Why-query support in graph databases

Date(s) - 04/02/2016
1:00 am - 2:00 am



Graph database systems based on the property graph model are used in multiple domains including: social networks, biology, logistics, and data integration. They provide schema-flexible storage for data of a different degree of a structure and support complex, expressive queries like shortest path calculation or subgraph isomorphism queries. Similar to a relational database, the result of a query in a graph database can be unexpected, because no answer, too few, or too many results can be returned. The flexibility and expressiveness of a graph database increase the likelihood of these situations and at the same time make it even harder for the user to resolve such issues manually.
In traditional database research, such unexpected answers are tackled by the concept of Why-queries. Based on an original query or data, they give an explanation why an expected answer was not delivered, and propose new refined queries delivering the expected result sets.
The goal of this thesis is to transfer existing Why-query approaches to graph databases and extend the concept to the underlying graph data model. Therefore, in the talk we will focus on how to generate explanations for unexpected results of pattern matching queries. Especially, two kinds of explanations will be presented: subgraph-based and modification-based approaches. In addition, user integration techniques will be discussed, which allow to generate user-specific explanations.


Speaker: Elena Vasilyeva

Elena Vasilyeva is a PhD student at Technische Universität Dresden, Germany. She obtained her master degree at the Petrozavodsk State University (Russia) and Technische Universität Dresden (Germany). She writes her PhD thesis about “Why-query support in graph databases” in the cooperation with SAP SE Germany. Her research interests include efficient graph processing, pattern matching in graph databases, and time series analysis.