Qualitative simulation predicts the set of possible behaviors...

to perturbations in its parmneters, and why. For example, comparative analysis could be asked to explain why the period of an oscillating spring/block system would increase if the mass of the block were larger. This paper formalizes the problem of comparative analysis and presents a technique, differential qualitative (DQ) analysis, which solves the task, providing explanations suitable for use by design systems, automated diagnosis, intelligent tutoring systems, and explanation based generalization.

Artificial intelligence has traditionally used constraint satisfaction and logic to frame a wide range of problems, including planning, diagnosis, cognitive robotics and embedded systems control. However, many decision making problems are now being re-framed as optimization problems, involving a search over a discrete space for the best solution that satisfies a set of constraints. The best methods for finding optimal solutions, such as A*, explore the space of solutions one state at a time. This paper introduces conflict-directed A*, a method for solving optimal constraint satisfaction problems. Conflict-directed A* searches the state space in best first order, but accelerates the search process by eliminating subspaces around each state that are inconsistent. This elimination process builds upon the concepts of conflict and kernel diagnosis used in model-based diagnosis[1,2] and in dependency-directed search[3--6]. Conflict-directed A* is a fundamental tool for building model-based embedded systems, and has been used to solve a range of problems, including fault isolation[1], diagnosis[7], mode estimation and repair[8], model-compilation[9] and model-based programming[10].

Shortcomings of qualitative simulation and of quantitative simulation motivate combining them to do simulations exhibiting strengths of both. The resulting class of techniques is called semi-quantitative simulation. One approach to semi-quantitative simulation is to use numeric intervals to represent incomplete quantitative information. In this research we demonstrate semiquantitative simulation using intervals in an implemented semi-quantitative simulator called Q3. Q3 progressively refines a qualitative simulation, providing increasingly specific quantitative predictions which can converge to a numerical simulation in the limit while retaining important correctness guarantees from qualitative and interval simulation techniques. Q3's simulations are based on a technique we call step size refinement. While a pure qualitative simulation has a very coarse step size, representing the state of a system trajectory at relatively few qualitatively distinct states, Q3 interpolates newly expl...

Qualitative simulation of behavior from structure is a valuable method for reasoning about partially known physical systems. Unfortunately, in many realistic situations, a qualitative description of structure is consistent with an intractibly large number of behavioral predictions. We present two complementary methods, representing different trade-offs between generality and power, for taming an important case of intractible branching. The first method applies to the most general case of the problem. It changes the level of the behavioral description to aggregate an exponentially exploding tree of behaviors into a few distinct possibilities The second method draws on additional mathematical knowledge, and assumptions about the smoothness of partially known functional relationships, to derive a correspondingly stronger result. Higher-order derivative constraints are automatically derived by manipulating the structural constraint model algebraically, and applied to eliminate impossible branches These methods have been implemented as extensions to QSIM and tested on a substantial number of examples They move us significantly closer to the goal of reasoning qualitatively about complex physical systems

based on theories by both Hornstein (in the spirit of Reichenbach) and Allen, with lexical-semantic information based on an extended version of Jackendoff's theory that includes a verb classification system proposed by Dowry and Vendlet. The model is intended to be extensible to realms outside of the temporal domain (e.g., the spatial domain). The integration of tense and aspect with lexical-semantics is especially critical in machine translation because of the lexical selection process during generation: there is often a number of lexical connective and tense/aspect possibilities that may be produced from a lexical semantic representation, which, as defined in the model presented here, is largely underspecified. The use of tense and aspect information allows the choice of target-language terms to be more finely tuned and the combination of event structures to be more carefully constrained.

This article is concerned with the general question of how to represent biological knowledge in computers such that it may be used in multiple problem solving tasks. In particular, I present a model of a bacterial gene regulation system that is used by a program that simulates gene regulation experiments,

Generating language that reflects the temporal organization of represented knowledge requires a language generation model that integrates contemporary theories of tense and aspect, temporal representations, and methods to plan text. This paper presents a model that produces complex sentences that reflect temporal relations present in underlying temporal concepts. The main result of this work is the successful application of constrained linguistic theories of tense and aspect to a generator which produces meaningful event combinations and selects appropriate connecting words that relate them. 1 Introduction Reasoning about temporal knowledge and formulating answers to questions that involve time necessitate the presentation of temporal information to users. One approach is to incorporate the temporal information directly into natural language paraphrases of the represented knowledge. This requires a method to plan language that contains not only tense selections, but aspect selections...

Understanding people's commonsense knowledge about physical world is a fundamental problem in building intelligent systems. If this knowledge can be represented and used by computers, they can duplicate people's ability to understand and interact with the world. Qualitative physics is the attempt to capture and formalize this knowledge. An important aspect of qualitative reasoning is the ability to derive the possible behaviors of a given physical system from the structure of the system, using minimal initial information. This thesis investigates qualitative domain theories and reasoning techniques which will en-able computers to analyze the qualitative behaviors of physical systems which include both me-chanical mechanisms and fluids, such as internal combustion engines and hydraulic lift pumps. We have developed a domain theory which integrates richer models of mechanics, fluids, and geometry than previous research in qualitative physics. These theories and inference techniques are embodied in QSA, a program that produces possible behaviors of physical systems. iii To My Parents iv I would like to thank: ACKNOWLEDGMENTS • Professor Kenneth Forbus, my advisor, for initiating and directing this research, for stim-ulating me to think creatively and to stand on my own legs, and for giving me a great example of researcher.

The problems and solutions adressed in this paper belong to the area of knowledge-based process supervision and control. The focus is on the interaction between a therapy planning and execution system, on the one hand, and a simulation module, on the other hand. Generating therapy plans for seriously disturbed dynamic processes inevitably needs at least approximate knowledge about the future. At a first glance, there is a simple scenario of interaction where a therapy planning system is asking a simulation system for the expected values of certain process parameters within some time interval of interest. A closer look exhibits a number of difficulties. The key problem is that the intermediate results of planning are essential for the work of the simulation module. As a consequence, there has to be a proper interaction of these knowledge processing components. This requires extra efforts of control. The interaction itself is becoming subject to in-depth investigations. There is a tradeo...