Master Projects

Student's Finals

Below is a list of all the Master Projects related to Human-IST Research Center and its team. The "Open Proposals" category contains proposals: if you are interested in a Master Project with Human-IST, please apply for one of these subjects.

 

LateX template for Master Project Report

Open Proposals

  • Resemblance Relations for Cognitive Systems

    To build a cognitive system, it is crucial to develop a memory in the form of a knowledge base from which the system is able to retrieve information as a human being would. A potential approach is to use resemblance relations. With the help of this technique, the system will be able to find similar information, even if at first sight the connection is not immediately obvious. Based on fuzzy logic, it is possible to build knowledge maps (e.g., fuzzy cognitive maps or neural networks) in which nodes would be connected through edges/connectors. Nodes consist of concepts or objects, and edges/connectors illustrate the relationship between nodes. Due to the network-like memory the system will be able to understand what information is similar, and so, by using resemblance relations, the system should be able to match the question with the best possible answer.

    Therefore, this master thesis should achieve the following goals:

    • Conceptual/mathematical description of resemblance relations (e.g., documentation)
    • Elaboration of possible ways to implement resemblance relations within an existing knowledge base (e.g., building a framework/prototype)
    • Testing and evaluating the use of resemblance relations in a knowledge base of a real project

    This master thesis helps to find out if resemblance relations could be applied to optimizing the matching of questions and answers in an existing knowledge database. During the project, the participant will also have the opportunity to gain an insight into projects running at IcT Post and to benefit from the units approach and competences.

    Teachers: 
  • Visualizing the Threads of Legal Articles from a 111-Years Old Constitution

    Between 2019 and 2023, the canton of Valais will be rewriting its constitution, which dates back to 1907. As a fundamental text defining the general principles of the State, the Constitution represents the source of all the cantonal laws promulgated between 1907 and 2018. From a data perspective, the links between constitutional articles and laws thus represent a network of nodes and links.

    The aim of this MSc work is to design an application supported by data visualization that allows the exploration of this network of laws (including data available from parliamentary debates prior the finalisation of legal articles)  through the prism of the 1907 Constitution and from different perspectives, eg. thematic, temporal, etc. 

    In terms of research, this work offers opportunities to contribute to the field of visualization of networks of citations, and visualization of temporal changes in complex networks. Following tasks are envisioned:

    1. State of the art: a list of visual methods and authors, especially focused on citation networks and changes in networks
    2. A visualization taxonomy: a diagram that describes the features that these visual methods have in common and their differences.
    3. Identification of a relevant research question, and setup of research methodology
    4. Implementation and integration of relevant visual methods
    5. Evaluation to comprehend strength and weakness of each method.

    The project is done in collaboration with the Cantonal Archives of the Canton of Valais.

    Teacher: 
  • Evaluating Experience Sampling Techniques in Large Scale Mobile Comfort Perception Sensing

    Maintaining comfort is the purpose of buildings and energy is often needed for it. With more and more severe energy efficiency requirements and ever increasing building automation to achieve it, there is a need to gather the actual comfort sensation of building users, to optimise buidling control.

    Yet comfort sensation is a subjective perception, that shows high variability among users, depending on their activity, their clothing, their environement, etc. The experience sampling approach allows to gather user perception about a given while they go about their normal doings.

    The goal of this master project is to develop further a multi-platform mobile application to allow large scale gathering of comfort perception in real buildings through an experience sampling approach. Building on the outcomes of the project Relationship between subjective comfort perception and smartphone sensor data , the project will bring a similar apporach to a larger scale.

    It will consist in:

    1. Developing further an infrastructure composed of multi-platform mobile app (with availability in app stores) and data servers, possibly using indoor positioning methods relying on BT beacons/Wifi fingerprinting, 
    2. Testing the relevance and usability of several sampling techniques and strategies with real users
    3. Analyse comfort data gathered in large scale
    Teachers: 
  • Matching Smart Home Data Exploitation Results with Users' Real Behaviour

    Home-based digital devices increasingly record data and send it to cloud services (vacuum robot data for sale). Beyond the obvious privacy problem, there are open questions of how this data matches with the self-representation users have of their own behaviour, or even the degree to which the interpretation of the data faithfully represents their actual behaviour.

    In collaboration with a swiss startup which already equipped more than 2000 flats with smart home technology, the project will explore means to exploit the data generated by the occupants interacting with the smarthome application and appliances. Based on electricity, heating, and water consumption and device interaction data, machine learning algorithms will be used to categorize/clusterize occupants' behaviour. These data-centric categories will then be evaluated with the help of qualitative questionnaires to produce personas that faithfully represents users' needs and behaviours.

    During the project, the participant will have the opportunity to integrate the startup team to benefit from their approach and competences.

    Teachers: 
  • Visualisation of the History of ACM IHM Publications

    Context

    This data science project aims at exploring, consolidating, and visualizing the metadata covering 30 years of scientific publications in Human-Computer Interaction at the ACM IHM conference, in prevision for the ACM IHM 2016 Conference that will be held in Fribourg (http://ihm2016.afihm.org/) The dataset produced as a result of this thesis should also be made available online for further use by the research community, along with some visualisations of interesting data.

    The project is advised in collaboration with Dario Rodighiero, data designer at the EPFL.

    The results of the work will be published and presented at the ACM IHM 2016 conference.

    Description

    The tasks to carry out in this project are the following:

    1. State of the art of similar initiatives
    2. Exploration of the available data (based on dataset provided by ACM)
    3. Development of tools to extract / complete relevant metadata
    4. Choice of structure for publishing the dataset
    5. Online publication of the dataset, with documentation
    6. Design and development of relevant visualisations
    7. Evaluation of the visualizations based on specific research question (to be determined)

    As an example of similar work, consider the infovis publications dataset that was visualized e.g. here.

    Technologies

    The dataset should be published online, using standard web technologies. The final visualisations should also be developed for the web, e.g. with the d3.js visualization framework.

    Technology choices for wrangling the data are left to the student (e.g. Python with Pandas, R, ...). For the visual exploration of the data at early stages, the use of Tableau & Gephi is encouraged.

    Teachers: 
  • (Multi)3 Socio-Affective Behaviours

    Context

    Affective and personality computing aim at recognising automatically various emotional states and personality traits from multimodal cues, such as facial expressions, gestures, speech prosody and physiology. It makes it possible to develop systems that convey more natural and user-friendly human-machine interactions, by adapting the behaviour of the machine to the user’s idiosyncrasies. Beside the fact that emotional states and personality traits are highly multimodal, they also strongly depend on the culture of the individuals and may be correlated as well with the language. However, such multimodal, multicultural and multilingual influences on socio-affective behaviours could not have been studied so far with automatic recognition systems, because there is no database that includes all of these criterions.

    The goal of the project is to extend and improve the framework that was developed for the recording of the RECOLA database in order to collect a new corpus of multimodal (audio, visual and physiological), multicultural and multilingual (French, Italian, German) data of naturalistic socio-affective interactions. Annotation of the data will be performed with the existing software ANNEMO. Finally, automatic processing and analysis of the data will be performed using state-of-the-art techniques for both features extraction and machine learning.

    The ideal candidate will have a strong background in informatics (data acquisition, java, Matlab), a strong interest for multimodal data processing and a high motivation.

    Goals

    1. Get familiar with multimodal data acquisition constraints
    2. Adapt the interaction scenario to improve quality of recordings
    3. Recruit participants and collect data (department of psychology)
    4. Perform annotation and analyse the data
    5. Apply state-of-the-art automatic recognition
    Teachers: 
  • Augmented Parallel Coordinates Visualization
    Description

    Parallel coordinates is a popular visualization technique that allows the representation of multidimensional data in a compact way. In particular, it allows analysts to identify correlations between dimensions of the data [1]. Extensions of parallel coordinates have been proposed, e.g. to help identify frequencies of values along each axis of the visualization [3], or to include the display of non-numeric categorical data [4]. However mixed data types are still an issue for parallel coordinates visualizations and several of the proposed improvements have never been actually tested with users. This project aims at bridging this gap.

     

    Overall the goal of this master project are:

    1. Synthetize the research in this area
    2. Create original solutions to improve existing alternatives to parallel coordinates visualisation, in particular for mixed numeric-categorical data
    3. Evaluate several alternatives using crowd-sourced user experiment [2]

     

    References
    1. INSELBERG A., DIMSDALE B.: Parallel coordinates: a tool for visualizing multi-dimensional geometry, 1990. arXiv:arXiv: 1011.1669v3, doi:10.1109/VISUAL.1990.146402. 1
    2. HEER J., BOSTOCK M.: Crowdsourcing Graphical Perception: Using Mechanical Turk to Assess Visualization Design. Proceedings of the 28th Annual Chi Conference on Human Factors in Computing Systems (2010), 203–212. doi:10.1145/1753326.1753357. 4
    3. JOHANSSON J., FORSELL C.: Evaluation of Parallel Coordinates: Overview, Categorization and Guidelines for Future Research. IEEE Transactions on Visualization and Computer Graph- ics 22, 1 (2016), 579–588. URL: http://www.ieee.org/publications{_}standards/publications/rights/ index.html, doi:10.1109/TVCG.2015.2466992. 1
    4. KOSARA R., BENDIX F., HAUSER H.: Parallel sets: Interactive exploration and visual analysis of categorical data. In IEEE Transac- tions on Visualization and Computer Graphics (jul 2006), vol. 12, IEEE, pp. 558–568. URL: http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=1634321, doi:10.1109/ TVCG.2006.76. 1, 2
    Teachers: 
  • Explaining and Encouraging Shifting as a Way to Reduce One’s Carbon Footprint

    When most people are asked ‘How can you reduce your carbon footprint?’, the answer is along the lines of ‘By not doing X, Y and Z, or by doing less of A, B, and C.’ Few people know that the question of CO2 emissions and carbon footprints is much more fine-grained than that, and that it’s in fact possible to reduce carbon emissions simply by doing certain activities at a different time, for example, using electronic devices at times when the electricity comes form renewable resources such as solar panels.

    The aim of the this project will be to use the User Centered Design approach, combined with mental models, to design and implement a number of prototype mobile apps that will help users a) understand the notion and effects of shift rather than reduction, and b) will help users figure out when certain activities can be carried out to reduce the carbon emissions that they generate.

    Concretely, the steps of the project will be:

    • interviews with potential users about the understanding of carbon emissions and shifting, and requirements they might have for an app such as described above
    • design and implementation of 2-3 prototypes
    • evaluation of the prototypes

    Prerequisites: Having taken the User Centered Design course that is part of the Swiss Joint Master in Computer Science, or demonstrating similar knowledge.

    Teachers: 

Ongoing Projects

Year Title Student Supervisor(s)
2017 Autonomous Vehicles and Us : Human-AV Interaction Design Guillaume Pythoud Florian Evéquoz, Grace Eden, Manchester Metropolitan University, Himanshu Verma, Denis Lalanne
2016 Detection and Visualization of Human-Human and Human-Artifact Interactions Using Proximity Sensors Nico Färber Himanshu Verma, Hamed Alavi, Denis Lalanne
2016 Visualizing the User in the Building Data Roberto Sanchez Julien Nembrini, Denis Lalanne

Finished Projects