Suggested Theses

 

 

KNOWLEDGE DISCOVERY
  • Bookmark Search (Bachelor Thesis / Master Project)

    Task: Develop a browser add-on that crawls all bookmarked web pages and indexes the content and provide a search interface.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Evaluation of Annotation Strategies (Bachelor Thesis)

    Task: Evaluate and compare two ways to annotate textual documents: i) visual, and ii) KWIC style.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Automatic Generation of Regular Expressions (Bachelor Thesis / Master Project)

    Task: The task is to automatically generate regular expressions from lists of different types of names, e.g. product names. For example for names of countries such a regular expression could be found: “\w+land”.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Text Segmentation via Bayesian Inference (Bachelor Thesis / Master Project)

    Task: The goal of text segmentation is to split a long text into smaller segments. In this thesis Bayesian inference should be used to achieve such a segmentation.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Evaluation of an Existing Ensemble Classifier (Bachelor Thesis / Master Project)

    Task: Evaluate the quality of an existing ensemble classifiers on a number of data-sets.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Dataset Replication / Anonymisation (Bachelor Thesis / Master Project)

    Task: Develop a tool to replicate a data-set without actually replicating the content, but replicating the key characteristics.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Weka Plugin for Hypothesis Generation (Bachelor Thesis / Master Project / Master Thesis)

    Task: Develop a plugin for Weka that automatically generated hypothesis from a given data set.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Actor Based Simulation Framework (Master Thesis)

    Task: Development of a actor based simulation of the movement of people, based on Open Street Map and Wikipedia demographics information.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Stacked Classifiers (Master Thesis)

    Task: Implement a stack of classifiers and evaluate the usefulness (ensemble classifiers).
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Sensor Data Framework (Master Thesis)

    Task: Develop an online platform for the collection of distributed sensor data.
    For more information please visit: Further Information
    Contact: Roman Kern, rkern@know-center.at

  • Validation of Information (Bachelor Thesis / Master Project)

    Task: Evaluation and implementation of information quality measures (i.e. redundancy) for content on Web pages.
    For more information please visit: Further Information
    Contact: Mark Kröll, mkroell@tugraz.at

  • Business Intent (Bachelor Thesis / Master Project)

    Task: Implement an algorithm to extact goals of companies from Web pages and to compare different company goals.
    For more information please visit: Further Information
    Contact: Mark Kröll, mkroell@tugraz.at

  • Open Information Extraction for German (Master Thesis)

    Task: Evaluation of the state-of-the-art in the area of Open Information Extraction and a prototypical implementation of such a system for the German language.
    For more information please visit: Further Information
    Contact: Mark Kröll, mkroell@tugraz.at

  • Language Agnostic Information Extraction (Master Thesis)

    Task: Evaluation of cross-lingual and language agnistic methods for Information Extraction and an implementation of such a method for text written in English and German.
    For more information please visit: Further Information
    Contact: Mark Kröll, mkroell@tugraz.at

Back to Top

 

KNOWLEDGE VISUALISATION
  • Visualization Toolbox (Master Thesis) New topic available

    In this work you will develop a Visualization Toolbox which is an application that allows users to discover, explore, and visually analyze datasets from different sources, e.g. publications from digital libraries such as ACM and IEEE, cultural content from services such as Europeana, or the local file system. The Toolbox relies on semantic data models (vocabularies) and their integration (mapping) to provide foundations for automatically generating and configuring suitable visualizations. In a nutshell, the role of the Toolbox is to support automated process of providing visualizations for exploring and analyzing data sets from different sources.
    For more information please visit:
    Further Information
    or contact:vsabol@know-center.at

  • Tracking User Behavior for Content-based Recommendations (Master Thesis) New topic available

    We provide a Recommendation dashboard that includes several interactive visualizations that is connected to the Content-based Recommender System (CB-RS). In this work you will implement an algorithm which tracks and collects users’ interaction with the visualizations and which defines behavioral patterns based on those collections. The patterns will be used to infer users’ next action using CB-RS and in turn to recommend visualizations which address users’ task in the best possible way.
    For more information please visit:
    Further Information
    or contact:vsabol@know-center.at

  • Adaptive time visualisations for recommendation data (Bachelor Thesis)

    In this work you’ll develop time visualisations that present an overview of recommended items and their distribution in time (e.g., happenings that lead to a particular event). The time visualisation should adaptively compress to accommodate restricted screen space, and interactively expand when the user needs to explore data in depth. We are looking for a method that makes the best use of the available screen space at each stage. The data visualised comprises recommendations of cultural, educational, and scientific nature, that should catch the attention of the user while reading, surfing, blogging,etc. You will have to deal with Web-based visualisation technologies, the decision on which technology to use for implementation is yours.
    For more information or to suggest your own topics please contact:
    Eduardo Veas, eveas@know-center.at

  • Scalable Web-based Information Visualization using WebGL and HTML5 (Bachelor Thesis)

    In this work you will develop selected information visualization components using WebGL and HTML5 technologies. WebGL is a standardized JavaScript API based on OpenGL ES which, as a subset of the full OpenGL specification, provides access to hardware accelerated rendering of interactive 3D and 2D graphics within a Web browser (without any plug-ins). You will beging by performing an analysis of available WebGL-based visualization frameworks. After that a decision will be taken together with your advisor on whether an existing framework will be used or not. Following that, you will start with main task, which is the implementation of selected visual representations, such as a stream graph, hierarchical timeline, scatterplot matrices, coocurrence matrices, sankey diagramms etc. for large scale data. The goal is to achieve a higher scalability compared to what available SVG or canvas-based implementations support. The resulting visualization components shall handle data sets containing well over 10000 data elements, ranging potentially into millions.
    For more information or to suggest your own topics please contact:
    Vedran Sabol, vsabol@know-center.at

  • 3D Knowledge Visualisation using WebGL (Master Thesis)

    In this work you will develop a 3D knowledge visualization component using WebGL, which a JavaScript API based on OpenGL ES providing access to hardware accelerated 3D rendering within a Web browser. Knowledge Visualisation is a discipline dealing with presenting and communicating knowledge through visual interfaces. Your tasks will include the implementation and evaluation of a visualization component for visualising 3D models, such as a building, electronic device or a car, and enriching these with knowledge originating from different sources. For example, encoding temperature information in a building using color, superimposing a sequence of usage instructions on a device, or showing technical information (dimensions, rotation direction, maintenance interval etc.) of various car components. You will develop approaches for retrieving information (such as numeric, textual, semantic or time-dependant data) from the knowledge bases and mapping it onto appropriate visual properties of the 3D model. Finally you will perform an evaluation involving several test users to eliminate usability issues and identify directions for further improvements.
    For more information or to suggest your own topics please contact:
    Eduardo Veas, eveas@know-center.at

  • Topical Landscape Visualisation (Master Thesis)

    In this work you will develop methods for visualizing document collections using the topical landscape visual metaphor. A topical landscape resembles a geographic map where text documents are placed in such a way that similar ones are close to each other, while dissimilar ones are far apart – the so called “similarity layout”. The resulting landscape consist of peaks representing agglomerations of thematically similar documents. Your tasks include one or more of the following (TBD): i) increasing the quality of algorithms for computing the similarity layout, ii) development of a scalable Web-based thematic landscape visualization component iii) implementing and evaluating novel interaction and exploration mechanisms for the landscape. Depending on this choice your development work may target a server component for number crunching and/or a client visualization component. The client component shall be realized using Web technologies (JavaScript, HTML5 Canvas or WebGL).
    For more information or to suggest your own topics please contact:
    Vedran Sabol, vsabol@know-center.at

Back to Top

 

SOCIAL COMPUTING
  • Analysis of a Collaborative Health Information System

    Task: to crawl an online health community and analyze social network aspects (e.g., identification of groups of interests and experts) or various aspects of information quality (e.g., conflicting statements, trust) with the help of statistical and/or semantic technologies.
    For more information or to suggest your own topics please contact:
    Elisabeth Lex, elex@know-center.at

    —–

    Analyse eines Collaborative Gesundheitsinformationssystem

    Aufgabe: ein Online-Gesundheits-Community zu crawlen und soziale Netzwerkaspekte (zB Identifikation von Interessengruppen und Experten) oder verschiedene Aspekte der Informationsqualität (zB widersprüchliche Aussagen, Vertrauen) mit Hilfe von statistischen und/oder semantischen Technologien zu analysieren.
    Für weitere Informationen oder eigene Themen vorzuschlagen, kontaktieren Sie bitte:
    Elisabeth Lex, elex@know-center.at

  • Analyzing interaction patterns in Online communities

    Task: Analyze interaction patterns in Online communities in respect to sentiment and social support.
    For more information or to suggest your own topics please contact:
    Elisabeth Lex, elex@know-center.at

  • Information Quality using SNA in Online communities and Social Media

    Task: Use methods and tools from the field of Social Network Analysis to investigate information quality aspects in Online communities and Social Media (e.g. analysis of connectivity structures, trust, …)
    For more information or to suggest your own topics please contact:
    Elisabeth Lex, elex@know-center.at

  • Predicting Event Participation from Social Networks and Media (Master Thesis)

    Task: Social Media and Networks bear great potential to predict crowd behavior. This was shown in many contexts in the past such as for instance in detecting earth quakes, predicting elections, etc. The goal of this project is a similar one but focuses on a different kind of problem namely known as predicting event participation from social media and networks. The goal of this project is to analyze large scale social media datasets (over time) obtained for instance from Social Media platforms such as Twitter and to implement a predictive model (based on the observations made) that is able to forcast event participation with high accuracy.
    For more information or to suggest your own topics please contact:
    Christoph Trattner, Know-Center, ctrattner@know-center.at

  • Where will we be going next? Predicting crowd behavior from Dwolla & Foursquare (Master Thesis)

    Task: Location-based social service platforms such as Dwolla have gained tremendously in popularity recently. As shown these data sources bear a great potential to predict not only social interactions between user but also tie strength such as partnership. The goal of this project is to compare different sources of location-based social networks derived for instance from Dwolla Check-ins or Foursquare with each other and to implement a high accurate predictive model that is able to recommend places to users which they most likely want to visit in the near future.
    For more information or to suggest your own topics please contact:
    Christoph Trattner, Know-Center, ctrattner@know-center.at

  • Utilizing Social Stream Data to predict purchases in online marketplaces (Master Thesis)

    Task: Social data – such as for instance Facebook likes – bear a great potential to increase the predictive power of Online Recommender engines, as for instance used in Amazon. The goal of this project is to implement a novel social recommender engine for online marketplaces by utilizing not only the users likes but also the users social streams.
    For more information or to suggest your own topics please contact:
    Christoph Trattner, Know-Center, ctrattner@know-center.at

Back to Top

 

UBIQUITOUS PERSONAL COMPUTING
    • Ubiquitous sensing for (collaborative) working, learning and creativity processes / Ubiquitäre Sensortechnik für (kollaborative) Arbeits-, Lern- und Kreativitätsprozesse
    • Human-computer interaction for (collaborative) working, learning and creativity processes / Mensch-Computer-Interaktion für (kollaborative) Arbeits-, Lern- und Kreativitätsprozesse

Current topics and contact/Aktuelle Themen und Kontakt: Viktoria Pammer

 

COGNITIVE SCIENCE
      • Implementierung statistischer Verfahren zur Analyse von Evaluationsdaten (Bachelor Arbeit oder Master Projekt)

        Verschiedene statistische Analyseverfahren sollen implementiert werden, die Evaluationsdaten (Fragebögen, Logdaten, Navigationsverhalten, etc) statistisch auswerten.
        Kontakt: Alexander Nussbaumer, alexander.nussbaumer@tugraz.at oder Simone Kopeinik, simone.kopeinik@tugraz.at
        —–

        Implementation of Statistical Methods for Evaluation Data Analysis (Bachelor Theses or Master Project)

        Different statistical methods should be implemented that analyze evaluation data, such as questionnaires, log data, user navigation data, etc.
        Contact: Alexander Nussbaumer, alexander.nussbaumer@tugraz.at or Simone Kopeinik, simone.kopeinik@tugraz.at

      • Erstellen eines Learning Analytics Plugins in Moodle (Bachelor Arbeit oder Master Projekt)

        Basierend auf einer Moodle Erweiterung, die im Rahmen eines Forschungsprojekts (INNOVRET) erstellt wurde, ist es die Aufgabe, ein Plugin zu entwickeln, das Lernerdaten ausertet und graphisch anzeigt.
        Kontakt: Alexander Nussbaumer, alexander.nussbaumer@tugraz.at oder Simone Kopeinik, simone.kopeinik@tugraz.at
        —–

        Development of a Learning Analytics Plugin for Moodle (Bachelor Thesis or Master Project)

        Based on a Moodle extension that has been developed in a research project (INNOVRET), a further plugin should be developed that analyses learner data and visualises them
        Contact: Alexander Nussbaumer, alexander.nussbaumer@tugraz.at or Simone Kopeinik, simone.kopeinik@tugraz.at

      • Implementierung verschiedener Recommender Strategien (Bachelor Arbeit oder Diplomarbeit)

        Basierend auf Lernerdaten (inkl. Social Networks), Kontentinformationen und konzeptioneller Modelle soll eine Java Library erstellt werden, die verschiednen Recommender Ansätze implementiert.
        Kontakt: Alexander Nussbaumer, alexander.nussbaumer@tugraz.at oder Simone Kopeinik, simone.kopeinik@tugraz.at
        —–

        Implementation Various Recommender Strategies (Bachelor Thesis or Master Thesis)

        Based on learner data (including social networks), content information, and conceptual models, a Java library should be developed that implements various recommender approaches.
        Contact: Alexander Nussbaumer, alexander.nussbaumer@tugraz.at or Simone Kopeinik, simone.kopeinik@tugraz.at

Back to Top

 

NETWORK SCIENCE
      • Link Navigation in Recommender Systems

        Task: Taking the content of a recommender system (using recommendation data for movies, books or music artists) as nodes and the recommendations as links, we can create a network. Given this network, we can then try to answer several questions, such as: Is the network connected? What degree distribution does it have? How does it evolve when we add more links? In this work, we will try to answer these questions as well as looking at the navigability of these systems – i.e., how well can users reach parts of the network by following recommendations?
        For more information or to suggest your own topics please contact:
        Daniel Lamprecht, daniel.lamprecht@tugraz.at

        —–

        (Link-)Navigation in Empfehlungsnetzwerken

        Aufgabe: Die Knoten (z.B. Filme, Bücher, Interpreten) in einem Recommender-System bilden zusammen mit den Empfehlungen als Kanten ein Empfehlungsnetzwerk. Anhand eines solchen Netzwerks, können wir verschiedenen Fragestellungen nachgehen, etwa: Ist das Netzwerk zusammenhängend? Welche Gradverteilung weist es auf? Wie entwickelt es sich, wenn Link hinzukommen? In dieser Arbeit werden wir diesen Fragestellungen nachgehen, und uns darüber hinaus auch speziell mit der Navigierbarkeit dieser Netzwerke beschäftigen – d.h. wie gut können User darin anhand der Empfehlungen Dinge auffinden?
        Für weitere Informationen oder um eigene Themen vorzuschlagen:
        Daniel Lamprecht, daniel.lamprecht@tugraz.at

      • Topic-biased PageRank

        Task: PageRank is an algorithm to rank pages based on the “pages are important if important pages link to them” principle. In this work, wewill explore a topic-biased version of PageRank, where rank is determined not only by links but also by page content. We will explore the factors important for high topic-biased PageRank (such as the relation to the topic or the distribution of links) on a set of Wikipedia articles.
        For more information or to suggest your own topics please contact:
        Daniel Lamprecht, daniel.lamprecht@tugraz.at

        —–

        Topic-biased PageRank

        Aufgabe: PageRank ist ein Algorithmus zur Reihung von Websites, und basiert auf dem Prinzip “Seiten sind wichtig, wenn wichtige Seiten auf sie verlinken.” In dieser Arbeit werden wir eine inhaltsbezogene Variante von PageRank untersuchen, bei der die Reihung auch durch den Seiteninhalt bestimmt wird. Wir werden verschiedene Faktoren für gutbewertete Seiten untersuchen (z.B. Zusammenhang mit dem gewünschten Thema oder die Verteilung der Links), und dabei eine Menge von Wikipediaartikeln als Datensatz verwenden.
        Für weitere Informationen oder um eigene Themen vorzuschlagen:
        Daniel Lamprecht, daniel.lamprecht@tugraz.at

      • Design, simulation and evaluation of team-meetings

        Task: Nearly in every field people are working together in teams. Therefore meetings have become very important in the last few decades. Related to that we ask our self how these meetings should be designed to increase performance and efficiency of teams. Should meetings only be used to apportion work or should team members also transfer newly gained knowledge to all other members? Which new gained knowledge should be shared with other members and what role does the structure of the team (i.e.: ratio of specialists and generalists in the team) play in meetings? From another point of view: Which type of meetings fits perfectly for different structured teams? To answer all this questions several types of meetings need to be designed, simulated and evaluated.
        For more information or to suggest your own topics please contact:
        Florian Geigl, florian.geigl@tugraz.at

        —–

        Design, Simulation und Evaluierung von Meetings

        Aufgabe: In fast allen Bereichen wird in Gruppen gearbeitet. In den letzten Jahrzehnten sind dadurch Meetings immer wichtiger geworden. Es stellt sich nun die Frage wie Meetings abgehalten werden sollen um die Performance und Effizienz des Teams zu steigern. Sollen Meetings nur zur Arbeitsaufteilung oder auch um neu erworbenes Wissen mit Teamkollegen zu teilen verwendet werden? Welches neu erworbene Wissen macht Sinn an Kollegen weitergegeben zu werden? Welche Rolle spielt in diesem Zusammenhang die Struktur des Teams (z.B.: das Verhältnis von Spezialisten und Generalisten im Team)? Um diese Fragen beantworten zu können, müssen verschiedene Meetings designt, simuliert und ausgewertet werden.
        Für weitere Informationen oder um eigene Themen vorzuschlagen:
        Florian Geigl, florian.geigl@tugraz.at

      • Correlation of Network structure and content

        Task: Wikipedia is probably the most common information system in the world. People use it every day to find answers to several questions or to extend their personal knowledge. To find a specific answer in such a system, people need to navigate through the underlying network, where each article represents a node whereas links between articles connect two nodes. Different studies have shown that people are extremely efficient in navigating through this type of networks. Thus we think that there is a correlation between network structure and content of the articles. To investigate whether this correlation exists or not, different network and content analysis have to be performed. A starting point will be the examination of network structure communities in connection to content communities/categories.
        For more information or to suggest your own topics please contact:
        Florian Geigl, florian.geigl@tugraz.at

        —–

        Korrelation zwischen Netzwerkstruktur und Inhalt

        Aufgabe: Die Wikipedia ist wahrscheinlich das bekannteste Informationsnetzwerk weltweit. Menschen verwenden sie täglich um Antworten auf Fragen zu finden oder um ihr persönliches Wissen zu erweitern. Um eine Antwort auf eine spezielle Frage zu finden, muss man durch das zugrunde liegende Netzwerk, wo jeder Artikel einen Knoten und jeder Link eine Verbindungen zwischen zwei Knoten darstellt, navigieren. Verschiedene Studien haben bereits gezeigt, dass Menschen extrem effizient im Navigieren durch solche Art von Netzwerken sind. Daher denken wir, dass ein Zusammenhang zwischen Netzwerkstruktur und Inhalt der Seiten existiert. Um herauszufinden ob diese Korrelation existiert, müssen verschiedene Netzwerk- und Inhaltsanaylsen durchgeführt werden. Ein Einstiegspunkt wird sein, den Zusammenhang zwischen Netzwerk-Communties und Inhalt-Communities/Kategorien zu untersuchen.
        Für weitere Informationen oder um eigene Themen vorzuschlagen:
        Florian Geigl, florian.geigl@tugraz.at

Back to Top