Robert B. Ross

ProbView: a flexible probabilistic database system

Probability theory is mathematically the best understood paradigm for modeling and manipulating uncertain information. Probabilities of complex events can be computed from those of basic events on which they depend, using any of a number of strategies. Which strategy is appropriate depends very much on the known interdependencies among the events involved. Previous work on probabilistic databases has assumed a fixed and restrictivecombination strategy (e.g., assuming all events are pairwise independent). In this article, we characterize, using postulates, whole classes of strategies for conjunction, disjunction, and negation, meaningful from the viewpoint of probability theory. (1) We propose a probabilistic relational data model and a genericprobabilistic relational algebra that neatly captures various strategiessatisfying the postulates, within a single unified framework. (2) We show that as long as the chosen strategies can be computed in polynomial time, queries in the positive fragment of the probabilistic relational algebra have essentially the same data complexity as classical relational algebra. (3) We establish various containments and equivalences between algebraic expressions, similar in spirit to those in classical algebra. (4) We develop algorithms for maintaining materialized probabilistic views. (5) Based on these ideas, we have developed a prototype probabilistic database system called ProbView on top of Dbase V.0. We validate our complexity results with experiments and show that rewriting certain types of queries to other equivalent forms often yields substantial savings.

Impact: a platform for collaborating agents

The paper discusses the Impact platform (Interactive Maryland Platform) for collaborative agents. Impact agents consist of software code with an associated wrapper that agentizes the code. A set of specialized servers facilitate agent interoperability in an application-independent manner.

Probabilistic temporal databases, I: algebra

Dyreson and Snodgrass have drawn attention to the fact that, in many temporal database applications, there is often uncertainty about the start time of events, the end time of events, and the duration of events. When the granularity of time is small (e.g., milliseconds), a statement such as “Packet p was shipped sometime during the first 5 days of January, 1998” leads to a massive amount of uncertainty (5×24×60×60×1000) possibilities. As noted in Zaniolo et al. [1997], past attempts to deal with uncertainty in databases have been restricted to relatively small amounts of uncertainty in attributes. Dyreson and Snodgrass have taken an important first step towards solving this problem. In this article, we first introduce the syntax of Temporal-Probabilistic (TP) relations and then show how they can be converted to an explicit, significantly more space-consuming form, called Annotated Relations. We then present a theoretical annotated temporal algebra (TATA). Being explicit, TATA is convenient for specifying how the algebraic operations should behave, but is impractical to use because annotated relations are overwhelmingly large. Next, we present a temporal probabilistic algebra (TPA). We show that our definition of the TP-algebra provides a correct implementation of TATA despite the fact that it operates on implicit, succinct TP-relations instead of overwhemingly large annotated relations. Finally, we report on timings for an implementation of the TP-Algebra built on top of ODBC.

Aggregate operators in probabilistic databases

Though extensions to the relational data model have been proposed in order to handle probabilistic information, there has been very little work to date on handling aggregate operators in such databases. In this article, we present a very general notion of an aggregate operator and show how classical aggregation operators (such as COUNT, SUM, etc.) as well as statistical operators (such as percentiles, variance, etc.) are special cases of this general definition. We devise a formal linear programming based semantics for computing aggregates over probabilistic DBMSs, develop algorithms that satisfy this semantics, analyze their complexity, and introduce several families of approximation algorithms that run in polynomial time. We implemented all of these algorithms and tested them on a large set of data to help determine when each one is preferable.

IMPACT: A System for Building Agent Applications

IMPACT (Interactive Maryland Platform for Agents Collaborating Together) provides a platform and environment for agent and software interoperability being developed as a joint, multinational effort with participants from the University of Maryland, the Technische Universität Wien, Bar-Ilan University, the University of Koblenz, and the Universita di Torino. Here, we describe the overall architecture of the IMPACT system, and outline how this architecture (i) allows agents to be developed either from scratch, or by extending legacy code-bases, (ii) allows agents to interact with one another, (iii) allows agents to have a variety of capabilities (reactive, autonomous, intelligent, mobile, replicating) and behaviors, and (iv) how IMPACT provides a variety of infrastructural services that may be used by agents to interact with one another.