Durch die wachsende Bedeutung des Onlinehandels und der Zunahme an Benutzer-generierten Inhalten werden neue Ansätze benötigt, um Konsumenten bei ihrer Entscheidungsfindung zu unterstützen. Wie Studien zeigen, werden im Onlinehandel häufig Kundenrezensionen und Gesamtbewertungen eingesetzt. Allerdings sind diese beiden Werkzeuge für die Entscheidungsfindung von Konsumenten nur begrenzt hilfreich. Gesamtbewertungen zeigen zwar eine oberflächliche Zufriedenheit der Kunden, geben jedoch keine Auskunft über die Bewertung bestimmter Produktaspekte, z.B. den Tragekomfort von Kopfhörern. Diese Aspekte werden von Kunden häufig in Rezensionen beschrieben, welche jedoch aufgrund ihrer unstrukturierten Weise nicht automatisiert aufbereitet werden können. Konsumenten sind daher gezwungen Rezensionen zu lesen und die darin diskutierten Merkmale manuell zu extrahieren. Die vorliegende Arbeit leistet mehrere Beiträge zur Adressierung des oben genannten Problems und beschäftigt sich dabei mit der Konzeptionierung, Evaluierung und Dienst-orientierten Bereitstellung einer interaktiven Entscheidungshilfe für den E-Commerce. Zunächst wird anhand einer empirischen Untersuchung der umsatzstärksten Onlineshops aus Deutschland der aktuelle Einsatz von Social Media Features analysiert. Dabei zeigt sich, dass die o.g. Problematik von keinem untersuchten Onlineshop adressiert wird. Ein weiterer Beitrag ist der Entwurf sowie die prototypische Implementierung einer interaktiven Entscheidungshilfe mit der Bezeichnung Rated Tags. Rated Tags erlaubt die Benutzer-generierte Definition von bewertbaren Schlagwörtern (Tags) und kombiniert dabei Methoden aus den Bereichen Social Tagging und Bewertungssysteme. Eine nachfolgende Evaluierung des Konzepts im Rahmen einer Anwenderstudie zeigt, dass der Einsatz von Rated Tags die Entscheidungsqualität verbessern sowie den Entscheidungsaufwand von Konsumenten reduzieren kann. Zur Optimierung des Lösungsansatzes wird dann ein Ensemble-Klassifikator aus dem Bereich des überwachten Lernens zur semiautomatisierten Vereinheitlichung von semantisch ähnlichen Tags entworfen, prototypisch implementiert und evaluiert. Die Ergebnisse der Evaluierung zeigen, dass die Leistung des Klassifikators den aktuellen Stand der Technik übersteigt. Als Abschluss der Arbeit wird ein Modell mit der Bezeichnung Rated Tags as a Service vorgestellt, welches die Service-orientierte Bereitstellung des Rated Tags-Ansatzes für Onlineshops oder Bewertungsportale beschreibt.

Scientific data life cycles define how data is created, handled, accessed, and analyzed by users. Such data life cycles become increasingly sophisticated as the sciences they deal with become more and more demanding and complex with the coming advent of exascale data and computing. The overarching data life cycle management background includes multiple abstraction categories with data sources, data and metadata management, computing and workflow management, security, data sinks, and methods on how to enable utilization. Challenges in this context are manifold. One is to hide the complexity from the user and to enable seamlessness in using resources to usability and efficiency. Another one is to enable generic metadata management that is not restricted to one use case but can be adapted with limited effort to further ones. Metadata management is essential to enable scientists to save time by avoiding the need for manually keeping track of data, meaning for example by its content and location. As the number of files grows into the millions, managing data without metadata becomes increasingly difficult. Thus, the solution is to employ metadata management to enable the organization of data based on information about it. Previously, use cases tended to only support highly specific or no metadata management at all. Now, a generic metadata management concept is available that can be used to efficiently integrate metadata capabilities with use cases. The concept was implemented within the MoSGrid data life cycle that enables molecular simulations on distributed HPC-enabled data and computing infrastructures. The implementation enables easy-to-use and effective metadata management. Automated extraction, annotation, and indexing of metadata was designed, developed, integrated, and search capabilities provided via a seamless user interface. Further analysis runs can be directly started based on search results. A complete evaluation of the concept both in general and along the example implementation is presented. In conclusion, generic metadata management concept advances the state of the art in scientific date life cycle management.

Integration flows are increasingly used to specify and execute data-intensive integration tasks between heterogeneous systems and applications. There are many different application areas such as real-time ETL and data synchronization between operational systems. For the reasons of an increasing amount of data, highly distributed IT infrastructures, and high requirements for data consistency and up-to-dateness of query results, many instances of integration flows are executed over time. Due to this high load and blocking synchronous source systems, the performance of the central integration platform is crucial for an IT infrastructure. To tackle these high performance requirements, we introduce the concept of cost-based optimization of imperative integration flows that relies on incremental statistics maintenance and inter-instance plan re-optimization. As a foundation, we introduce the concept of periodical re-optimization including novel cost-based optimization techniques that are tailor-made for integration flows. Furthermore, we refine the periodical re-optimization to on-demand re-optimization in order to overcome the problems of many unnecessary re-optimization steps and adaptation delays, where we miss optimization opportunities. This approach ensures low optimization overhead and fast workload adaptation.

Data streams in the form of potentially unbounded sequences of tuples arise naturally in a large variety of domains including finance markets, sensor networks, social media, and network traffic management. The increasing number of applications that require processing data streams with high throughput and low latency have promoted the development of data stream processing systems (DSPS). A DSPS processes data streams with continuous queries, which are issued once and return query results to users continuously as new tuples arrive. For stream-based applications, both the query-execution performance (in terms of, e.g., throughput and end-to-end latency) and the quality of produced query results (in terms of, e.g., accuracy and completeness) are important. However, a DSPS often needs to make tradeoffs between these two requirements, either because of the data imperfection within the streams, or because of the limited computation capacity of the DSPS itself. Performance versus result-quality tradeoffs caused by data imperfection are inevitable, because the quality of the incoming data is beyond the control of a DSPS, whereas tradeoffs caused by system limitations can be alleviated—even erased—by enhancing the DSPS itself. This dissertation seeks to advance the state of the art on handling the performance versus result-quality tradeoffs in data stream processing caused by the above two aspects of reasons. For tradeoffs caused by data imperfection, this dissertation focuses on the typical data-imperfection problem of stream disorder and proposes the concept of quality-driven disorder handling (QDDH). QDDH enables a DSPS to make flexible and user-configurable tradeoffs between the end-to-end latency and the query-result quality when dealing with stream disorder. Moreover, compared to existing disorder handling approaches, QDDH can significantly reduce the end-to-end latency, and at the same time provide users with desired query-result quality. In this dissertation, a generic buffer-based QDDH framework and three instantiations of the generic framework for distinct query types are presented. For tradeoffs caused by system limitations, this dissertation proposes a system-enhancement approach that combines the row-oriented and the column-oriented data layout and processing techniques in data stream processing to improve the throughput. To fully exploit the potential of such hybrid execution of continuous queries, a static, cost-based query optimizer is introduced. The optimizer works at the operator level and takes the unique property of execution plans of continuous queries—feasibility—into account.

Business Intelligence (BI) enables organizations to collect and analyze internal and external business data to generate knowledge and business value, and provide decision support at the strategic, tactical, and operational levels. The consolidation of data coming from many sources as a result of managerial and operational business processes, usually referred to as Extract-Transform-Load (ETL) is itself a statically defined process and knowledge workers have little to no control over the characteristics of the presentable data to which they have access. There are two main reasons that dictate the reassessment of this stiff approach in context of modern business environments. The first reason is that the service-oriented nature of today’s business combined with the increasing volume of available data make it impossible for an organization to proactively design efficient data management processes. The second reason is that enterprises can benefit significantly from analyzing the behavior of their business processes fostering their optimization. Hence, we took a first step towards quality-aware ETL process design automation by defining through a systematic literature review a set of ETL process quality characteristics and the relationships between them, as well as by providing quantitative measures for each characteristic. Subsequently, we produced a model that represents ETL process quality characteristics and the dependencies among them and we showcased through the application of a Goal Model with quantitative components (i.e., indicators) how our model can provide the basis for subsequent analysis to reason and make informed ETL design decisions. In addition, we introduced our holistic view for a quality-aware design of ETL processes by presenting a framework for user-centered declarative ETL. This included the definition of an architecture and methodology for the rapid, incremental, qualitative improvement of ETL process models, promoting automation and reducing complexity, as well as a clear separation of business users and IT roles where each user is presented with appropriate views and assigned with fitting tasks. In this direction, we built a tool —POIESIS— which facilitates incremental, quantitative improvement of ETL process models with users being the key participants through well-defined collaborative interfaces. When it comes to evaluating different quality characteristics of the ETL process design, we proposed an automated data generation framework for evaluating ETL processes (i.e., Bijoux). To this end, we classified the operations based on the part of input data they access for processing, which facilitated Bijoux during data generation processes both for identifying the constraints that specific operation semantics imply over input data, as well as for deciding at which level the data should be generated (e.g., single field, single tuple, complete dataset). Bijoux offers data generation capabilities in a modular and configurable manner, which can be used to evaluate the quality of different parts of an ETL process. Moreover, we introduced a methodology that can apply to concrete contexts, building a repository of patterns and rules. This generated knowledge base can be used during the design and maintenance phases of ETL processes, automatically exposing understandable conceptual representations of the processes and providing useful insight for design decisions. Collectively, these contributions have raised the level of abstraction of ETL process components, revealing their quality characteristics in a granular level and allowing for evaluation and automated (re-)design, taking under consideration business users’ quality goals.

Modern agile software development methods allow to continuously evolve software systems by easily adding new features, fixing bugs, and adapting the software to changing requirements and conditions while it is continuously used by the users. A major obstacle in the agile evolution is the underlying database that persists the software system’s data from day one on. Hence, evolving the database schema requires to evolve the existing data accordingly—at this point, the currently established solutions are very expensive and error-prone and far from agile. In this thesis, we present InVerDa, a multi-schema-version database system to facilitate agile database development. Multi-schema-version database systems provide multiple schema versions within the same database, where each schema version itself behaves like a regular single-schema database. Creating new schema versions is very simple to provide the desired agility for database development. All created schema versions can co-exist and write operations are immediately propagated between schema versions with a best-effort strategy. Developers do not have to implement the propagation logic of data accesses between schema versions by hand, but InVerDa automatically generates it. To facilitate multi-schema-version database systems, we equip developers with a relational complete and bidirectional database evolution language (BiDEL) that allows to easily evolve existing schema versions to new ones. BiDEL allows to express the evolution of both the schema and the data both forwards and backwards in intuitive and consistent operations; the BiDEL evolution scripts are orders of magnitude shorter than implementing the same behavior with standard SQL and are even less likely to be erroneous, since they describe a developer’s intention of the evolution exclusively on the level of tables without further technical details. Having the developers’ intentions explicitly given in the BiDEL scripts further allows to create a new schema version by merging already existing ones. Having multiple co-existing schema versions in one database raises the need for a sophisticated physical materialization. Multi-schema-version database systems provide full data independence, hence the database administrator can choose a feasible materialization, whereby the multi-schema-version database system internally ensures that no data is lost. The search space of possible materializations can grow exponentially with the number of schema versions. Therefore, we present an adviser that releases the database administrator from diving into the complex performance characteristics of multi-schema-version database systems and merely proposes an optimized materialization for a given workload within seconds. Optimized materializations have shown to improve the performance for a given workload by orders of magnitude. We formally guarantee data independence for multi-schema-version database systems. To this end, we show that every single schema version behaves like a regular single-schema database independent of the chosen physical materialization. This important guarantee allows to easily evolve and access the database in agile software development—all the important features of relational databases, such as transaction guarantees, are preserved. To the best of our knowledge, we are the first to realize such a multi-schema-version database system that allows agile evolution of production databases with full support of co-existing schema versions and formally guaranteed data independence.

In the last few decades, database management systems became powerful tools for storing large amount of data and executing complex queries over them. In addition to extended functionality, novel types of databases appear like triple stores, distributed databases, etc. Graph databases implementing the property-graph model belong to this development branch and provide a new way for storing and processing data in the form of a graph with nodes representing some entities and edges describing connections between them. This consideration makes them suitable for keeping data without a rigid schema for use cases like social-network processing or data integration. In addition to a flexible storage, graph databases provide new querying possibilities in the form of path queries, detection of connected components, pattern matching, etc. However, the schema flexibility and graph queries come with additional costs. With limited knowledge about data and little experience in constructing the complex queries, users can create such ones, which deliver unexpected results. Forced to debug queries manually and overwhelmed by the amount of query constraints, users can get frustrated by using graph databases. What is really needed, is to improve usability of graph databases by providing debugging and explaining functionality for such situations. We have to assist users in the discovery of what were the reasons of unexpected results and what can be done in order to fix them. The unexpectedness of result sets can be expressed in terms of their size or content. In the first case, users have to solve the empty-answer, too-many-, or too-few-answers problems. In the second case, users care about the result content and miss some expected answers or wonder about presence of some unexpected ones. Considering the typical problems of receiving no or too many results by querying graph databases, in this thesis we focus on investigating the problems of the first group, whose solutions are usually represented by why-empty, why-so-few, and why-so-many queries. Our objective is to extend graph databases with debugging functionality in the form of why-queries for unexpected query results on the example of pattern matching queries, which are one of general graph-query types. We present a comprehensive analysis of existing debugging tools in the state-of-the-art research and identify their common properties. From them, we formulate the following features of why-queries, which we discuss in this thesis, namely: holistic support of different cardinality-based problems, explanation of unexpected results and query reformulation, comprehensive analysis of explanations, and non-intrusive user integration. To support different cardinality-based problems, we develop methods for explaining no, too few, and too many results. To cover different kinds of explanations, we present two types: subgraph- and modification-based explanations. The first type identifies the reasons of unexpectedness in terms of query subgraphs and delivers differential graphs as answers. The second one reformulates queries in such a way that they produce better results. Considering graph queries to be complex structures with multiple constraints, we investigate different ways of generating explanations starting from the most general one that considers only a query topology through coarse-grained rewriting up to fine-grained modification that allows fine changes of predicates and topology. To provide a comprehensive analysis of explanations, we propose to compare them on three levels including a syntactic description, a content, and a size of a result set. In order to deliver user-aware explanations, we discuss two models for non-intrusive user integration in the generation process. With the techniques proposed in this thesis, we are able to provide fundamentals for debugging of pattern-matching queries, which deliver no, too few, or too many results, in graph databases implementing the property-graph model.

Over the past decade, the public awareness and availability as well as methods for the creation and use of spatial data on the Web have steadily increased. Besides the establishment of governmental Spatial Data Infrastructures (SDIs), numerous volunteered and commercial initiatives had a major impact on that development. Nevertheless, data isolation still poses a major challenge. Whereas the majority of approaches focuses on data provision, means to dynamically link and combine spatial data from distributed, often heterogeneous data sources in an ad hoc manner are still very limited. However, such capabilities are essential to support and enhance information retrieval for comprehensive spatial decision making. To facilitate spatial data fusion in current SDIs, this thesis has two main objectives. First, it focuses on the conceptualization of a service-based fusion process to functionally extend current SDI and to allow for the combination of spatial data from different spatial data services. It mainly addresses the decomposition of the fusion process into well-defined and reusable functional building blocks and their implementation as services, which can be used to dynamically compose meaningful application-specific processing workflows. Moreover, geoprocessing patterns, i.e. service chains that are commonly used to solve certain fusion subtasks, are designed to simplify and automate workflow composition. Second, the thesis deals with the determination, description and exploitation of spatial data relations, which play a decisive role for spatial data fusion. The approach adopted is based on the Linked Data paradigm and therefore bridges SDI and Semantic Web developments. Whereas the original spatial data remains within SDI structures, relations between those sources can be used to infer spatial information by means of Semantic Web standards and software tools. A number of use cases were developed, implemented and evaluated to underpin the proposed concepts. Particular emphasis was put on the use of established open standards to realize an interoperable, transparent and extensible spatial data fusion process and to support the formalized description of spatial data relations. The developed software, which is based on a modular architecture, is available online as open source. It allows for the development and seamless integration of new functionality as well as the use of external data and processing services during workflow composition on the Web. / Die Entwicklung des Internet im Laufe des letzten Jahrzehnts hat die Verfügbarkeit und öffentliche Wahrnehmung von Geodaten, sowie Möglichkeiten zu deren Erfassung und Nutzung, wesentlich verbessert. Dies liegt sowohl an der Etablierung amtlicher Geodateninfrastrukturen (GDI), als auch an der steigenden Anzahl Communitybasierter und kommerzieller Angebote. Da der Fokus zumeist auf der Bereitstellung von Geodaten liegt, gibt es jedoch kaum Möglichkeiten die Menge an, über das Internet verteilten, Datensätzen ad hoc zu verlinken und zusammenzuführen, was mitunter zur Isolation von Geodatenbeständen führt. Möglichkeiten zu deren Fusion sind allerdings essentiell, um Informationen zur Entscheidungsunterstützung in Bezug auf raum-zeitliche Fragestellungen zu extrahieren. Um eine ad hoc Fusion von Geodaten im Internet zu ermöglichen, behandelt diese Arbeit zwei Themenschwerpunkte. Zunächst wird eine dienstebasierten Umsetzung des Fusionsprozesses konzipiert, um bestehende GDI funktional zu erweitern. Dafür werden wohldefinierte, wiederverwendbare Funktionsblöcke beschrieben und über standardisierte Diensteschnittstellen bereitgestellt. Dies ermöglicht eine dynamische Komposition anwendungsbezogener Fusionsprozesse über das Internet. Des weiteren werden Geoprozessierungspatterns definiert, um populäre und häufig eingesetzte Diensteketten zur Bewältigung bestimmter Teilaufgaben der Geodatenfusion zu beschreiben und die Komposition und Automatisierung von Fusionsprozessen zu vereinfachen. Als zweiten Schwerpunkt beschäftigt sich die Arbeit mit der Frage, wie Relationen zwischen Geodatenbeständen im Internet erstellt, beschrieben und genutzt werden können. Der gewählte Ansatz basiert auf Linked Data Prinzipien und schlägt eine Brücke zwischen diensteorientierten GDI und dem Semantic Web. Während somit Geodaten in bestehenden GDI verbleiben, können Werkzeuge und Standards des Semantic Web genutzt werden, um Informationen aus den ermittelten Geodatenrelationen abzuleiten. Zur Überprüfung der entwickelten Konzepte wurde eine Reihe von Anwendungsfällen konzipiert und mit Hilfe einer prototypischen Implementierung umgesetzt und anschließend evaluiert. Der Schwerpunkt lag dabei auf einer interoperablen, transparenten und erweiterbaren Umsetzung dienstebasierter Fusionsprozesse, sowie einer formalisierten Beschreibung von Datenrelationen, unter Nutzung offener und etablierter Standards. Die Software folgt einer modularen Struktur und ist als Open Source frei verfügbar. Sie erlaubt sowohl die Entwicklung neuer Funktionalität durch Entwickler als auch die Einbindung existierender Daten- und Prozessierungsdienste während der Komposition eines Fusionsprozesses.

Role-based modeling has been proposed in 1977 by Charles W. Bachman, as a means to model complex and dynamic domains, because roles are able to capture both context-dependent and collaborative behavior of objects. Consequently, they were introduced in various fields of research ranging from data modeling via conceptual modeling through to programming languages. More importantly, because current software systems are characterized by increased complexity and context-dependence, there is a strong demand for new concepts beyond object-oriented design. Although mainstream modeling languages, i.e., Entity-Relationship Model, Unified Modeling Language, are good at capturing a system's structure, they lack ways to model the system's behavior, as it dynamically emerges through collaborating objects. In turn, roles are a natural concept capturing the behavior of participants in a collaboration. Moreover, roles permit the specification of interactions independent from the interacting objects. Similarly, more recent approaches use roles to capture context-dependent properties of objects. The notion of roles can help to tame the increased complexity and context-dependence. Despite all that, these years of research had almost no influence on current software development practice. To make things worse, until now there is no common understanding of roles in the research community and no approach fully incorporates both the context-dependent and the relational nature of roles. In this thesis, I will devise a formal model for a family of role-based modeling languages to capture the various notions of roles. Together with a software product line of Role Modeling Editors, this, in turn, enables the generation of a role-based language family for Role-based Software Infrastructures (RoSI).

The Web provides a platform for people to share their data, leading to an abundance of accessible information. In recent years, significant research effort has been directed especially at tables on the Web, which form a rich resource for factual and relational data. Applications such as fact search and knowledge base construction benefit from this data, as it is often less ambiguous than unstructured text. However, many traditional information extraction and retrieval techniques are not well suited for Web tables, as they generally do not consider the role of the table structure in reflecting the semantics of the content. Tables provide a compact representation of similarly structured data. Yet, on the Web, tables are very heterogeneous, often with ambiguous semantics and inconsistencies in the quality of the data. Consequently, recognizing the structure and inferring the semantics of these tables is a challenging task that requires a designated table recovery and understanding process. In the literature, many important contributions have been made to implement such a table understanding process that specifically targets Web tables, addressing tasks such as table detection or header recovery. However, the precision and coverage of the data extracted from Web tables is often still quite limited. Due to the complexity of Web table understanding, many techniques developed so far make simplifying assumptions about the table layout or content to limit the amount of contributing factors that must be considered. Thanks to these assumptions, many sub-tasks become manageable. However, the resulting algorithms and techniques often have a limited scope, leading to imprecise or inaccurate results when applied to tables that do not conform to these assumptions. In this thesis, our objective is to extend the Web table understanding process with techniques that enable some of these assumptions to be relaxed, thus improving the scope and accuracy. We have conducted a comprehensive analysis of tables available on the Web to examine the characteristic features of these tables, but also identify unique challenges that arise from these characteristics in the table understanding process. To extend the scope of the table understanding process, we introduce extensions to the sub-tasks of table classification and conceptualization. First, we review various table layouts and evaluate alternative approaches to incorporate layout classification into the process. Instead of assuming a single, uniform layout across all tables, recognizing different table layouts enables a wide range of tables to be analyzed in a more accurate and systematic fashion. In addition to the layout, we also consider the conceptual level. To relax the single concept assumption, which expects all attributes in a table to describe the same semantic concept, we propose a semantic normalization approach. By decomposing multi-concept tables into several single-concept tables, we further extend the range of Web tables that can be processed correctly, enabling existing techniques to be applied without significant changes. Furthermore, we address the quality of data extracted from Web tables, by studying the role of context information. Supplementary information from the context is often required to correctly understand the table content, however, the verbosity of the surrounding text can also mislead any table relevance decisions. We first propose a selection algorithm to evaluate the relevance of context information with respect to the table content in order to reduce the noise. Then, we introduce a set of extraction techniques to recover attribute-specific information from the relevant context in order to provide a richer description of the table content. With the extensions proposed in this thesis, we increase the scope and accuracy of Web table understanding, leading to a better utilization of the information contained in tables on the Web.