here]

Ferguson, G. (1995). Knowledge Representation and Reasoning for MixedInitiative Planning. PhD thesis, University of Rochester. Also available as TR 562, Department of Computer Science, University of Rochester.

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