Recent developments have clarified the process of generating partially ordered, partially specified sequences of actions whose execution will achieve an agent's goal. This article summarizes a progression of least commitment planners, starting with one that handles the simple STRIPS representation and ending with UCPOP, a planner that manages actions with disjunctive precondition, conditional effects, and universal quantification over dynamic universes. Along the way, I explain how Chapman's formulation of the modal truth criterion is misleading and why his NP-completeness result for reasoning about plans with conditional effects does not apply to UCPOP.

The vast amounts of on-line text now available have ledto renewed interest in information extraction (IE) systems thatanalyze unrestricted text, producing a structured representation ofselected information from the text. This paper presents a novel approachthat uses machine learning to acquire knowledge for some of the higher level IE processing. Wrap-Up is a trainable IE discourse component that makes intersentential inferences and identifies logicalrelations among information extracted from the text. Previous corpus-based approaches were limited to lower level processing such as part-of-speech tagging, lexical disambiguation, and dictionary construction. Wrap-Up is fully trainable, and not onlyautomatically decides what classifiers are needed, but even derives the featureset for each classifier automatically. Performance equals that of a partially trainable discourse module requiring manual customization for each domain.

This paper is a multidisciplinary review of empirical, statistical learning from a graphical model perspective. Well-known examples of graphical models include Bayesian networks, directed graphs representing a Markov chain, and undirected networks representing a Markov field. These graphical models are extended to model data analysis and empirical learning using the notation of plates. Graphical operations for simplifying and manipulating a problem are provided including decomposition, differentiation, andthe manipulation of probability models from the exponential family. Two standard algorithm schemas for learning are reviewed in a graphical framework: Gibbs sampling and the expectation maximizationalgorithm. Using these operations and schemas, some popular algorithms can be synthesized from their graphical specification. This includes versions of linear regression, techniques for feed-forward networks, and learning Gaussian and discrete Bayesian networks from data. The paper concludes by sketching some implications for data analysis and summarizing how some popular algorithms fall within the framework presented. The main original contributions here are the decompositiontechniques and the demonstration that graphical models provide a framework for understanding and developing complex learning algorithms.

These software between telephone, television, agents are on their way, and they're going to The be getting a lot of things accomplished by basic idea is that the networks that constitute interacting with each other. The question is, our telephone infrastructure, our television How will these agents be cooperating with (particularly cable) infrastructure, and our each other, competing with each other, and computer infrastructure will be coalescing into negotiating with each other? Now, the agents that we are interested in Another example is routing among looking at are heterogeneous, self-motivated telecommunication networks. The systems are not assumed to be packets, can pass over a network controlled by centrally designed. For example, if you have a one company onto another network controlled personal digital assistant, you might have one by another company, or it can pass that was built by IBM, but the next person through one country on through another. Computers that control a telecommunications They don't necessarily have a notion of global network might find it beneficial to enter into utility. Each personal digital assistant or agreements with other computers that control each agent operating from your machine is other networks about routing packets more interested in what your idea of utility is and efficiently from source to destination. The in how to further your notion of goodness. We're other agents ask them to do unless they have Another example is the proliferation of shared databases, where there's information They have sprung up with a vengeance in the last decade.

Several factors Chapter 2 describes a small fragment Chapters 8, 9, and 10 describe have led to the increase in interest in of probability and information recent research on more isolated this field, which is heavily influenced theory, including brief coverage of aspects of parsing and language analysis.

In Greed in the Quest for Machines That the first wave of AI businesses were addition, when expert systems began Think, Harvey P. Newquist, Sams Publishing, researchers, sneering dominates over to be written in The author's aversion to places away they could implement their applications from the executive suite distorts the in house at a lower cost. Gold Hill, and other took root as an academic companies founded by Ed Feigenbaum. Inc., marketing a symbolic mathematics because pioneering companies making who covered the field during the small assembly robots and industrial program that was once a 1980s when academic researchers vision systems failed just as the robots minor product. Teknowledge was went commercial in one of the 1980's became essential to manufacturing reduced to a small division. Alan Newell's world-leading of traditional companies now use AI begins with a history spanning Babbage but unmarketed reasoning program techniques in house for such things to Turing to Minsky, McCarthy, research at Carnegie Mellon University, as geological exploration, financial Newell, Simon, Samuel, and others at conducted vigorously through the decision making, medical advice, factory the 1956 Dartmouth meeting and 1980s, is dismissed.

Chevron Canada is decentralizing its computer environment from mainframes to PCs e Emission Reduction Research mounted infantry virtual environment) and Sun workstations. The DIVE belt provides the Chevron oil exploration crews will be N.J.) has developed the Batch Design ability to operate inside a virtual environment able to retrieve various types of well Kit, an expert system for optimizing without becoming tangled batch processes and minimizing pollution. Funded by the U.S. The system will help eliminate Army, the DIVE project is designed to ADVANTA Mortgage (San Diego, avoidable pollution and save pharmaceutical allow soldiers to operate within a virtual Cal.) has signed a license agreement and chemical manufacturers battlefield. VR is being used to demonstrate used as the focal point of exhibition fire engineering principles such The Santa Fe Institute (Santa Fe, stands designed by Photosound for as means of escape theory, fire modeling, N.M.) has won an ARPA grant of such pharmaceutical firms as Smith-human behavior, and spatial $323,000 for research on complex Kline Beecham. The system will be designed advanced computin arena.

Over 60 researchers from 10 countries took part in the Third Knowledge Discovery in Databases (KDD) Workshop, held during the Eleventh National Conference on Artificial Intelligence in Washington, D.C. A major trend evident at the workshop was the transition to applications in the core KDD area of discovery of relatively simple patterns in relational databases; the most successful applications are appearing in the areas of greatest need, where the databases are so large that manual analysis is impossible. Progress has been facilitated by the availability of commercial KDD tools for both generic discovery and domain-specific applications such as marketing. At the same time, progress has been slowed by problems such as lack of statistical rigor, overabundance of patterns, and poor integration. Besides applications, the main themes of this workshop were (1) the discovery of dependencies and models and (2) integrated and interactive KDD systems.

The field of knowledge engineering has been one of the most visible successes of AI to date. Knowledge acquisition is the main bottleneck in the knowledge engineer's work. Machine-learning tools have contributed positively to the process of trying to eliminate or open up this bottleneck, but how do we know whether the field is progressing? How can we determine the progress made in any of its branches? How can we be sure of an advance and take advantage of it? This article proposes a benchmark as a classificatory, comparative, and metric criterion for machine-learning tools. The benchmark centers on the knowledge engineering viewpoint, covering some of the characteristics the knowledge engineer wants to find in a machine-learning tool. The proposed model has been applied to a set of machine-learning tools, comparing expected and obtained results. Experimentation validated the model and led to interesting results.

This report summarizes this year's workshop and outlines WET to underwrite and support these workshops. Information Systems is also acknowledged. The Defense Advanced this year's workshop and outlines the philosophy behind this annual event. Computer-Supported Cooperative and present the best research Finally, I would like to thank V. Work gathering, which takes in that has a bearing on the "repersonalization Jagannathan for his great help and everyone from anthropologists to of computing," as Fernando expertise in workshop management futurists, this workshop focuses on flores, founder of Action Technologies, and Mary Carriger for relieving me of hardware and software that enables puts it.