Browsing by Subject "recommender systems"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Understanding How People Use Natural Language to Ask for Recommendations: Query Dataset
    (2017-06-29) Kang, Jie; Condiff, Kyle; Chang, Shuo; Konstan, Joseph A; Terveen, Loren; Harper, F Maxwell; max@umn.edu; Harper, F Maxwell; GroupLens Center for Social and Human-Centered Computing; University of Minnesota Department of Computer Science and Engineering
    This dataset describes subjects' initial and follow-up queries from the research paper "Understanding How People Use Natural Language to Ask for Recommendations", published in the ACM Conference on Recommender Systems (RecSys), 2017. The data were collected on movielens.org between May 12 and May 24, 2016.
  • Thumbnail Image
    Item
    User-Centric Design and Evaluation of Online Interactive Recommender Systems
    (2018-05) Zhao, Qian
    User interaction is present in all user interfaces including recommender systems. Understanding user factors in interactive recommender systems is important for achieving better user experience and overall user satisfaction. Many prior works in recommender systems consider recommendation as a content selection process and there is not much prior work focusing on studying user interaction, except user on-boarding interaction design, rating interface design etc. Even for the content selection part, however, it seems obvious that there are a fair amount of factors lying in the scope of user interaction as well, to name a few, visual attention and item exposure, perceived temporal change, reactivity, confusion; i.e., factors regarding content browsing in a typical information system. My research studies several factors while real users are interacting with online recommender systems and answers a series of questions regarding those factors. Specifically, my research focuses on gaining a better understanding on a) whether users pay attention to grids of recommendations displayed in modern recommender interfaces; b) how to interpret and infer user inaction after we show those recommendations to users and further utilize this inaction model to improve recommendation; c) how to organize and present the top-N recommendations to better utilize user attention and increase user engagement; d) how does recommenders optimizing for being engaging (i.e., as many user interactions as possible) affect user experience compared with recommenders optimizing for being right in estimating user preference and maximizing the preference of users on recommendations displayed; e) how to better support work that combines user-centric design, evaluation and building complex, scalable recommendation models going from offline settings into the online environments of providing interactive real-time responses to user recommendation requests, by building a generic recommender server framework.