Viewpoint Diversity in Search Results
Tim Draws (TU Delft - Web Information Systems)
Nirmal Roy (TU Delft - Web Information Systems)
Oana Inel (Universitat Zurich)
A. Rieger (TU Delft - Web Information Systems)
Rishav Hada (Microsoft Research)
Mehmet Orcun Yalcin (Independent researcher)
Benjamin Timmermans (IBM Benelux)
N. Tintarev (Universiteit Maastricht)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Adverse phenomena such as the search engine manipulation effect (SEME), where web search users change their attitude on a topic following whatever most highly-ranked search results promote, represent crucial challenges for research and industry. However, the current lack of automatic methods to comprehensively measure or increase viewpoint diversity in search results complicates the understanding and mitigation of such effects. This paper proposes a viewpoint bias metric that evaluates the divergence from a pre-defined scenario of ideal viewpoint diversity considering two essential viewpoint dimensions (i.e., stance and logic of evaluation). In a case study, we apply this metric to actual search results and find considerable viewpoint bias in search results across queries, topics, and search engines that could lead to adverse effects such as SEME. We subsequently demonstrate that viewpoint diversity in search results can be dramatically increased using existing diversification algorithms. The methods proposed in this paper can assist researchers and practitioners in evaluating and improving viewpoint diversity in search results.