Structurally Tractable Uncertain Data

TL;DR

This paper investigates how restricting the structure of uncertain data to tree-like forms can make complex query evaluation tasks computationally feasible, providing new tractability results and a framework for probabilistic reasoning.

Contribution

It introduces tractability results for tree-structured uncertain data and proposes a framework for probabilistic rule reasoning and uncertainty about order.

Findings

Tractability results for tree and tree-like uncertain data

A new representation for uncertainty about order

Analysis of uncertain data conditioned by observations

Abstract

Many data management applications must deal with data which is uncertain, incomplete, or noisy. However, on existing uncertain data representations, we cannot tractably perform the important query evaluation tasks of determining query possibility, certainty, or probability: these problems are hard on arbitrary uncertain input instances. We thus ask whether we could restrict the structure of uncertain data so as to guarantee the tractability of exact query evaluation. We present our tractability results for tree and tree-like uncertain data, and a vision for probabilistic rule reasoning. We also study uncertainty about order, proposing a suitable representation, and study uncertain data conditioned by additional observations.