TY - THES A1 - Kottek, Nick T1 - Echtzeit-Erkennung von Gesten des deutschen Fingeralphabets mithilfe eines Convolutional Neural Networks N2 - Sign language is an important factor in the integration of deaf and hard of hearing people into society. Due to the limited number of people who speak sign language, there is a communication barrier that needs to be addressed. In recent years, sign language has been an important field of research. There have been numerous attempts at trying to find a solution to this problem. This thesis focuses on the German finger alphabet, which has not been researched as much. It examines whether it is possible to recognize the gestures of the German finger alphabet in real time with a convolutional neural network. For that, literature is consulted, and a prototype is developed. The prototype includes a newly created dataset with 2,500 images distributed over 25 gestures, a convolutional neural network, and software for real-time translation of a video stream. The prototype demonstrates the feasibility of realizing such a task. The convolutional neural network achieves an accuracy of 99.61%. On a powerful desktop computer, the prediction of a single frame takes between 36 and 40 ms. However, there are some limitations and restrictions to the quality of prediction caused by factors such as lighting and background. Additionally, certain similar gestures, such as M and N, are difficult to distinguish. Based on these results, ideas and suggestions for future studies are presented. KW - convolutional neural network KW - sign language KW - German finger alphabet KW - image classification KW - real-time recognition Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:hbz:1383-opus4-19768 ER - TY - THES A1 - Shrestha, Sabita T1 - Named Entity Recognition for Nepali Text Using Pre-Trained BERT-Based Model N2 - The popularity of using transformer-based models like Bidirectional Encoder Representations from Transformers (BERT) for various Natural Language Processing (NLP) tasks is increasing rapidly. Unfortunately, the research is very limited for low-resource languages like Nepali. This study examines the utilisation of pre-trained BERT-based NERNepal for Named Entity Recognition (NER) tasks in Nepali text. The main goal is to investigate the efficiency of the NERNepal model, which has been pre-trained and fine-tuned on Nepali corpus data for NER. This research provides new insights by evaluating the NERNepal model on two distinct datasets. It addresses unique linguistic challenges specific to Nepali. The study also offers a detailed analysis of the model's strengths and weaknesses. By focusing on diverse datasets, this study shows how adaptable the model is and how its performance varies. These aspects have not been explored extensively before. The EverestNER dataset is one of the largest human-annotated datasets in Nepal so far, and the Nepali_NER dataset is also BIO-annotated for the NER task, which helped to compare the model’s prediction. The performance was better on the Nepali_NER dataset in comparison with another selected dataset. The EverestNER dataset contained many complex words for ORG connected with many tokens for a single entity with different contextual meanings and different annotations for the same word in different tokens as per context. Because of this, it created more confusion for the prediction, especially for the ORG entity. It had a similar issue with another dataset as well, but the label annotation was better in comparison. Furthermore, the research tries to clarify the difficulties and constraints related to utilising pre-trained BERT models for NER in low-resource languages such as Nepali. The study focuses on research areas on the efficacy of the model and its performance on two different datasets. Post-training with the EverestNER train dataset was attempted, but due to computational resource limitations, only a maximum of 3 epochs was possible, which did not improve the evaluation. It also has implications for enhancing language processing tools for Nepali. The results show the ability of a pre-trained BERT-based model to improve NER skills for Nepali text. However, further study is required to overcome the current obstacles to identifying complex words. The availability of high computational resources and the possibility of combining other NER approaches with a transformer-based model could increase the performance and robustness of the model. Keywords: Named Entity Recognition, Natural Language Processing, BERT, low resources language KW - Named Entity Recognition KW - Natural Language Processing KW - BERT KW - low resources language Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:hbz:1383-opus4-19870 ER - TY - THES A1 - Nacke, Marius T1 - Künstliche Intelligenz zur Vorhersage von Kreditentscheidungen N2 - Seit der Verschärfung der Kreditrichtlinien im Jahr 2022 ist es zunehmend schwieriger geworden, einen Kredit in Deutschland und im europäischen Raum zu erhalten. Das Interesse an Krediten ist zuletzt ungeachtet dieser Erschwernisse angestiegen. Vergleichs- und Vermittlungsportale für Kredite können potenziellen Darlehensnehmern aktuell nur in begrenztem Maße dabei unterstützen, eine genehmigte Kreditanfrage zu erhalten. Die Vorhersage der Kreditentscheidungen von Banken kann als Basis für weitere Methoden fungieren, die Kreditnehmern helfen können, die Wahrscheinlichkeit einer Genehmigung zu erhöhen. In der vorliegenden Arbeit wird daher untersucht, welche Verfahren des Machine-Learning am ehesten für die Vorhersage von Kreditentscheidungen geeignet sind. Dazu werden die Daten der Yareto GmbH, eines Vergleichs- und Vermittlungsportals für Fahrzeugkredite, für das Training sowohl eines Logistischen-Regressions- und XGBoost-Modells als auch eines Multilayer-Perceptrons verwendet. Das XGBoost-Modell erreicht dabei die höchste Accuracy und Precision mit 73,6\% bzw. 76,1\% bei einer Trainingszeit von lediglich 55 Minuten. Das Multilayer Perceptron kann eine ähnlich hohe Accuracy erreichen, bedarf allerdings mehr als zehn Stunden für das Training. Demnach ist XGBoost am ehesten von den drei untersuchten Verfahren zur Vorhersage von Kreditentscheidungen geeignet. KW - Kreditvergleichsportale KW - Prognosen KW - Künstliche Intelligenz KW - Ensemble-Modelle KW - Kreditentscheidungen Y1 - 2024 ER - TY - THES A1 - Akter, Most Sarmin T1 - Data Analysis of the "Loyalty Program-Valeo Specialist Club" to Improve the Independent Aftermarket: Designing an Efficient Marketing Strategy N2 - Nowadays, a loyalty program is one of the most significant marketing tools for winning genuine customer loyalty. Valeo, a global automotive supplier, also offers the Valeo Specialist Club loyalty program and rewards garage or workshop members who only purchase Valeo spare parts. The core intent of this study is to understand the importance of the Valeo loyalty application, members’ insights, and their experiences for improving the future planning, aftermarket performance, and management of the Value Specialist Club. In today’s competitive market, products, services, and customer loyalty are valuable assets, so the program was created to enhance member satisfaction and loyalty with the Value Specialist Club. A mixed-method approach was used to answer the study question and findings, improve member satisfaction, and refine marketing strategies, using a survey as primary data and Qlik Sense as secondary data. Besides, using a survey methodology, the study collected 79 responses from different countries (the United Kingdom, Germany, Belgium, and the Netherlands) and analyzed the data with only 58 Specialist Club member's responses. A reliability examination was additionally accomplished to measure the accuracy of the collected survey data. Qlik Sense, Microsoft 365, and Excel 2016 were used for data analysis and visualization, which involved descriptive analysis, correlation coefficients, and text sentiment analysis. The study results show positive trends in member satisfaction across four markets, but a decreasing engagement rate is noticeable. In general, members from all countries were satisfied with Valeo products and services, contrasting with moderate dissatisfaction with the Valeo Specialist Club app and customer service. The study also found a strong correlation between the Valeo Specialist Club app, loyalty service, and customer service, suggesting improvements to meet member expectations and preferences. Nevertheless, the study also emphasizes the need to enhance the Valeo Specialist Club registration process, loyalty program features, and benefits to align with current trends, foster increased member satisfaction, and sustain loyalty within the independent aftermarket. Furthermore, the conclusions of this study will aid in developing upcoming marketing plans. KW - Loyalty Program KW - Insights KW - Member Loyalty KW - Marketing KW - Valeo Specialist Club Y1 - 2024 ER - TY - THES A1 - Grigo, Tobias T1 - Das Homeoffice in der Verwaltung – Eine Zwischenbilanz und Erfahrungsberichte zu Best Practices, Weiterentwicklung und Herausforderungen N2 - The home office has become established in the offices of numerous companies and administrations since the events of the global COVID-19 pandemic and has posed numerous challenges. Particularly in Germany, administrative authorities were confronted with the challenge of quickly responding to the events of the pandemic and taking appropriate measures to curb the spread of the virus in the offices of administrations. This paper examines the use of the home office in administration during the pandemic by combining quantitative and qualitative methods such as statistical analyses, interviews, and case studies. In doing so, it sheds light on the experiences of employees as well as the challenges and potentials of home office use. The context aims to demonstrate that the investigation of home office use in administration during the pandemic was crucial in understanding the effects of the newly designed work practices on administrative organizations to better shape future work models. The aim of this study is to gain insights that contribute to the further development of the home office concept in administration. This includes evaluating the long-term effects on work efficiency, work-life balance, and corporate culture. These insights contribute to assessing the effectiveness and efficiency of the home office in administration and lay the groundwork for how optimized work arrangements can be created. They also provide valuable insight into the adaptability of administrative structures to continuously changing work environments. The results of this study will help develop practical solutions and optimize the implementation of the home office in administration. The findings lay new foundations for future research that could address further aspects of home office use in administration, such as long-term effects on organizational performance and employee satisfaction, as well as the development of guidelines for effective implementation of hybrid work models. KW - Homeoffice KW - Administration KW - COVID-19 pandemic KW - Work practices KW - Efficiency Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:hbz:1383-opus4-20272 ER - TY - THES A1 - Regmi, Khem Raj T1 - Interactive Visualisation Tool for Teaching Environmental Data: A Guideline of how to set up the tool taking user feedback into account N2 - This thesis will study the creation of an interactive data visualization tool for environmental dataset that concentrates specifically on Nitrate concentrations in groundwater. This project is intended to create a tool for exploratory/experiential learning - one that allows learners to investigate regional data and pursue genuine learning opportunities. The plan is to use regional data in a specific user journey to create this interactive learning experience. The project uses open environmental data: Hygris dataset of LANUV(Landesamt für Natur, Umwelt und Verbraucherschutz) related to the regions Kleve and Wesel in Germany, and realizes this with help of python programming language and python streamlit framework, web development tools and data visualization libraries (pandas , plotly.express, streamlite.components.v1). This study explores tailored visualization strategies that address key challenges such as the inherent complexity of environmental datasets and the need for more engaging and comprehensible educational tools. These strategies include deploying interactive visualizations that allow hands-on data exploration, and temporal visualizations. The role of user feedback in refining educational tools for teaching environmental data is examined, particularly in designing effective visualization techniques like line char, bar chart, bubble chart, pie chart, scatter plot and intensity graph for representing groundwater nitrate concentrations. The impact of different visualization techniques on the understanding and interpretation of open environmental data among educators and students is analyzed, with an emphasis on the spatial distribution of groundwater nitrate concentrations over time. This approach aims to increase user engagement and deepen understanding of environmental data, facilitating a more profound embedding of information within learners' cognitive processes. KW - Interactive data visualization KW - Data representation KW - Environmental data KW - Nitrate concentration KW - Groundwater KW - Open data KW - User feedback KW - Education KW - Data accessibility KW - Data inclusivity KW - Educational tools KW - Digital teaching Y1 - 2024 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:hbz:1383-opus4-21594 ER - TY - THES A1 - Boulmane, Aya T1 - Development of a Web-Based Tool for Estimating Lifecycle CO₂ Emissions of Electric vs. Diesel Vehicles in Germany N2 - Due to rapid technological development, the transport sector has emerged as one of the most difficult areas to reduce decarbonization, balancing its high global greenhouse gas emissions. In Germany, the shift from internal combustion engine vehicles (ICEVs) to electric vehicles (EVs) has become a key focus of national and EU climate policies. Nevertheless, the transition to EVs is associated with a multitude of factors that need to be considered, including emissions from manufacturing, the energy used during operations, and post-operational treatment. This thesis focuses on the design, development, and refinement of a web-based CO₂ Lifecycle Emission Calculation tool that compares cradle-to-grave emissions of EVs and diesel ICEVs based on different driving scenarios in Germany. It is based on Life Cycle Assessment (LCA) Methodology as indicated in ISO 14040/14044. With the tool, the manufacturing, operational, and end-of-life emissions are estimated based on validated datasets from peer-reviewed literature and authoritative bodies (IPCC, IEA, UBA). The tool’s modular structure, which was developed in Python with the Streamlit framework, also allows for other languages to be added, selection of the region’s grid mix, scenario comparisons, and results visualization in multi-level dashboards. Findings of the literature review show that the tool is accurate for both types of vehicles as the known lifecycle emissions ranges for both types are captured. Carbon payback periods for EVs ranged from less than three years to over eight years depending on the grid context, from renewable-rich to coal-intensive. Sensitivity analysis demonstrates that both grid CO₂ intensity and annual mileage hold the greatest influence over lifecycle results. This research contributes to sustainability as it fills the void that exists between academic LCA models and consumer-facing applications with a transparent, reproducible, and user-configurable decision-support tool. Improvements can be made by incorporating real-time grid emission data, expanding the vehicle datasets, and including additional CO₂ environmental impact categories. Y1 - 2025 ER - TY - THES A1 - Okos, Martin T1 - Ethische und rechtliche Rahmenbedingungen des Maschinellen Lernens - eine systematische Analyse zur Konzeptualisierung und Regulation innerhalb der EU N2 - Die Entwicklungen des maschinellen Lernens und damit einhergehend auch der künstlichen Intelligenz umfassen immer mehr Anwendungsfälle im täglichen Leben. Je komplexer und autonomer diese Systeme werden, desto schwieriger lässt sich ihr Verhalten prognostizieren und es ist nicht auszuschließen, dass Situationen entstehen, die ethische und rechtliche Fragen aufwerfen. Das Ziel dieser Arbeit ist die Untersuchung der erforderlichen Rahmenbedingungen für den Umgang mit diesen Technologien. Neben der Erklärung wesentlicher Begriffe aus dem Bereich der KI und Ethik werden durch die Recherche geeigneter Literatur über Pflegeroboter und dem autonomen Fahren zwei Anwendungsbereiche vorgestellt, die einen Bedarf an moralischen Maschinen aufweisen. Maßnahmen zur Regulierung innerhalb der Europäischen Union erfolgen durch das Anwenden bestehender und Erlassen neuer Gesetze sowie über die Definition ethischer Leitlinien, die ebenfalls als Teil der vorliegenden Arbeit behandelt werden. N2 - Developments in machine learning along with artificial intelligence are encompassing increasingly more use cases in everyday life. The more complex and autonomous these systems become, the harder it is to predict their behaviour, and it cannot be ruled out that situations emerge where ethical and legal questions arise. This thesis aims to investigate the necessary framework conditions for dealing with these technologies. In addition to explaining essential terminology from the domain of AI and ethics, two application areas that show a need for moral machines are presented by referencing contemporary literature on care robots and autonomous driving. The European Union provides regulatory measures by applying existing laws and enacting new ones, as well as by defining ethical guidelines, which are also addressed as part of this thesis. KW - Maschinelles Lernen KW - Künstliche Intelligenz KW - Ethik KW - Pflegeroboter KW - Autonomes Fahren Y1 - 2022 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:hbz:1383-opus4-15170 ER - TY - THES A1 - Salobir, Jan T1 - Relaunch eines Kommunalen Webauftritts beim KRZN N2 - In der Arbeit wird der Prozess des Relaunches eines kommunalen Webauftritts Schritt für Schritt analysiert und besonders darauf geachtet, wie die gesetzlichen Anforderungen für so einen Webauftritt umgesetzt werden. KW - E-Government KW - Onlinezugangsgesetz KW - Webauftritt KW - Verwaltungsinformatik KW - Verwaltungsdigitalisierung Y1 - 2022 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:hbz:1383-opus4-15214 ER - TY - THES A1 - Ponten, Marvin T1 - Visualisierung von Kulturgütern am Beispiel der römischen Kaiserthermen in Trier als 3D Modell. N2 - This paper is a summary of the “Kaiserthermen Project'' for the “Landesmuseum Trier” and shows in detail how to convert a 3Ds Max 3D model for usage into an open source program like Blender. It tries to explain the basics of 3D modelling and compares it to the 3D modelling of cultural heritages. It should work as a form of manual to further convert different models into the FBX file format. It also aims to explain the high value of cultural heritage as a 3D model and compares it to the 3D model of the “Trier Kaiserthermen”. It furthermore gives examples and reasons for the usage of 3D models in different fields. It also presents the final 3D renderings and shows how they were created. At the end it explains how the perfect model for scientific research should be and how this project and paper improved this for the “Kaiserthermen”. KW - 3D Model KW - Cultural Heritage KW - Converting to FBX KW - 3D Render Y1 - 2022 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:hbz:1383-opus4-15889 ER -