The task of inferring a set of classes and class descriptions most likely to explain a given data set can be placed on a firm theoretical foundation using Bayesian statistics. Within this framework, and using various mathematical and algorithmic approximations, the AutoClass system searches for the most probable classifications, automatically choosing the number of classes and complexity of class descriptions. Simpler versions of AutoClass have been applied to many large real data sets, have discovered new independently-verified phenomena, and have been released as a robust software package. Recent extensions allow attributes to be selectively correlated within particular classes, and allow classes to inherit, or share, model parameters though a class hierarchy. 1

In this paper, we review AI research on computational discovery and its recent application to the discovery of new scientific knowledge. We characterize five historical stages of the scientific discovery process, which we use as an organizational framework in describing applications. We also identify five distinct steps during which developers or users can influence the behavior of a computational discovery system. Rather than criticizing such intervention, as done in the past, we recommend it as the preferred approach to using discovery software. As evidence for the advantages of such human-computer cooperation, we report seven examples of novel, computer-aided discoveries that have appeared in the scientific literature. We consider briefly the role that humans played in each case, then examine one such interaction in more detail. We close by recommending that future systems provide more explicit support for human intervention in the discovery process. Running head: Computational Sci...

. In this paper, we review AI research on computational discovery and its recent application to the discovery of new scientific knowledge. We characterize five historical stages of the scientific discovery process, which we use as an organizational framework in describing applications. We also identify five distinct steps during which developers or users can influence the behavior of a computational discovery system. Rather than criticizing such intervention, as done in the past, we recommend it as the preferred approach to using discovery software. As evidence for the advantages of such human-computer cooperation, we report seven examples of novel, computer-aided discoveries that have appeared in the scientific literature, along with the role that humans played in each case. We close by recommending that future systems provide more explicit support for human intervention in the discovery process. 1 Introduction The process of scientific discovery has long been viewed as the pinnacle ...

Summary. Clustering algorithms can identify groups in large data sets, such as star catalogs and hyperspectral images. In general, clustering methods cannot analyze items that have missing data values. Common solutions either fill in the missing values (imputation) or ignore the missing data (marginalization). Imputed values are treated as just as reliable as the truly observed data, but they are only as good as the assumptions used to create them. In contrast, we present a method for encoding partially observed features as a set of supplemental soft constraints and introduce the KSC algorithm, which incorporates constraints into the clustering process. In experiments on artificial data and data from the Sloan Digital Sky Survey, we show that soft constraints are an effective way to enable clustering with missing values. 1