Elena Vasilyeva

Top-k Differential Queries in Graph Databases

The sheer volume as well as the schema complexity of today’s graph databases impede the users in formulating queries against these databases and often cause queries to “fail” by delivering empty answers. To support users in such situations, the concept of differential queries can be used to bridge the gap between an unexpected result (e.g. an empty result set) and the query intention of users. These queries deliver missing parts of a query graph and, therefore, work with such scenarios that require users to specify a query graph. Based on the discovered information about a missing query subgraph, users may understand which vertices and edges are the reasons for queries that unexpectedly return empty answers, and thus can reformulate the queries if needed. A study showed that the result sets of differential queries are often too large to be manually introspected by users and thus a reduction of the number of results and their ranking is required. To address these issues, we extend the concept of differential queries and introduce top-k differential queries that calculate the ranking based on users’ preferences and therefore significantly support the users’ understanding of query database management systems. The idea consists of assigning relevance weights to vertices or edges of a query graph by users that steer the graph search and are used in the scoring function for top-k differential results. Along with the novel concept of the top-k differential queries, we further propose a strategy for propagating relevance weights and we model the search along the most relevant paths.

Relaxation of subgraph queries delivering empty results

Graph databases with the property graph model are used in multiple domains including social networks, biology, and data integration. They provide schema-flexible storage for data of a different degree of a structure and support complex, expressive queries such as subgraph isomorphism queries. The exibility and expressiveness of graph databases make it difficult for the users to express queries correctly and can lead to unexpected query results, e.g. empty results. Therefore, we propose a relaxation approach for subgraph isomorphism queries that is able to automatically rewrite a graph query, such that the rewritten query is similar to the original query and returns a non-empty result set. In detail, we present relaxation operations applicable to a query, cardinality estimation heuristics, and strategies for prioritizing graph query elements to be relaxed. To determine the similarity between the original query and its relaxed variants, we propose a novel cardinality-based graph edit distance. The feasibility of our approach is shown by using real-world queries from the DBpedia query log.

Efficient integration of external information into forecast models from the energy domain

Forecasting is an important analysis technique to support decisions and functionalities in many application domains. While the employed statistical models often provide a sufficient accuracy, recent developments pose new challenges to the forecasting process. Typically the available time for estimating the forecast models and providing accurate predictions is significantly decreasing. This is especially an issue in the energy domain, where forecast models often consider external influences to provide a high accuracy. As a result, these models exhibit a higher number of parameters, resulting in increased estimation efforts. Also, in the energy domain new measurements are constantly appended to the time series, requiring a continuous adaptation of the models to new developments. This typically involves a parameter re-estimation, which is often almost as expensive as the initial estimation, conflicting with the requirement for fast forecast computation. To address these challenges, we present a framework that allows a more efficient integration of external information. First, external information are handled in a separate model, because their linear and non-linear relationships are more stable and thus, they can be excluded from most forecast model adaptations. Second, we directly optimize the separate model using feature selection and dimension reduction techniques. Our evaluation shows that our approach allows an efficient integration of external information and thus, an increased forecasting accuracy, while reducing the re-estimation efforts.

Why-query support in graph databases

Graph databases implementing a property graph model allow storing of heterogeneous information in the form of a graph and support complex graph-specific queries like shortest path, pattern matching, etc. Their flexibility and a rich spectrum of supported queries make it difficult for a user to create correct queries. As a consequence, a user can get unexpected results like too many, too few, or even empty answers. This research aims at providing a basic debugging functionality to a user in order to discover the reasons of a failure and to fix a query. The main goals of this thesis include (1) studying the reasons of a failure in terms of a graph with the focus on cardinality-based problems like too few, too many, and empty results; (2) developing methods for query refinement in order to derive expected answers with considering specifics of a property graph model, and (3) proposing a set of strategies for integrating user intention into the debugging process.

Robust Cardinality Estimation for Subgraph Isomorphism Queries on Property Graphs

With an increasing popularity of graph data and graph processing systems, the need of efficient graph processing and graph query optimization becomes more important. Subgraph isomorphism queries, one of the fundamental graph query types, rely on an accurate cardinality estimation of a single edge of a pattern for efficient query processing. State of the art approaches do not consider two important aspects for cardinality estimation of graph queries on property graphs: the existence of nodes with a high outdegree and functional dependencies between attributes. In this paper we focus on these two challenges and integrate the detection of high-outdegree nodes and functional dependency analysis into the cardinality estimation. We evaluate our approach on two real data sets and compare it against a state-of-the-art query optimizer for property graphs as implemented in Neo4j.