Logo des Repositoriums

Auflistung nach Autor:in "Conrad, Stefan"

1 - 6 von 6
  • Konferenzbeitrag
    Automated Extraction of Icon-based Tables
    (INFORMATIK 2024, 2024) Thome, Boris; Hertweck, Friederike; Jonas, Lukas; Yasar, Serife; Conrad, Stefan
    This paper introduces a novel algorithm specifically designed to address the unique complexities of icon-based tables in digital documents. Our algorithm leverages a combination of computer vision techniques to accurately detect table grids, identify symbols, and insert them into their relative positions.
  • Konferenzbeitrag
    Detection and Implicit Classification of Outliers via Different Feature Sets in Polygonal Chains
    (Datenbanksysteme für Business, Technologie und Web (BTW 2017), 2017) Singhof, Michael; Klassen, Gerhard; Braun, Daniel; Conrad, Stefan
    Many outlier detection tasks involve a classification of outliers of di erent types. Most standard procedures solve this problem in two steps: First, an outlier detection algorithm is carried out, which is normally trained on outlier free data, only, since the samples of outliers are limited. Second, the outliers detected in that step, are classified with a conventional classification algorithm, that needs samples for all classes. However, often the quality of the classification is lowered due to the small number of available samples. Therefore, in this work, we introduce an outlier detection and classification algorithm, that does not depend on training data for the classification process. Instead, we assume, that di erent kinds of outliers are inferred by di erent processes and as such should be detected by different outlier detection approaches. This work focuses on the example of outliers in mountain silhouettes.
  • Konferenzbeitrag
    Image landmark recognition with hierarchical K-means tree
    (Datenbanksysteme für Business, Technologie und Web (BTW 2015), 2015) Rischka, Magdalena; Conrad, Stefan
    Today's giant-sized image databases require content-based techniques to handle the exploration of image content on a large scale. A special part of image content retrieval is the domain of landmark recognition in images as it constitutes a basis for a lot of interesting applications on web images, personal image collections and mobile devices. We build an automatic landmark recognition system for images using the Bag-of-Words model in combination with the Hierarchical K-Means index structure. Our experiments on a test set of landmark and non-landmark images with a recognition engine supporting 900 landmarks show that large visual dictionaries of size about 1M achieve the best recognition results.
  • Konferenzbeitrag
    RAPP: A Responsible Academic Performance Prediction Tool for Decision-Making in Educational Institutes
    (BTW 2023, 2023) Duong, Manh Khoi; Dunkelau, Jannik; Cordova, José Andrés; Conrad, Stefan
    Due to the increasing importance of educational data mining for the early intervention of at-risk students and the growth of performance data collected in educational institutes, it becomes natural to employ machine learning models to predict student's performances based off prior data. Although machine learning pipelines are often similar, developing one for a specific target prediction of academic success can become a daunting task. In this work, we present a graphical user interface which implements a customisable machine learning pipeline which allows the training and evaluation of machine learning models for different definitions of academic success, \eg, collected credits, average grade, number of passed exams, etc. The evaluation is exported in PDF format after finishing training. As this tool serves as a decision support system for socially responsible AI systems, fairness notions were included in the evaluation to detect potential discrimination in the data and prediction space.
  • Zeitschriftenartikel
    Text Mining für Online-Partizipationsverfahren: Die Notwendigkeit einer maschinell unterstützten Auswertung
    (HMD Praxis der Wirtschaftsinformatik: Vol. 54, No. 4, 2017) Liebeck, Matthias; Esau, Katharina; Conrad, Stefan
    Online-Partizipationsverfahren werden in den letzten Jahren vermehrt von Städten und Gemeinden eingesetzt, um ihre Bürger in politische Entscheidungsprozesse einzubeziehen. Der vorliegende Beitrag beginnt mit einer Kategorisierung von Online-Partizipationsverfahren im politischen Kontext in Deutschland und fokussiert auf das Beteiligungsprojekt Tempelhofer Feld in Berlin. Dazu werden die Probleme einer manuellen Auswertung und die Notwendigkeit einer maschinell unterstützten Auswertung von Textbeiträgen aus Partizipationsverfahren beschrieben.Im Beitrag wird auf die Probleme und Lösungsmöglichkeiten in den drei Analysebereichen Argument Mining, Themenextraktion und Erkennung von Emotionen eingegangen. Für den Bereich Argument Mining wird ein geeignetes dreiteiliges Argumentationsmodell, welches auf das Online-Partizipationsverfahren Tempelhofer Feld der Stadt Berlin angewendet wird, diskutiert. Zudem wird der Einsatz von word embeddings als Features für eine Support Vector Machine zur automatisierten Klassifikation von Argumentationskomponenten evaluiert. Anschließend wird ein Einblick in das Aufgabengebiet der Themenextraktion, dessen Ziel die Erstellung eines groben Überblicks über die diskutierten Themen eines Online-Partizipationsverfahrens ist, gegeben und die Ergebnisse zweier Verfahren werden diskutiert. Danach erfolgt eine Diskussion über die Einsatzmöglichkeiten einer automatisierten Emotionserkennung im Kontext von Online-Partizipationsverfahren.AbstractIn recent years cities and municipalities rely increasingly on online participation processes to involve their citizens in political decision-making processes. This paper opens by categorizing political online participation processes in Germany before focusing on one participation project in particular, namely the Tempelhofer Feld in Berlin. In addition, the problems of a manual analysis of text contributions from participation processes are outlined in order to highlight the necessity for automatically supported evaluations.We discuss problems and possible solutions in three areas of analysis: argument mining, topic extraction, and emotion mining. For argument mining, a suitable three-part argumentation model is discussed which is applied to the online participation process Tempelhofer Feld. Furthermore, word embeddings are being evaluated as features for a support vector machine tasked with the automated classification of argumentation components. Subsequently, we focus on topic extraction, which aims to provide a rough overview of the topics discussed in an online participation process, and present the results of two methods. The paper concludes with a discussion on possible applications of an automated recognition of emotions in the context of online participation processes.
  • Zeitschriftenartikel
    The First Data Science Challenge at BTW 2017
    (Datenbank-Spektrum: Vol. 17, No. 3, 2017) Hirmer, Pascal; Waizenegger, Tim; Falazi, Ghareeb; Abdo, Majd; Volga, Yuliya; Askinadze, Alexander; Liebeck, Matthias; Conrad, Stefan; Hildebrandt, Tobias; Indiono, Conrad; Rinderle-Ma, Stefanie; Grimmer, Martin; Kricke, Matthias; Peukert, Eric
    The 17th Conference on Database Systems for Business, Technology, and Web (BTW2017) of the German Informatics Society (GI) took place in March 2017 at the University of Stuttgart in Germany. A Data Science Challenge was organized for the first time at a BTW conference by the University of Stuttgart and Sponsor IBM. We challenged the participants to solve a data analysis task within one month and present their results at the BTW. In this article, we give an overview of the organizational process surrounding the Challenge, and introduce the task that the participants had to solve. In the subsequent sections, the final four competitor groups describe their approaches and results.