"Visual Computing Cafe Seminar: Trust, Recommendation, and the User Interface"

John O'Donovan, Research Scientist, CS, UCSB

December 6th (Friday), 12:30pm
Harold Frank Hall 1132 Auditorium

This talk focuses on communicating trust and credibility information on the web, with a particular emphasis on recommender systems. To really understand this topic, we must examine the broad picture of mining, tailoring, representing and communicating information on the web that can impact on a trust-decision. This includes design and evaluation of interactive web-based interface mechanisms for controlling complex algorithms such as recommender systems (RS). I will discuss ways that our research integrates methods, ideas and practices from diverse research areas such as UI and Interaction design, HCI and the behavioral/cognitive sciences, for the purpose of understanding trust in complex information spaces. I will talk about the roles that factual information, in tandem with user-specific subjective metrics such as trust and competence can play in the process of recommendation, particularly in emerging “social” recommender systems –those where predictions are made from groups with pre-existing social connections. Visualization and the user interface also plays an important role in these systems, since background information on a recommendation can reveal potentially large and impactful amounts of pre-existing knowledge from the a-priori social relationship.

About John O'Donovan:

photo of john o'donovan John is a research scientist in the Department of Computer Science, UCSB. He received his PhD from University College Dublin in 2008. John has published more than 40 technical papers and has successfully secured external research funding from sources such as ARL, KDD and NSF over the past several years. He currently leads a team of computer scientists and cognitive science researchers from CMU, UIUC, and ARL as part of a 5-year network science project. His well cited thesis work on “Trust in recommender systems” was nominated for a national doctoral dissertation award in Science and Engineering, and his most recent work on information credibility won a best paper award at IEEE SocialCom 2013. John is an active Editor, PC member and reviewer for more than 10 conferences, workshops, and journals. He is a member of ACM and IEEE.

Hosted by: UCSB Center for Visual Computing