Refine
Year of publication
Document Type
- Doctoral Thesis (226)
- Article (27)
- Preprint (10)
- Conference Proceeding (9)
- Report (4)
- Book (3)
- Part of Periodical (3)
- Master's Thesis (1)
- Other (1)
Language
- English (284) (remove)
Has Fulltext
- yes (284)
Keywords
- Computersicherheit (8)
- Maßtheorie (8)
- Graphenzeichnen (6)
- Iteriertes Funktionensystem (5)
- Marketing (5)
- Multimedia (5)
- Software Engineering (5)
- Information Retrieval (4)
- Kryptologie (4)
- Modellierung (4)
Institute
- Fakultät für Informatik und Mathematik (100)
- Wirtschaftswissenschaftliche Fakultät (52)
- Mitarbeiter Lehrstuhl/Einrichtung der Fakultät für Informatik und Mathematik (47)
- Philosophische Fakultät (36)
- Mitarbeiter Lehrstuhl/Einrichtung der Wirtschaftswissenschaftlichen Fakultät (9)
- Sonstiger Autor der Fakultät für Informatik und Mathematik (9)
- Sozial- und Bildungswissenschaftliche Fakultät (8)
- Philosophische Fakultät / Südostasienkunde (6)
- Juristische Fakultät (5)
- Sonstiger Autor der Wirtschaftswissenschaftlichen Fakultät (5)
Entity disambiguation is the task of mapping ambiguous terms in natural-language text to its entities in a knowledge base. It finds its application in the extraction of structured data in RDF (Resource Description Framework) from textual documents, but equally so in facilitating artificial intelligence applications, such as Semantic Search, Reasoning and Question & Answering. We propose a new collective, graph-based disambiguation algorithm utilizing semantic entity and document embeddings for robust entity disambiguation. Robust thereby refers to the property of achieving better than state-of-the-art results over a wide range of very different data sets. Our approach is also able to abstain if no appropriate entity can be found for a specific surface form. Our evaluation shows, that our approach achieves significantly (>5%) better results than all other publicly available disambiguation algorithms on 7 of 9 datasets without data set specific tuning. Moreover, we discuss the influence of the quality of the knowledge base on the disambiguation accuracy and indicate that our algorithm achieves better results than non-publicly available state-of-the-art algorithms.
Entity Linking is the task of mapping terms in arbitrary documents to entities in a knowledge base by identifying the correct semantic meaning. It is applied in the extraction of structured data in RDF (Resource Description Framework) from textual documents, but equally so in facilitating artificial intelligence applications, such as Semantic Search, Reasoning and Question and Answering. Most existing Entity Linking systems were optimized for specific domains (e.g., general domain, biomedical domain), knowledge base types (e.g., DBpedia, Wikipedia), or document structures (e.g., tables) and types (e.g., news articles, tweets). This led to very specialized systems that lack robustness and are only applicable for very specific tasks. In this regard, this work focuses on the research and development of a robust Entity Linking system in terms of domains, knowledge base types, and document structures and types.
To create a robust Entity Linking system, we first analyze the following three crucial components of an Entity Linking algorithm in terms of robustness criteria: (i) the underlying knowledge base, (ii) the entity relatedness measure, and (iii) the textual context matching technique. Based on the analyzed components, our scientific contributions are three-fold. First, we show that a federated approach leveraging knowledge from various knowledge base types can significantly improve robustness in Entity Linking systems. Second, we propose a new state-of-the-art, robust entity relatedness measure for topical coherence computation based on semantic entity embeddings. Third, we present the neural-network-based approach Doc2Vec as a textual context matching technique for robust Entity Linking.
Based on our previous findings and outcomes, our main contribution in this work is DoSeR (Disambiguation of Semantic Resources). DoSeR is a robust, knowledge-base-agnostic Entity Linking framework that extracts relevant entity information from multiple knowledge bases in a fully automatic way. The integrated algorithm represents a collective, graph-based approach that utilizes semantic entity and document embeddings for entity relatedness and textual context matching computation. Our evaluation shows, that DoSeR achieves state-of-the-art results over a wide range of different document structures (e.g., tables), document types (e.g., news documents) and domains (e.g., general domain, biomedical domain). In this context, DoSeR outperforms all other (publicly available) Entity Linking algorithms on most data sets.
The well-founded semantics has been accepted as the most relevant semantics for logic-based information systems. In this dissertation a framework based on a set of program transformations is presented that generalizes all major computation approaches for the well-founded semantics using a common data structure and provides a common language to describe their evaluation strategy. This rewriting system gives the formal background to analyze and combine different evaluation strategies in a common framework, or to design new algorithms and prove the correctness of its implementations at a high level just by changing the order of program transformations.
Data has become a necessary resource for firm operations in the modern digital world, explaining their growing data gathering efforts. Due to this development, consumers are confronted with decisions to disclose personal data on a daily basis, and have become increasingly intentional about data sharing. While this reluctance to disclose personal data poses challenges for firms, at the same time, it also creates new opportunities for improving privacy-related interactions with customers. This dissertation advocates for a more holistic perspective on consumers’ privacy-related decision-making and introduces the consumer privacy journey consisting of three subsequent phases: pre data disclosure, data disclosure, post data disclosure. In three independent essays, I stress the importance of investigating data requests (i.e., the first step of this journey) as they represent a largely neglected, yet, potentially powerful means to influence consumers’ decision-making and decision-evaluation processes. Based on dual-processing models of decision-making, this dissertation focuses on both consumers’ cognitive and affective evaluations of privacy-related information: First, Essay 1 offers novel conceptualizations and operationalizations of consumers’ perceived behavioral control over personal data (i.e., cognitive processing) in the context of Artificial Intelligence (AI)-based data disclosure processes. Next, Essay 2 examines consumers’ cognitive and affective processing of a data request that entails relevance arguments as well as relevance-illustrating game elements. Finally, Essay 3 categorizes affective cues that trigger consumers’ affective processing of a data request and proposes that such cues need to fit with a specific data disclosure situation to foster long-term decision satisfaction. Collectively, my findings provide research and practice with new insights into consumers’ privacy perceptions and behaviors, which are particularly valuable in the context of complex, new (technology-enabled) data disclosure situations.
Governments around the world currently focus on shaping the digital economy. Particular attention is paid to Internet platforms, Internet infrastructure and data as essential components of the digital economy. The three studies in this thesis contribute to the understanding of the behavior of firms in each of these domains and derive insights for future regulations and business projects.
The first study deals with the ranking of content on Internet platforms and how it affects the incentives of content providers to invest in content quality. The focus of the study is on sponsored ranking and organic ranking, but the case that a vertically integrated content provider is favored by an Internet platform is also taken into account. Using a game theoretic model, it is shown that there is no ranking design that strictly leads to more investment compared to the other designs. It is also shown that the Internet platform usually chooses the type of ranking that, from the perspective of the Internet platform and consumers, yields the best expected overall content quality. The second study deals with the incentive of Internet service providers to throttle specific Internet content. The key finding is that Internet service providers use this instrument to utilize the capacity of their telecommunications network more efficiently. This leads not only to more benefits for Internet users, but also to a higher incentive to invest in network capacity due to better monetization. The third study examines the circumstances under which firms are willing to share data with other firms. By means of an economic laboratory experiment, it is shown that more data is shared if the firms have control over who exactly they share data with. Thus, for example, data pools that grant unrestricted data access to all participating firms can be expected to perform worse than data pools that give their participating firms control over with whom their uploaded data is shared. In addition, the third study finds that established relationships are characterized by more data sharing and less volatility in the amount of shared data than new relationships. The study concludes that data sharing projects should not be expected to work optimally right away.
In summary, the studies in this thesis identify a number of costs that may arise when digital firms' choice is restricted by regulation or design. The ability of Internet service providers to throttle certain content and the ability of Internet platforms to choose the ranking design are usually used in the best interests of consumers. Data sharing also works best when firms are free to decide who gets their data.
Die Forschung zu Lehrkrafturteilen hat in den letzten drei Jahrzehnten beträchtliche Fortschritte gemacht. Die Bedeutung des Lehrkrafturteils und die Variabilität in der Urteilsgenauigkeit erfordern eine eingehendere Untersuchung. Basierend auf der Überprüfung früherer Studien wurde ein systematischer analytischer Rahmen vorbereitet, der aus drei Hauptstudien besteht, um das Verständnis der Prozesse und Merkmale von Lehrkrafturteilen zu erweitern. In den drei vorgestellten Studien wurde insbesondere untersucht, wie Lehrkrafturteile durch verschiedene Schülermerkmale generiert werden, welche Möglichkeiten es gibt, die Urteilsgenauigkeit von Lehrkräften zu verbessern und ob die Urteilsgenauigkeit von Lehrkräften im Laufe der Zeit stabil bleiben kann.
In der ersten Studie wurde das Linsenmodell der Theorie der sozialen Beurteilung angewendet, um die Einschätzungen von Lehrpersonen über die Leistung von Schülerinnen und Schülern und ihre Strategien der Informationsverarbeitung besser zu verstehen. 260 Lehrkräfte aus sieben chinesischen Grundschulen wurden gebeten, aus sieben Informationsquellen Schülermerkmale auszuwählen und zu bewerten, anhand derer sie die Leistungen der Schüler beurteilen könnten. Die Lehrpersonen entwickelten eine klare Hierarchie der verwendeten Datenquellen. Die besten Informationen wurden aus den Fähigkeiten und Einstellungen der Schülerinnen und Schüler gewonnen und die am wenigsten wichtigen Informationen aus der sozialen Interaktion mit anderen sowie aus der Schüler-Demografie. Um genauere Einschätzungen zu treffen, sollten die Lehrkräfte über gültige Indikatoren für die Schülerleistung informiert werden.
Die zweite Studie zielte darauf ab, die Urteilsgenauigkeit von Lehrkräften und die Leistung der Schülerinnen und Schüler durch den Einsatz von Classroom-Response-Systemen („Clickern“) zu fördern. 20 Schulklassen mit 459 Schülerinnen und Schülern der sechsten Klasse und ihren Mathematiklehrkräften wurden für eine fünfwöchige quasi-experimentelle Interventionsstudie mit einem Pre- und Post-Test in drei Gruppen eingeteilt. Die Ergebnisse zeigen, dass beide Ziele weitgehend erreicht werden konnten. Schülerinnen und Schüler der Clicker-Gruppe haben durch die Intervention mehr mathematisches Wissen erworben als Studenten der Tagebuch- und Kontrollgruppe. Die Lehrkrafturteile aller drei Gruppen wurden vom Pre- zum Post-Test genauer. Lehrpersonen, die Clicker verwendeten, beurteilten jedoch mit höchster Genauigkeit. Clicker können als wertvolles Werkzeug zur Verbesserung der Urteilsgenauigkeit von Lehrkräften empfohlen werden.
In der dritten Studie wurde die zeitliche Stabilität der Urteilsgenauigkeit der Lehrkräfte hinsichtlich Motivation, Emotion und Leistung der Schülerinnen und Schüler untersucht. Neun Klassen mit 326 Sechstklässlern einer chinesischen Grundschule und ihren Mathematiklehrpersonen nahmen an der Studie teil. Die Schüler arbeiteten an einem standardisierten Mathematik-Test und einem Selbstbeschreibungsfragebogen zu Motivation und Emotion. Die Lehrpersonen beurteilten die Motivation, Emotion und Leistung jedes einzelnen Schülers anhand einzelner Items. Das Lehrkrafturteil und die Eigenschaften der Schülerinnen und Schüler wurden innerhalb von vier Wochen zweimal gemessen. Die Ergebnisse zeigten, dass die Lehrkräfte in der Lage waren, die Schülerleistungen mit hoher Genauigkeit, die Motivation der Schülerinnen und Schüler mit mäßiger bis hoher Genauigkeit und die Emotion der Schülerinnen und Schüler meist mit geringer Genauigkeit zu bewerten. Die Urteilsgenauigkeit der Lehrpersonen war sehr stabil mit nur geringen Veränderungen an den verschiedenen Genauigkeitskomponenten. Es kann gefolgert werden, dass chinesische Grundschullehrkräfte in der Lage sind, zu verschiedenen Zeitpunkten faire Urteile über Schülerleistungen und der Motivation ihrer Schülerinnen und Schüler zu treffen. Die Emotionen der Schülerinnen und Schüler sind für Lehrpersonen jedoch schwer zu erfassen.
While in some Eastern European countries a wave of colored revolutions challenged existing political orders, Belarus has remained largely untouched by mass protests. In Minsk, the diffusion of democratic ideas leading to the mobilization of population meets a stable authoritarian regime. Nevertheless, the stagnating democratization process cannot be only attributed to the strong authoritarian rule and abuse of power. Indeed, Belarusian president Alexander Lukashenko still enjoys popularity by a large part of the population. Although international observers report that elections in Belarus have never been free and fair, few commentators doubt that Lukashenko would not have won in democratic elections. This evidence suggests that the regime succeeded in building a strong legitimizing basis, which has not been seriously challenged during the last two decades. This paper explores the authoritarian stability in Belarus by looking at the patterns of state ideology. The government effectively spreads state ideology since the early 2000s. Ideology departments have been created in almost all state institutions. The education sector has been affected by the introduction of the compulsory course "The Fundamentals of Belarusian State Ideology" at all universities, and increasing attention to the patriotic education at schools. Based on document analysis, I trace the creation of "ideological vertical" in Belarus and focuse on the issue of ideology in education and youth policy sectors.
This cumulative thesis consists of six single contributions: five independent essays and an introductory chapter. All of the conributions have been published or accepted for publication. The overarching scope of these essays is to analyze different factors accounting for the stability of post-Soviet authoritarian regimes and the obstacles, which Western democracy promoters can face when dealing with autocrats. The starting point for these enquiries has been the striking inability of Western democracies and in particular of the European Union to encourage and to assist political transformation in the majority of the post-Soviet republics.
Commutative Gröbner bases have a lot of applications in theory and practice, because they have many nice properties, they are computable, and there exist many efficient improvements of their computations. Non-commutative Gröbner bases also have many useful properties. However, applications of non-commutative Gröbner bases are rarely considered due to high complexity of computations. The purpose of this study was to improve the computation of non-commutative Gröbner bases and investigate the applications of non-commutative Gröbner bases. Gröbner basis theory in free monoid rings was carefully revised and Gröbner bases were precisely characterized in great detail. For the computations of Gröbner bases, the Buchberger Procedure was formulated. Three methods, say interreduction on obstructions, Gebauer-Möller criteria, and detecting redundant generators, were developed for efficiently improving the Buchberger Procedure. Further, the same approach was applied to study Gröbner basis theory in free bimodules over free monoid rings. The Buchberger Procedure was also formulated and improved in this setting. Moreover, J.-C. Faugere's F4 algorithm was generalized to this setting. Finally, many meaningful applications of non-commutative Gröbner bases were developed. Enumerating procedures were proposed to semi-decide some interesting undecidable problems. All the examples in the thesis were computed using the package gbmr of the computer algebra system ApCoCoA. The package was developed by the author. It contains dozens of functions for Gröbner basis computations and many concrete applications. The package gbmr and a collection of interesting examples are available at http://www.apcocoa.org/.
This thesis investigates the suitability of state-of-the-art protocols for large-scale and long-term environmental event monitoring using wireless sensor networks based on the application scenario of early forest fire detection. By suitable combination of energy-efficient protocol mechanisms a novel communication protocol, referred to as cross-layer message-merging protocol (XLMMP), is developed. Qualitative and quantitative protocol analyses are carried out to confirm that XLMMP is particularly suitable for this application area. The quantitative analysis is mainly based on finite-source retrial queues with multiple unreliable servers. While this queueing model is widely applicable in various research areas even beyond communication networks, this thesis is the first to determine the distribution of the response time in this model. The model evaluation is mainly carried out using Markovian analysis and the method of phases. The obtained quantitative results show that XLMMP is a feasible basis to design scalable wireless sensor networks that (1) may comprise hundreds of thousands of tiny sensor nodes with reduced node complexity, (2) are suitable to monitor an area of tens of square kilometers, (3) achieve a lifetime of several years. The deduced quantifiable relationships between key network parameters — e.g., node size, node density, size of the monitored area, aspired lifetime, and the maximum end-to-end communication delay — enable application-specific optimization of the protocol.