GR User Interaction with Recommender Systems

Recommender systems recommend movies, restaurants or other items to an active user based on information about users and items in various application domains. For example, the domain of travel and tourism is more complex than others, because the products to be recommended have many facets and context - such as the current time and location - plays a larger role.  Research in this area does not only have to consider the underlying algorithms, but also how users interact with the systems. User interaction is mostly concerned with submitting initial preferences and giving feedback on suggested items [1].

The goal of this Guided Research in the Informatics (or related) Master’s program (IN2169) is to investigate improved solutions for user interaction in recommender system applications. The actual focus can be discussed and will be specific, options include:
- Adapting heuristic evaluation methods for early-stage recommender systems user interfaces
- Comparing different methods for preference elicitation, e.g. questionnaire with features vs. pairwise comparison of pictures
- Investigating the trade-off between controllability (e.g. showing more features and options) and cognitive load
- User interface issues in the decision-making process of users, e.g. whether how to present recommended items does change the users' choice satisfaction
- Explanations for sequence-aware tourism recommender systems
- Novel approaches for visualization of recommendation results, including map-based options
- Comparing different methods for user feedback on recommended items, e.g. rating-based vs. critiquing items
- Investigating distributed and migratory user interfaces

The proposed course of action is in general as follows:
* Overview of the state-of-the-art, both in existing applications and research literature
* Designing a solution, which can be very focused but should include some novel aspect(s)
* Implementing a prototype to allow test users to interact with the system (but no backend or actual recommender system needed)
* Conducting a user study using an appropriate method, comparing the proposed solution(s) with a baseline

The project has to be documented in a brief scientific report that can possibly be submitted to a conference or workshop. The interaction prototype can be developed in any platform, for example Web based (e.g. with React) or mobile with Android or iOS. Prerequisites are high motivation and good programming skills in the selected mobile platform. Please send your application (brief CV, transcript of records and short motivation statement) to Wolfgang Wörndl (woerndl@in.tum.de). The start of the project is flexible, but a guided research module has to be formally registered by the first week of a semester.

[1] M. Jugovac, D. Jannach: Interacting with Recommenders - Overview and Research Directions. ACM Transactions on Interactive Intelligent Systems (TiiS), 7 (3), 2017