Researchers dissect algorithm curation of video recommendations on YouTube
With nearly two billion active users roaming its platform, the video-sharing giant YouTube is regarded as one of the most visited sites on the web. Yet, user reactions of the platform’s algorithmic curation of recommendations have yielded mixed reception.
In a recent study, published in PLOS One, a research coalition affiliated with the French National Centre for Scientific Research explored the algorithmic curation of video recommendations on the social networking site.
For some users, the presence of a filter bubble during their venture of video viewing habits inhibited diversity and serendipity. This argument of online user confinement has been the subject of many past studies.
“Recent empirical studies generally suggest that filter bubbles may principally be observed in the case of explicit recommendation (based on user-declared preferences) rather than implicit recommendation (based on user activity),” the study’s co-authors explained in the findings.
“We focus on YouTube which has become a major online content provider but where confinement has until now been little-studied in a systematic manner. We aim to contribute to the above literature by showing whether recommendation on YouTube exhibits phenomena typical of filter bubbles, tending to lower the diversity of consumed content.”
For the study, researchers examined a substantial number of seed videos, focusing on the sets of recommendations based on the platform’s algorithmic curation.
The findings showed that YouTube recommendations are generally prone to confinement dynamics. Purported filter bubbles were evident among sets of videos organized based on viewing retention and total audience count.
“On the whole, the analysis of the graphs we extracted nonetheless demonstrate the diversity of navigation anisotropy on YouTube in a variety of dimensions. They also suggest that the most confined graphs i.e., potential bubbles, are organized around videos that garner the highest audience and plausibly viewing time.”
“Future work should certainly appraise a variety of other modes of recommendation (such as personalized suggestions), other types of behavior (such as organic navigation, whereby users search for videos by themselves) and a mix thereof (such as browsing on subscription-based channels),” the study’s co-authors concluded.