A Problem Representation Approach for Decision Support Systems

AAAI Conferences

The goal of mixed-initiative systems is to synergistically combine the capabilities of the human user and the computer system with the intent to produce higher quality solutions than either the human or the computer can produce independently (Cox & Veloso, 1997; Ferguson, Allen & Miller, 1996; Oates & Cohen, 1994). In one form mixed-initiative systems act as decision support aids, assisting the user in reaching the most optimal problem solution. Supporting the user's ability to monitor the actions of the system and to guide the decision process of the system are two key considerations in the successful design of a decision support system. Both of these points rely on the correct specification of human-computer interaction points. Traditional, computer-centered system design approaches do not do this well, if at all, and are insufficient for the design of decision support systems.


Generating Satellite Control Schedules Using Case-Based Scheduling

AAAI Conferences

Costas Tsatsoulis and Julian Holtzman Lawrence Applied Research Corporation (LARC) Lawrence, KS 66047 {tsatsoul,holtzman}@larc.com Abstract ICARUS is an intelligent system that integrates Case-Based Reasoning and utility theory to remove conflicts from control and task schedules of the Air Force's Satellite Control Network. We describe the methodology that integrates Case-Based Reasoning with utility theory, the problem domain of satellite task scheduling, and how ICARUS is applied to the problem.


The Integration of Planning into Scheduling with OMP

AAAI Conferences

Current planning and scheduling (P&S) literature exposes a trend to integrate both planning and scheduling features in the aim of addressing more challenging real-world problems. Even if planning and scheduling have been usually handled independently using different methods and technologies, it could be easy to find strict connections between them. While in certain application domains the subdivision of the two problems as separate entities is quite motivated (see for example (Srivastava, Kambhampati, & Do 2001; Pecora & Cesta 2005)), in other domains such a clear separation of the planning and scheduling phase is more questionable. From one hand specialized scheduling capabilities are required in a planning engine as soon as non trivial resource constraints have to be taken in account: even if some approaches have pursued the idea to embed resource models directly into a planning engine, it is still not clear how non trivial problems over multi capacity and consumable resources can be afforded without exploiting the powerful capabilities of scheduling reasoners. From the other hand planning capabilities are required in a scheduling environment in order to deal with complex situations where paying attention only to resources capacity constraints is not enough to solve the problem. Very often, it could be necessary to guarantee also some logical orders between the activities that cannot be decided in advance when the scheduling problem is initially formulated. For instance, in several domains a resource could require setup activities whose position cannot be decided all at once but rather depends on the order that the scheduler chooses for other activities during problem solving. Several architectural approaches to integrate planning and scheduling problems exists: for instance O-PLAN (Currie & Tate 1991), IxTeT (Laborie & Ghallab 1995), HSTS (Muscettola et al. 1992), R


TALplanner in the Third International Planning Competition: Extensions and Control Rules

AAAI Conferences

TALplanner is a forward-chaining planner that relies on domain knowledge in the shape of temporal logic formulas in order to prune irrelevant parts of the search space. TALplanner recently participated in the third International Planning Competition, which had a clear emphasis on increasing the complexity of the problem domains being used as benchmark tests and the expressivity required to represent these domains in a planning system. Like many other planners, TALplanner had support for some but not all aspects of this increase in expressivity, and a number of changes to the planner were required. After a short introduction to TALplanner, this article describes some of the changes that were made before and during the competition. We also describe the process of introducing suitable domain knowledge for several of the competition domains.


TALplanner in IPC-2002: Extensions and Control Rules

arXiv.org Artificial Intelligence

TALplanner is a forward-chaining planner that relies on domain knowledge in the shape of temporal logic formulas in order to prune irrelevant parts of the search space. TALplanner recently participated in the third International Planning Competition, which had a clear emphasis on increasing the complexity of the problem domains being used as benchmark tests and the expressivity required to represent these domains in a planning system. Like many other planners, TALplanner had support for some but not all aspects of this increase in expressivity, and a number of changes to the planner were required. After a short introduction to TALplanner, this article describes some of the changes that were made before and during the competition. We also describe the process of introducing suitable domain knowledge for several of the competition domains.