A User Study Evaluating Recommender Systems in a Digital Library for Older Adults

• Activity Number: 39
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Computing
• Author(s): Amy K. Atwood* and Adam Maus
• Companies: University of Wisconsin - Madison and University of Wisconsin - Madison
• Keywords: recommender systems ; user engagement ; user study ; evaluation ; older adults ; information and communication technology

Abstract

ElderTree, an information and communication technology designed to address challenges older adults face in maintaining their independence, consists of multiple online services, including a discussion forum, health tracker, route planner, and library of expert-written Active Living Tips. Because the Tips directly address important issues (e.g., falls prevention, medication management, driving) which could lead to reduced independence due to hospital or nursing home admission, we will test if tailoring the Featured Tip shown on each user's homepage can increase user engagement with Tips (e.g., number of: Tips viewed, days viewing Tips). Existing users (194 adults aged 65 plus) will be randomized to the current system, which randomly selects a tip from those the user has not read, or one of two new recommender systems that use k-nearest neighbor user-based collaborative filtering: an implicit feedback system (based on Tips viewed) or an explicit feedback system (based on 1 to 5-star ratings of Tip helpfulness). We will compare systems on user engagement (controlling for prior engagement) and on survey assessments of user perceived system usefulness, usability, and trustworthiness.