Query-driven Data Completeness Management (PhD Thesis)

TL;DR

This thesis develops a metadata-based framework for assessing and inferring data completeness across various data models, including relational, RDF, and spatial data, with applications in decision making and process verification.

Contribution

It introduces formal methods for expressing and inferring data completeness, extending these techniques to null values, RDF, spatial data, and process-aware verification.

Findings

Framework for relational database completeness inference

Extension to RDF and spatial data completeness

Verification techniques for data completeness in processes

Abstract

Knowledge about data completeness is essentially in data-supported decision making. In this thesis we present a framework for metadata-based assessment of database completeness. We discuss how to express information about data completeness and how to use such information to draw conclusions about the completeness of query answers. In particular, we introduce formalisms for stating completeness for parts of relational databases. We then present techniques for drawing inferences between such statements and statements about the completeness of query answers, and show how the techniques can be extended to databases that contain null values. We show that the framework for relational databases can be transferred to RDF data, and that a similar framework can also be applied to spatial data. We also discuss how completeness information can be verified over processes, and introduce a data-aware process model that allows this verification.