The Resource Description Framework (RDF) is widely used to model web data. The scale and complexity of the modeled data emphasized performance challenges on the RDF-triple stores. Workload adaption is one important strategy to deal with those challenges on the storage level. Current workload-adaption approaches lack the necessary generalization of the problem and only optimize part of the storage layer with the workload (mostly the replication). This creates a big performance gap within other data structures (e.g. indexes and cache) that could heavily benefit from the same workload adaption strategy. Moreover, the workload statistics are built collectively in most of the current approaches. Thus, the analysis process is unaware of whether workloads’ items are old or recent. However, that does not simulate the temporal trends that exist naturally in user queries which causes the analysis process to lag behind the rapid workload development. We present a novel universal adaption approach to the storage management of a distributed RDF store. The system aims to find optimal data assignments to the different indexes, replications, and join cache within the limited storage space. We present a cost model based on the workload that often contains frequent patterns. The workload is dynamically and continuously analyzed to evaluate predefined rules considering the benefits and costs of all options of assigning data to the storage structures. The objective is to reduce query execution time by letting different data containers compete on the limited storage space. By modelling the workload statistics as time series, we can apply well-known smoothing techniques allowing the importance of the workload to decay over time. That allows the universal adaption to stay tuned with potential changes in the workload trends.

Data Science as a multidisciplinary discipline has seen a massive transformation in the direction of operationalisation of analysis workflows. Yet it can be observed that such a workflow consists of potentially many diverse components: like modules in different programming languages, database backends, or web frontends. In order to achieve high efficiency and reproducibility of the analysis, a sufficiently high level of software engineering for the different components as well as an overall software architecture that integrates and automates the different components is needed. For the use case of gene expression analysis, from a software quality point of view we analyze a newly developed web application that allows user-friendly access to the underlying workflow.

In this article we present the implementation and benchmarking of a medical information system on top of a distributed relational database system. We enhanced a distributed database system with the implementation of a clustering (based on similarity of disease terms) that induces a primary horizontal fragmentation of a data table and derived fragmentations of secondary tables. With our clustering-based fragmentation, data locality for similarity-based query answering is ensured so that data do not have to be sent unnecessarily over the network. In our benchmark we show that we achieve a significant efficiency gain when retrieving all relevant related answers.

In this paper we address the topic of identification of cohorts of similar patients in a database of electronic health records. We follow the conjecture that retrieval of similar patients can be supported by an underlying distributed database design. Hence we propose a fragmentation based on partitioning the health records and present a benchmark of two implementation variants in comparison to an off-the-shelf data distribution approach provided by Apache Ignite. While our main use case in this paper is cohort identification, our approach has advantages for taxonomy-based query answering in other (non-medical) domains.

The efficiency of Machine Learning (ML) models has widely been acknowledged in the healthcare area. However, the quality of the underlying medical data is a major challenge when applying ML in medical decision making. In particular, the imbalanced class distribution problem causes the ML model to be biased towards the majority class. Furthermore, the accuracy will be biased, too, which produces the Accuracy Paradox. In this paper, we identify an optimal ML model for predicting mortality risk for Intensive Care Units (ICU) patients. We comprehensively assess an approach that leverages the efficiency of ML ensemble learning (in particular, Gradient Boosting Decision Tree) and clustering-based data sampling to handle the imbalanced data problem that this model faces. We comprehensively compare different competitors (in terms of ML models as well as clustering methods) on a big real-world ICU dataset achieving a maximum area under the curve value of 0.956.

Confidentiality concerns are important in the context of cloud databases. In this paper, the technique of vertical fragmentation is explored to break sensitive associations between columns of several database tables according to confidentiality constraints. By storing insensitive portions of the database at different non-communicating servers it is possible to overcome confidentiality concerns. In addition, visibility constraints and data dependencies are supported. Moreover, to provide some control over the distribution of columns among different servers, novel closeness constraints are introduced. Finding confidentiality-preserving fragmentations is studied in the context of mathematical optimization and a corresponding integer linear program formulation is presented. Benchmarks were performed to evaluate the suitability of our approach.