TY - THES TI - Map matching methods for assigning vehicle positions to road networks AU - Wöltche, Adrian PY - 2026 PB - University of Bamberg Press CY - Bamberg T3 - Schriften aus der Fakultät Wirtschaftsinformatik und Angewandte Informatik der Otto-Friedrich-Universität Bamberg SN - 978-3-98989-134-0 DO - 10.20378/irb-116612 UR - https://fis.uni-bamberg.de/handle/uniba/116612 LA - eng AB - Map matching is a widely used technology for resolving discrepancies between sequences of geocoordinates from movements and road networks. During movement, geocoordinates can be measured with the help of global navigation satellite systems. However, due to the nature of these systems, the measured geopositions contain considerable noise. Consequently, discrepancies generally occur between the measured positions and the underlying roads on which the movement originally occurred. Map matching is a combination of defined models, algorithms, and metrics for correcting these discrepancies. The goal of map matching is to find routes within the road network that most accurately represent the recorded movements. The results of map matching exactly overlap with the road network, are free from the original noise, and are corrected concerning impossible detours and gaps from the measured movements. These results facilitate further analyses in the domain of mobility research. In this thesis, we present a novel map matching method based on Markov decision processes and reinforcement learning algorithms. Our novel method introduces various advances in multiple aspects of the map matching process pipeline for increasing the accuracy of our results in comparison to existing solutions. Our process includes an improved trajectory simplification preprocessing step for improving matching performance both in terms of accuracy and speed, a novel candidate adoption feature that enables the stochastic process to dynamically resolve outliers and noise clusters, a new metric that evaluates direction changes in routes for reducing detours and round trips in the results, and finally, a new comparison algorithm that can evaluate results from different map matching solutions with varying numerical coordinate precision. Our novel map matching solution is provided as ready-to-use open source software, and we compare our solution with multiple third-party open source solutions in terms of computational speed, resource consumption, and accuracy of the results. For achieving high performance, our open source reference implementation contains several specific data structures, algorithmic customizations, and software design decisions, that we explain in context. For our extensive benchmark, we use multiple existing map matching data sets, with a mixture of ground truth and hand-corrected solutions. Furthermore, we introduce and analyze a novel open source process for generating artificial ground truth map matching data. Our evaluation demonstrates that our novel map matching method is more robust to noise and outliers than existing methods and that it outperforms them in terms of accuracy and computational performance. Our new map matching approach contains several key concepts that can be individually transferred to other solutions. Additionally, our new approach generates further ideas for future research, some of which we present in this work. KW - map matching KW - open source KW - benchmark KW - geographic information science KW - markov decision process KW - reinforcement learning ER - TY - THES TI - LLMs and Generative AI for Everything? : A Case Analysis for Applications in Natural Language Processing AU - Jegan, Robin PY - 2026 PB - University of Bamberg Press CY - Bamberg T3 - Schriften aus der Fakultät Wirtschaftsinformatik und Angewandte Informatik der Otto-Friedrich-Universität Bamberg SN - 978-3-98989-130-2 DO - 10.20378/irb-116344 UR - https://fis.uni-bamberg.de/handle/uniba/116344 LA - eng AB - The research field of Natural Language Processing (NLP) has experienced a major shift since the introduction of Large Language Models (LLMs). All facets and application scenarios within NLP have been impacted by the use of LLMs. Current research as well as practice of text processing tools is focused mainly on the application and development of LLMs. Major investments, not only by LLM providers but also other companies applying LLMs in their workflows, have only solidified the role of LLMs in NLP - and in other research and application areas - as part of the artificial intelligence boom in recent years. However, limitations and downsides of the application of LLMs have also emerged. Problems regarding the generated texts as well as the environmental impact of the large-scale use of LLMs are just two of many factors that should be critically analyzed, despite the hype and the prevalence of LLMs for NLP tasks. These restrictions provide the main motivation for this thesis. Traditional models as alternatives to LLMs will be discussed from different perspectives. The characterization of traditional models will be progressively developed as features of alternatives to LLMs will emerge during the course of this thesis. This process will be grounded in experiments, observations and evaluations. Several NLP applications will be presented by surveying the state of the art with neural network-based models such as LLMs as well as the current usage of traditional models. The concrete NLP applications comprise information and relation extraction, text classification, text segmentation, text simplification and text summarization. The first half of this thesis will present the emergence of LLMs contextualized along previous developments within NLP. Characteristics of the selected NLP applications will be collected before a structured literature review will display the prevalence of LLMs regarding each application and will discuss if traditional models are still actively researched. Lessons from domains with long-standing development procedures and processes will also be taken into account to provide a purposeful and structured manner of approaching NLP tasks. A collection of challenges within current NLP will conclude the first half of the thesis, which will serve as motivation for the analysis of experiments and applications of the latter half. The second half of this thesis will present observations and evaluations from use cases, aligned towards the challenges recognized in the first half. Through the analysis of these use cases, benefits of applying traditional models will be collected and supported, in particular through the analysis of a text segmentation use case that is purposefully applied with the lessons drawn from the first half of the thesis in mind. The interpretation of these results will conclude in a discussion on the applicability of traditional models in contrast to LLMs and also give recommendations of both model types for different use cases. Concrete use cases for information extraction, entity matching, text classification and text segmentation will be presented, in which traditional models match or surpass the performance of modern methods. Through improved efficiency as well as enhanced explainability and reproducibility in comparison with neural network-based techniques, these showcases demonstrate the continued relevancy of traditional techniques in today's NLP landscape. Overall, this thesis discusses the role of traditional models in current NLP research and practice, especially in contrast and comparison to modern neural network-based approaches including LLMs. The applicability of modern and less modern techniques is analyzed through a case-based analysis of NLP tasks in a structured and purposeful manner. KW - Natural Language Processing KW - Large Language Models KW - Traditional Models ER - TY - THES TI - Nutzung von Adaptiven User Interfaces zur Personalisierung einer digitalen Studienplanungsassistenz AU - Notter, Rebecca PY - 2026 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-113909 UR - https://fis.uni-bamberg.de/handle/uniba/113909 LA - deu AB - Die Studienplanung an deutschen Hochschulen ist aufgrund fragmentierter Systeme und heterogener Studierendenbedürfnisse komplex. Digitale Studienassistenzsysteme sollen diese Prozesse unterstützen, berücksichtigen nach aktueller Forschung jedoch zumeist nur konkrete Nutzungssituationen ohne den Blick auf den Verlauf eines Semesters und die Veränderungen damit einhergehender Studierendenbedürfnisse. Somit war Ziel dieser Masterarbeit, einen Fokus auf den zeitlichen Wandel studienorganisatorischer Tätigkeiten und Bedürfnisse zu legen und mittels Konzepten adaptiver Benutzeroberflächen die Personalisierung des digitalen Studienassistenzsystems Baula der Otto-Friedrich-Universität Bamberg über den Verlauf eines Semesters zu ermöglichen. Hierfür wurde zunächst ein Gesamtbild studienorganisatorischer Tätigkeiten basierend auf Literaturanalyse und qualitativer Inhaltsanalyse erstellt. Deren Ergebnis stellt ein Semesterzeitstrahl mit den Tätigkeitsbereichen Stundenplan-, Lehrveranstaltungs-, Lern- und Prüfungsorganisation dar. Auf Grundlage dieses Zeitstrahls wurden anschließend ausführliche Anforderungsprofile und Adaptive User Interface-Konzepte entwickelt. Daraus wurden drei Funktionalitäten ausgewählt, welche den Verlauf eines Semesters abdecken: Modulwahl, Stundenplanung und Prüfungsorganisation. Diese Konzepte wurden in einem interaktiven Prototyp umgesetzt, der im GitLab der Otto-Friedrich-Universität Bamberg zur Verfügung steht (siehe https://gitlab.rz.uni-bamberg.de/minf/theses/ma-notter). Eine Evaluierung des Prototyps fand in Form von Nutzungstests unter Verwendung der Usability-Test-Methode, des Think Aloud-Protokolls und der qualitativen Inhaltsanalyse statt. Zudem wurden die Tests zur Evaluation und Überarbeitung des Semesterzeitstrahls genutzt. Die Evaluation zeigte einerseits, dass studienorganisatorische Tätigkeiten und Bedürfnisse diverse Veränderungen über den Semesterverlauf durchlaufen und damit verschiedene Ansätze für Adaptive User Interfaces bieten. Andererseits zeigte die Evaluation, dass die Personalisierung insbesondere dann erfolgreich adressiert werden konnte, wenn sie mit allgemeingültigen Kriterien erreichbar war. Die Personalisierung auf der Ebene individuellen Studier- und Lernverhaltens bleibt dagegen eine zentrale Herausforderung, deren Berücksichtigung und Granularität künftig stärker in den Fokus der Weiterentwicklung rücken sollte. Zusammenfassend liefert diese explorative Arbeit erste allgemeingültige Konzepte und Umsetzungsideen für semesterübergreifend adaptierbare UI im digitalen Studienassistenzsystem Baula. Sie legt damit eine Grundlage für weiterführende Forschung, welche beispielsweise die Personalisierung granularer betrachten, bestehende Funktionalitäten ausbauen, Nutzungsdaten vermehrt einbeziehen, eine technische Implementierung in die reale Baula-Umgebung vornehmen oder neue moralische Fragen zu den Grenzen von Adaptivität in DSA beantworten kann. Die abschließende Empfehlung dieser Arbeit zur Personalisierung des digitalen Studienassistenzsystems Baula mittels Adaptive User Interfaces liegt insbesondere in der Bereitstellung personalisierter Informationen und Erinnerungen zum jeweils richtigen Zeitpunkt im Verlauf des Semesters, um so Orientierung, Struktur und Sicherheit im Studienalltag zu bieten. KW - Adaptive User Interfaces KW - Digitale Studienplanung KW - Personalisierung ER - TY - THES TI - Automatic Generation of Misinformation : Natural Language Generators in the Application Scenario of Wikipedia AU - Matschat, David PY - 2026 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-113231 UR - https://fis.uni-bamberg.de/handle/uniba/113231 LA - eng AB - The spreading of misinformation increasingly poses a threat, especially in the online domain. With the advances in computational text generation techniques and in the context of Wikipedia, this thesis will answer the following question: Can contemporary Natural Language Generators be used to intentionally generate Wikipedia articles with embedded, authentic misinformation? For this purpose, the state of research about misinformation and the vulnerability of Wikipedia, as well as the foundations of Natural Language Generation (NLG) are reviewed. Five different, contemporary NLG models are implemented and trained on a dump of Wikipedia articles about English writers. Subsequently, they generate a variety of texts for both a quantitative and a qualitative evaluation. The quantitative outputs are then assessed by automatic NLG evaluation metrics, and the qualitative ones are reviewed manually. Even though two of the models, GPT-2 and GPT-2 (finetuned), achieve very good results concerning word use, grammar, authenticity, and can even produce a Wikipedia-like structure, overall the five implemented models do not pose a imminent threat to the online encyclopedia as they either lack the sophistication in sentence building or the stability in their performance. However, with these already good results, the further development of new NLG models should be closely monitored. KW - Natural Language Generation KW - Generation of Misinformation KW - Wikipedia KW - Natural Language Processing ER - TY - THES TI - An Approach Towards Unsupervised Text Simplification on Paragraph-Level for German Texts AU - Fruth, Leon PY - 2026 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-112994 UR - https://fis.uni-bamberg.de/handle/uniba/112994 LA - eng AB - Text simplification aims to make texts easier to read and comprehend for people. Recent approaches tackle this task by training neural models on large-scale parallel datasets. However, most languages only have limited simplification data available. Laban et al. [2021b] presented an unsupervised simplification approach to avoid the need for parallel data. They introduced the training algorithm k-SCST, which optimizes a reward by generating and scoring multiple candidate simplifications and encouraging the candidates that outperform the mean reward. This thesis adapts this approach to simplify short paragraphs from German Wikipedia articles. The individual scores of the reward regarding simplicity, fluency and meaning preservation are modified for the new domain and language. In addition, other aspects of the training method are explored. The results show some lexical and syntactic simplification phenomena but also problems regarding fluency and faithfulness. The findings are assessed, and suggestions for future improvements are presented. KW - Text Simplification KW - Natural Language Processing KW - Unsupervised Machine Learning KW - German Text Processing KW - Language Models ER - TY - THES TI - Multilingual Text Summarization Approaches : A Case Study on Generative and Extractive Methods AU - Dal Cin, Giulia PY - 2026 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-112942 UR - https://fis.uni-bamberg.de/handle/uniba/112942 LA - eng AB - In recent years, research in the field of automatic text summarization (ATS) has mainly focused on improving model performance, but it has rarely considered the context and the purpose for which summaries are produced. Therefore, in this master’s thesis, five multilingual ATS scenarios are defined, and each of them is associated with a purpose and some specific requirements. These scenarios are used to evaluate and compare the summaries produced by three ATS systems: extractive algorithm LexRank, pre-trained language model mLongT5, and large language model Mistral NeMo. Both quantitative and qualitative evaluation is performed. Results show that LexRank often fails at writing well-structured and coherent summaries; to a minor extend, mLongT5 does as well. In some of the five scenarios, both systems also produce summaries with insufficient information coverage. Additionally, mLongT5-generated summaries often contain factually incorrect statements or hallucinations. Problems linked to factually incorrect content, hallucinations and insufficient information coverage also occur in NeMo-generated summaries, but only rarely. Additionally, NeMo often does not respect length requirements, and it sometimes switches language in its summaries. Despite these problems, NeMo has good results in almost every scenario, outperforming the other two systems. However, performance differences between the systems vary based on the scenario. KW - Automatic Text Summarization KW - Natural Language Processing KW - Multilingual Text Summarization ER - TY - THES TI - Automatische Erzeugung von Marginalien AU - Lang, Aaron PY - 2026 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-112598 UR - https://fis.uni-bamberg.de/handle/uniba/112598 LA - deu AB - Im Gegensatz zu verwandten Aufgaben wie der Keyword Prediction oder Text Summarization wurde die automatische Erzeugung von Marginalien bisher noch nicht untersucht. Diese Masterarbeit adressiert diese Lücke, indem zunächst ein Datensatz aus Marginalien deutscher Informatik-Lehrbücher erstellt wird. Darauf aufbauend werden verschiedene Methoden der Keyword Prediction anhand dieses Datensatzes quantitativ und qualitativ evaluiert. KW - Marginalien KW - Keyword Extraction KW - Keyword Prediction KW - Keyword Generation KW - Keywords ER - TY - THES TI - Entity Matching for Person Records in Authority Files AU - Hebeis, Maximilian PY - 2026 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-112660 UR - https://fis.uni-bamberg.de/handle/uniba/112660 LA - eng AB - Authority control is of increasing importance in multiple research fields. To ensure reusability as promoted by the FAIR principles, authority files provide persistent identifiers to named entities such as persons, organisations, or geographical locations. Initially conceived as controlled vocabularies in the library context, in recent years authority files have evolved into broader knowledge bases with additional information on the entity referred to in each authority record. Reconciling authority files which contain records referencing the same entity remains a challenge. This thesis represents a case study into applying learning-based entity matching to person records from two large authority databases, namely the German national Integrated Authority File (GND) and the crowd-sourced open knowledge base Wikidata. A workflow is built to extract the person records from the data dumps of both authority files in the RDF/N-Triples format and store them in MongoDB collections. The person records with existing link attributes to a record from the respective other database are then used as source data to train, validate and test learning-based classifiers on. These classifiers are subsequently compared based on their performance classifying person record pairs correctly as matching (referring to the same person) or non-matching. The two approaches compared are classifiers based on classical machine learning and classifiers based on deep learning, i.e. a multilingual variant of the language model BERT. The results show that learning-based entity-matching architectures achieve high F1 scores on GND and Wikidata authority data; despite their semantic capabilities, BERT-based approaches don’t manage to outperform ML classifiers on the data used in this study. KW - Entity Matching KW - Entity Resolution KW - Authority Control KW - Authority File KW - BERT KW - Deep Learning KW - Machine Learning KW - GND KW - Wikidata ER - TY - THES TI - Ansätze für einen frageorientierten Zugang zu Studienordnungen AU - Möglich, Frauke PY - 2025 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-112943 UR - https://fis.uni-bamberg.de/handle/uniba/112943 LA - deu AB - Ziel dieser Arbeit war es, einen frageorientierten Zugang zu Studienordnungen für Studierende zu entwickeln. Dabei wurden verschiedene Ansätze untersucht, die es Studierenden ermöglichen, anhand typischer Fragestellungen relevante Stellen in ihren Studienordnungen zu finden und so den Zugang zu Informationen zu erleichtern. Zu Beginn wurde das Informationsbedürfnis der Zielgruppe analysiert und eine nutzenden-orientierte Evaluationsmetrik entwickelt. Anschließend wurden unterschiedliche Methoden – darunter TF-IDF, BERT-basierte Modelle sowie das Large Language Model Mistral Large – getestet und bewertet. Die Ergebnisse zeigten, dass unter bestimmten Bedingungen TF-IDF die besten Resultate lieferte, gefolgt von Mistral und den BERT-basierten Modellen. Es wurden weiterhin verschiedene Merkmale der Texte untersucht, mit dem Ziel, die Bedingungen, in denen die Ansätze gut oder schlecht funktionieren, herauszufinden. Die Ergebnisse, die mit Mistral Large gewonnen wurden, wurden außerdem nach Fehlern, die typisch für Mistral Large sein könnten, hin untersucht. Explorativ wurde ein BERT-basiertes Modell unter Verwendung von durch TF-IDF gefundenen Features getestet. KW - Frageorientierter Zugang KW - Studienordnungen KW - Natürliche Sprachverarbeitung ER - TY - THES TI - Classification and Class Search Using Voting Techniques AU - Wegmann, Markus ED - Bamberg PY - 2025 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-106578 UR - https://fis.uni-bamberg.de/handle/uniba/106578 LA - eng AB - Beyond the search for content and information, such as texts and documents of all types in information retrieval, class-based search provides information on suitable, superordinate categories, suitable sources, and appropriate generic terms or classifications based on document rankings and their properties. The starting point and basis of this thesis is existing research on voting techniques in expertise retrieval, which addresses the search for suitable experts on a searched topic based on the relevant and ranked documents of a collection. In this thesis, the concept of expert search is derived, generalized, and referred to as a class search. To find these suitable superordinate classes in response to a query, associated scored and ranked documents with their properties vote for them. This voting process does not correspond to the classic electoral voting systems, but uses voting techniques that aggregate the probative value of voting documents for each class in different ways. Following theoretic analyses, the properties of conventional and, in the course of the work, also new voting techniques are examined in a larger experimental setup, and their applicability in general scenarios is investigated. In addition to their use at the document level, voting techniques are further investigated at the level of document passages; the relevance of document rankings resulting from passages that vote for their documents is examined. In addition to similarity measures, the applicability of voting techniques is also examined and evaluated at the level of distance measures. Using the example of hierarchical clustering, the application of voting techniques is related to known clustering techniques, and their systemic behavior is analyzed. With its theoretical considerations and evaluations of practical scenarios, this work provides a broad overview of voting techniques, their characteristics, differences, and their diverse application scenarios. KW - Information Retrieval KW - Class Search KW - Classification KW - Voting Techniques ER - TY - THES TI - An Investigation of Three Novel Applications for RDF-based Knowledge Graphs : Unveiling Opportunities, Challenges, and Recommendations AU - Martin, Leon PY - 2024 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-105058 UR - https://fis.uni-bamberg.de/handle/uniba/105058 LA - eng AB - Various sources suggest that Knowledge Graphs (KGs) despite being built upon decades-old foundations, are currently in the midst of a hype. The reason for this is rooted in the variety of applications, for which KGs can be used. While KGs have been employed successfully in many application domains such as information retrieval and recommender systems, recent research in machine learning and artificial intelligence highlights the opportunities that KGs provide in these domains. However, like any other hyped technology, KGs have been subject to inflated expectations, which has led a certain disillusionment in the recent years. To contribute to the further maturation of KGs towards a state with a deep and realistic understanding of the technology, this cumulative doctoral thesis investigates three novel applications of RDF-based KGs. The first application leverages the large open-domain KG Wikidata to identify semantically useful paths between entities. These paths are then used as the basis for generating entity relationship explanations, which are to be displayed in the knowledge panels on the result pages of web search engines with the goal of satisfying the users' information need more quickly. The second application establishes a bidirectional knowledge exchange between (LaTeX-based) research publications and Scientific Knowledge Graphs (SKGs), one category of domain-specific KGs, through custom import and export commands. This technology is intended to narrow the gap between research publications and SKGs to facilitate the shift towards the envisaged KG augmented research. The third application leverages the reasoning capabilities of the RDF ecosystem to validate software repositories, represented as KG fragments, against quality criteria that they should meet according to the type of the respective project. In light of the still ongoing reproducibility crisis, the validation of software repositories against the FAIR principles is examined in particular. Based on the insights obtained from the investigation of the three applications, opportunities, challenges, and recommendations for the utilization of KGs are derived. Some of the addressed aspects elaborate on topics discussed in previous work while others add novel points to this line of research. To give some examples, the aspects include remarks on the ability of KGs to integrate and interconnect heterogeneous data, the complexity of the RDF ecosystem, and data access facilities and interfaces. KW - Knowledge Graphs KW - linked data KW - RDF KW - Wikidata KW - information retrieval KW - scientific knowledge graphs KW - FAIR KW - software repositories KW - quality assessment ER - TY - THES TI - Bewegte 3D-Objekte im urbanen Raum als Herausforderung für Augmented-Reality-Systeme AU - Öder, Bernhard PY - 2024 PB - Otto-Friedrich-Universität CY - Bamberg DO - 10.20378/irb-92165 UR - https://fis.uni-bamberg.de/handle/uniba/92165 LA - deu AB - Die Arbeit gibt einen Überblick über verschiedene Technologien, welche im Zusammenspiel die Grundlage für Augmented-Reality-Anwendungen darstellen. Vor diesem Hintergrund wird am Beispiel der ehemaligen Straßenbahn Bambergs gezeigt, dass der komplexe Fall von bewegten 3D-Objekten im urbanen Raum für Augmented-Reality-Systeme eingeschränkt realisierbar ist. Das Modell der Straßenbahn wird dazu mittels Bilderkennung in den Stadtraum projiziert und bewegt sich über vier Streckenabschnitte entlang einer Straße. Die Anwendung läuft auf herkömmlichen Smartphones. Als Hauptherausforderungen wurden die Bilderkennung im urbanen Raum, die präzise Objektplatzierung, die Visualisierungsstabilität der Straßenbahn gegenüber der realen Umwelt sowie das Verdeckungsverhalten erkannt. Als Faktoren, welche die Ausführung und die Visualisierungsqualität beeinflussen, wurden Belichtung, Auswahl des verwendeten Smartphones, das Nutzerverhalten und Verdeckungen erkannt. Aus den genannten Faktoren ergeben sich Einschränkungen bei der Erstellung einer solchen Anwendung, welche berücksichtigt werden müssen, um den Herausforderungen gerecht zu werden. Die Arbeit zeigt, wie die AR-Anwendung im Detail entstanden ist. Für eine einfache und präzise Objektplatzierung kommt Bilderkennung zum Einsatz. Umgesetzt wurde sie unter der Verwendung von Unity und Vuforia. Das Modell der Straßenbahn wurde in Blender erstellt. KW - Augmented Reality KW - Cultural Heritage KW - Welterbe KW - AR KW - Smart City ER - TY - THES TI - Effective and Efficient Summarization of Two-Dimensional Point Data : Approaches for Resource Description and Selection in Spatial Application Scenarios AU - Kufer, Stefan PY - 2019 PB - University of Bamberg Press CY - Bamberg T3 - Schriften aus der Fakultät Wirtschaftsinformatik und Angewandte Informatik der Otto-Friedrich-Universität Bamberg SN - 978-3-86309-672-4 DO - 10.20378/irbo-55137 UR - https://fis.uni-bamberg.de/handle/uniba/46917 LA - eng AB - Space is everywhere, and so is spatial data. The digital revolution has led to an enormous pool of available spatial data, and the amount of newly generated spatial data is increasing day by day. Often, this spatial data is two-dimensional point data that in addition is associated with data objects such as media objects (e.g. pictures or texts). As a consequence of the huge amount of data objects which is generated and then has to be maintained, there is a need for effective and efficient search systems which are capable of addressing the spatial properties of the data objects. In this context, different search scenarios exist: The data objects might be maintained in a distributed system (i.e. on a set of different machines) or in a centralized system (i.e. on a single machine) which, in each case, leads to varying requirements for appropriate search systems. However, in both scenarios, the concept of resource description and selection is an applicable paradigm for conducting similarity searches with regard to the spatial properties of the data objects. Hereby, when focusing on the spatial aspects, a resource is an abstract entity that administers spatial point data. For describing the spatial footprint of a resource, its spatial data point set can be ‘summarized' by means of geometrically delineated areas which cover this data point set. These resource descriptions are then usable for a targeted selection of the resources which administer the relevant data objects based on spatial properties. This thesis is concerned with the effective and efficient summarization of two-dimensional spatial point data as a means for resource description and selection in spatial application scenarios. The term ‘effective' refers to a spatially very accurate geometric delineation of the data point set to describe whereas the term ‘efficient' relates to its storage-space-efficient representation. Two search-based spatial application scenarios are assessed in this thesis, and in both, the search task can be modelled by adhering to the concept of resource description and selection. The first scenario is a distributed application scenario in which spatial point data is maintained by a set of independent resources such as peers in a peer-to-peer network. Given a concrete query, the resource descriptions shall enable the targeted selection of the peers administering the relevant data points while ignoring ‘irrelevant' resources. The distributed application scenario is the main part of this thesis, and our summarization approaches are specifically developed for this purpose. Hereby, a variety of concepts is considered which include the compression of the data points, the use of arbitrarily complex bounding volumes to delineate them, the representation of complex objects (such as complex bounding volumes) with a set of simple objects, and a division of the data point set to describe into groups and the subsequent concise delineation of each group. Overall, 14 summarization approaches are presented in this thesis which can be categorized into data partitioning approaches, space partitioning approaches, and hybrid approaches. Following the specification of suitable resource selection schemes which are based on the various summaries, an extensive evaluation is conducted by means of assessing the approaches' performances for k nearest neighbor (kNN) queries. The evaluation considers different data collections and varying environmental conditions in order to assess the robustness of the approaches. In the second application scenario, a utilization of our summarization approaches in a multidimensional data structure is assessed---which constitutes a centralized application scenario. More specifically, the two summarization approaches of the distributed application scenario which are most suitable for this purpose are integrated into an R-tree which administers sets of two-dimensional point data. The R-tree is a tree-based, centralized multidimensional data structure which has been the subject of intensive research over many years, starting in 1984. Within an R-tree, the ‘resources' are the nodes of the hierarchical tree structure. The resource descriptions shall enable the targeted traversal of paths in the tree structure when searching for the relevant data points given a specific query. Traditionally, these nodes are described by Minimum Bounding Rectangles (MBRs) which are very storage-space-efficient but rather coarse descriptions of a spatial footprint. After presenting the classical R-tree, the modifications to the R-tree that are necessary to integrate our summarization approaches are discussed. Also, appropriate algorithms for calculating summaries from a set of rectangles are outlined as this is a prerequisite for the efficient use of our sophisticated summarization approaches in R-trees. Following the specification of appropriate range query and kNN query algorithms, an extensive evaluation is conducted which assesses both the improvement potential over the traditional R-tree using MBR summaries as well as the degree of achievement by means of the straightforward approach to the integration that we pursue. Also in this evaluation, different data collections and varying environmental conditions are considered. Overall, the thesis presents a wide variety of summarization approaches for describing sets of two-dimensional point data. These summarization approaches are excellently suited for usage in the investigated distributed application scenario. Furthermore, we assess the utilization of the two summarization approaches which are most appropriate for this purpose in an already very intensively researched centralized application scenario. Here, we examine further improvement potentials and identify important obstacles which are to overcome for our summarization approaches in such an environment. KW - resource description KW - resource selection KW - summarization KW - spatial indexing KW - quadtree KW - r-tree KW - two-dimensional KW - spatial data ER - TY - THES TI - Korpusbasierte Verfahren zur Generierung lexikalischer Ressourcen für das Opinion Mining : statistische Ansätze und deren Einsatzmöglichkeiten AU - Reinel, Dirk PY - 2018 PB - University of Bamberg Press CY - Bamberg T3 - Schriften aus der Fakultät Wirtschaftsinformatik und Angewandte Informatik der Otto-Friedrich-Universität Bamberg SN - 978-3-86309-593-2 DO - 10.20378/irbo-52349 UR - https://fis.uni-bamberg.de/handle/uniba/44092 LA - deu AB - Die ständig zunehmende Anzahl nutzergenerierter textueller Daten im World Wide Web, wie beispielsweise Kundenrezensionen, und die Begehrlichkeiten diese Daten hinsichtlich der darin ausgedrückten Meinungen zu Unternehmen, Produkten, Dienstleistungen etc. maschinell auszuwerten, erfordern gut funktionierende, angepasste Methoden des Opinion Mining. Die Grundlage für viele dieser Methoden bilden lexikalische Ressourcen in Form von Lexika meinungstragender Wörter und Phrasen. Diese Lexika existieren bisher allerdings nur für ausgewählte Sprachen, haben diverse inhaltliche Lücken, und sind automatisch (für verschiedene Sprachen) nur mit großem Aufwand zu erzeugen. In dieser Arbeit wird deshalb ein neues Verfahren vorgestellt, das dazu beitragen soll, die benannten Probleme – durch den Einsatz statistischer Methoden – zu überwinden. Zudem wurde, mittels dieses Verfahrens, der Prototyp eines neuen Lexikons mit meinungstragenden Wörtern und Phrasen für die deutsche Sprache generiert und anschließend evaluiert. Dafür wurde im Rahmen eines Experiments mit 20 Teilnehmern ein Basis-Referenzlexikon für die deutsche Sprache manuell erzeugt. Klassische Einsatzgebiete der Opinion Mining Algorithmen und Ressourcen, und damit des vorgestellten Verfahrens, sind Systeme zur Erfassung von Kundenmeinungen zu verschiedenen Unternehmensbereichen zur Unterstützung des Beschwerde- und Reputationsmanagements. Allerdings sind die Möglichkeiten des neu entwickelten Verfahrens nicht auf diese klassischen Anwendungsfelder begrenzt. Auch der interdisziplinäre Einsatz, z.B. zur Untersuchung von Sprachvarietäten im Forschungsfeld der Sprachstatistik, ist denkbar. ER - TY - THES TI - Resource Description and Selection for Similarity Search in Metric Spaces AU - Blank, Daniel PY - 2015 PB - University of Bamberg Press CY - Bamberg T3 - Schriften aus der Fakultät Wirtschaftsinformatik und Angewandte Informatik der Otto-Friedrich-Universität Bamberg SN - 978-3-86309-310-5 DO - 10.20378/irb-21374 UR - https://fis.uni-bamberg.de/handle/uniba/21374 LA - eng AB - In times of an ever increasing amount of data and a growing diversity of data types in different application contexts, there is a strong need for large-scale and flexible indexing and search techniques. Metric access methods (MAMs) provide this flexibility, because they only assume that the dissimilarity between two data objects is modeled by a distance metric. Furthermore, scalable solutions can be built with the help of distributed MAMs. Both IF4MI and RS4MI, which are presented in this thesis, represent metric access methods. IF4MI belongs to the group of centralized MAMs. It is based on an inverted file and thus offers a hybrid access method providing text retrieval capabilities in addition to content-based search in arbitrary metric spaces. In opposition to IF4MI, RS4MI is a distributed MAM based on resource description and selection techniques. Here, data objects are physically distributed. However, RS4MI is by no means restricted to a certain type of distributed information retrieval system. Various application fields for the resource description and selection techniques are possible, for example in the context of visual analytics. Due to the metric space assumption, possible application fields go far beyond content-based image retrieval applications which provide the example scenario here. KW - resource description and selection KW - metric access methods KW - content-based image retrieval ER - TY - THES TI - Concept and implementation of a rule framework to dynamically transform data and queries for heterogeneous collections AU - Gradl, Tobias PY - 2014 PB - Zenodo CY - Genève DO - 10.5281/zenodo.804889 UR - https://fis.uni-bamberg.de/handle/uniba/42052 LA - eng KW - data federation KW - data integration KW - domain specific languages KW - digital humanities KW - language theory ER - TY - THES TI - Information Management for Digital Learners : Introduction, Challenges, and Concepts of Personal Information Management for Individual Learners AU - Gooren-Sieber, Stefanie PY - 2014 PB - University of Bamberg Press CY - Bamberg T3 - Schriften aus der Fakultät Wirtschaftsinformatik und Angewandte Informatik der Otto-Friedrich-Universität Bamberg SN - 978-3-86309-206-1 DO - 10.20378/irb-2861 UR - https://fis.uni-bamberg.de/handle/uniba/2861 LA - eng AB - The current cultural transition of our society into a digital society influences all aspects of human life. New technologies like the Internet and mobile devices enable an unobstructed access to knowledge in worldwide networks. These advancements bring with them a great freedom in decisions and actions of individuals but also a growing demand for an appropriate mastering of this freedom of choice and the amount of knowledge that has become available today. Naturally, this observable rise and progress of new technologies—gently but emphatically becoming part of people’s everyday lives—not only changes the way people work, communicate, and shape their leisure but also the way people learn. This thesis is dedicated to an examination of how learners can meet these requirements with the support that modern technology is able to provide to learners. More precisely, this thesis places a particular emphasis that is absent from previous work in the field and thus makes it distinctive: the explicit focus on individual learners. As a result, the main concern of this thesis can be described as the examination, development, and implementation of personal information management in learning. Altogether two different steps towards a solution have been chosen: the development of a theoretical framework and its practical implementation into a comprehensive concept. To establish a theoretical framework for personal information management in learning, the spheres of learning, e-learning, and personalised learning have been combined with theories of organisational and personal knowledge management to form a so far unique holistic view of personal information management in learning. The development of this framework involves the identification of characteristics, needs, and challenges that distinguish individual learners from within the larger crowd of uniform learners. The theoretical framework defined within the first part is transferred to a comprehensive technical concept for personal information management in learning. The realisation and design of this concept as well as its practical implementation are strongly characterised by the utilisation of information retrieval techniques to support individual learners. The characteristic feature of the resulting system is a flexible architecture that enables the unified acquisition, representation, and organisation of information related to an individual’s learning and supports an improved find-ability of personal information across all relevant sources of information. The most important results of this thesis have been validated by a comparison with current projects in related areas and within a user study. KW - - ER - TY - THES TI - Facettenbasierte Indexierung multipler Artefakte : ein Framework für vage Anfragen in der Produktentwicklung AU - Weber, Nadine PY - 2011 PB - University of Bamberg Press CY - Bamberg T3 - Schriften aus der Fakultät Wirtschaftsinformatik und Angewandte Informatik der Otto-Friedrich-Universität Bamberg SN - 978-3-86309-058-6 DO - 10.20378/irb-373 UR - https://fis.uni-bamberg.de/handle/uniba/373 LA - deu AB - Durch den zunehmenden Einsatz von Informations- und Kommunikationstechnologien sowie den schnellen Technologiefortschritt steht die Entwicklung technischer Produkte vor immer neuen Herausforderungen. Dabei ist die Aufgabe der Produktentwicklung selbst als Problemlösungsprozess zu betrachten, in dem Lösungen mittels intensiver Informationsverarbeitung gefunden werden. Somit werden täglich unterschiedlichste Arten von Informationen erstellt, benötigt und verarbeitet, die primär in digitaler Form vorliegen. Diese werden in heterogenen Anwendungssystemen verwaltet, was eine Wiederverwendung bereits existierender Informationen erschwert. Damit beansprucht die Suche nach Informationen noch immer einen erheblichen Anteil der Entwicklungszeit. Zur Verbesserung der Informationsversorgung im Bereich der technischen Produktentwicklung wird ein interaktives Information Retrieval-System – das LFRP-Framework – vorgestellt. Dieses kombiniert die vier Basiskonzepte der multiplen Ebenen, der facettierten Suche, des Rankings und der parallelen Koordinaten, um hochkomplexe Informationsbedürfnisse zu befriedigen. Seine Realisierung erfordert neben einer geeigneten Suchoberfläche die Entwicklung einer Indexierungskomponente, welche die vorhandenen Informationen in eine für das LFRP-Framework rechnerverarbeitbare Form transformiert. Dieser als Indexierung bezeichnete Prozess stellt die Grundvoraussetzung für die Funktionsfähigkeit eines Suchsystems dar und liegt daher im Fokus der Betrachtung. Es wird ein Lösungsansatz vorgestellt, welcher eine Indexierung in Form facettenbasierter Suchkriterien ermöglicht und dabei nicht nur Informationen aus heterogenen Anwendungssystemen, sondern insbesondere aus entwicklungsspezifischen Dokumenten, wie CAD-Modellen, technischen Zeichnungen oder Stücklisten, berücksichtigt. KW - FORFLOW, LFRP-Framework, Enterprise Search, Facettenbasierte Suche KW - FORFLOW, LFRP-Framework, Enterprise Search, Facet-based Search KW - LFRP-Framework KW - Enterprise Search KW - Facettenbasierte Suche KW - FORFLOW KW - Enterprise Search KW - Facet-based Search KW - Information Retrieval KW - Information-Retrieval-System KW - Unternehmen KW - CAD KW - Wissensmanagement ER - TY - THES TI - Interactive Search Processes in Complex Work Situations : a Retrieval Framework AU - Eckstein, Raiko PY - 2011 PB - University of Bamberg Press CY - Bamberg T3 - Schriften aus der Fakultät Wirtschaftsinformatik und Angewandte Informatik der Otto-Friedrich-Universität Bamberg SN - 978-3-86309-017-3 DO - 10.20378/irb-287 UR - https://fis.uni-bamberg.de/handle/uniba/287 LA - eng AB - In recent years a steady increase of information produced in organizations can be noticed. In order to stay competitive, companies have a growing interest in reusing existing knowledge from past projects. Furthermore, a complete picture of the available information is necessary to be able to make informed decisions. The variety and complexity of information in modern organizations often exceeds the capabilities of the currently deployed enterprise search solutions. The reasons for that are manifold and range from non-linked information from multiple software systems to missing functionality to support users during search tasks. Existing search engines often do not support the search paradigms necessary in these environments. On many occasions, users are not aware of the results they will find during the formulation of the search queries. Additionally, the aspect of knowledge building and the identification of new insights into the available data is a priority for the users. Therefore, search paradigms are useful to provide users with tools that support exploratory navigation in a data set and help them to recognize relationships between search results. The goal of this publication is the introduction of a framework that supports exploratory searches in an organizational setting. The described LFRP-framework is built on top of four pillars. 1. The multi-layer functionality allows users to formulate complex search queries referring to more than one result type. Therewith, it enables search queries that - starting from a set of relevant projects - allow selections of documents that are linked to these projects. 2. The search paradigm of faceted searching supports users in formulating search queries incrementally by offering dynamic and valid filter criteria that avoid empty result sets. 3. By combining the concept of faceted search with the capability to influence the search result order based on filter criteria, users can define in a fine-grained way which criteria values shall be weighted stronger or weaker in the search results. The interaction with the ranking is conducted transparently by the so-called user preference functions. 4. The last pillar consists of the visualization type of parallel coordinates covering two tasks in the search user interface of the LFRP-Framework. On the one hand, users formulate their search queries solely graphically in the parallel coordinates and on the other hand they obtain a visual representation of the search results and are able to discover relationships between search results and their facets. The framework is introduced formally from a query model point of view as well as a prototypical implementation. It enables users to access large linked data sets by navigation and constitutes a contribution to a comprehensive information strategy for organizations. KW - Unternehmensweite Suche , Information-Retrieval-System , Visualisierung , Produktentwicklung KW - Enterprise Search KW - Unternehmensweite Suche KW - Information-Retrieval-System KW - Visualisierung KW - Produktentwicklung KW - Enterprise Search ER -