Dataset Information

Tubes and bubbles topological confinement of YouTube recommendations.

ABSTRACT: The role of recommendation algorithms in online user confinement is at the heart of a fast-growing literature. 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). 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. Starting from a diverse number of seed videos, we first describe the properties of the sets of suggested videos in order to design a sound exploration protocol able to capture latent recommendation graphs recursively induced by these suggestions. These graphs form the background of potential user navigations along non-personalized recommendations. From there, be it in topological, topical or temporal terms, we show that the landscape of what we call mean-field YouTube recommendations is often prone to confinement dynamics. Moreover, the most confined recommendation graphs i.e., potential bubbles, seem to be organized around sets of videos that garner the highest audience and thus plausibly viewing time.

SUBMITTER: Roth C

PROVIDER: S-EPMC7173795 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Tubes and bubbles topological confinement of YouTube recommendations.

Roth Camille C Mazières Antoine A Menezes Telmo T

PloS one 20200421 4

The role of recommendation algorithms in online user confinement is at the heart of a fast-growing literature. 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). 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 c ...[more]

PMID: 32315329

Similar Datasets

Project description:ObjectivesThis study aimed to evaluate content quality and racial/ethnic representation, particularly of high-risk cohorts, of prostate cancer screening videos on YouTube (YT) and TikTok (TK).Materials and methodsThe top 50 videos populated for the search term 'prostate cancer screening' on YT and TK that met inclusion criteria were retrieved in a cache-cleared browser. Three reviewers analysed all videos using validated criteria for the quality of consumer health information (DISCERN and Patient Education Materials Assessment Tool [PEMAT]). High quality was defined as follows: DISCERN ≥ 4, PEMAT understandability ≥75% and PEMAT actionability ≥75%. A 5-point Likert scale was used to demonstrate the level of misinformation compared to American Urological Association and National Comprehensive Cancer Network guidelines. Perceived race and ethnicity of people in the videos were assessed by consensus approach.ResultsTK videos were shorter (median 3.7 vs. 0.5 min, p < 0.001) and had more views per month (5437.5 vs. 19.3, p = 0.03) than YT videos. Perceived Black and Hispanic representation was present in 10% and 6% of YT videos and 20% and 12% of TK videos, respectively. High-risk racial/ethnic groups were explicitly discussed in 46% of YT videos and 8% of TK videos. A total of 98% of YT videos and 100% of TK videos had low- to moderate-quality consumer health information, and 88% of YT videos and 100% of TK videos had moderate to high levels of misinformation based on screening guidelines.ConclusionsYT and TK videos about prostate cancer screening are widely viewed but do not provide quality consumer health information. Black and Hispanic men remain under-represented on both platforms, and high-risk racial groups were not discussed in most videos despite the importance for screening criteria. The low understandability and actionability, significant misinformation and lack of diversity in online videos support the need for higher quality videos with adequate attention to high-risk ethnic cohorts.

Project description:Objectives:The Internet in general, and YouTube in particular, is now one of the most popular sources of health-related information. Pain neuroscience education has become a primary tool for managing persistent pain, based in part on the discovery that information about pain can change pain. Our objective was to examine the availability, characteristics, and content of YouTube videos that address the neuroscience of pain. Methods:We conducted a systematic review of videos on YouTube using the search terms "pain education", "what is pain", and "pain brain" in January 2018. Videos were included if they were in English, were under 10 minutes long, and included information on the neuroscience of pain. Videos were coded for (i) descriptive characteristics (e.g., number of views, duration on YouTube), (ii) source and style, (iii) whether or not they addressed seven pre-determined target concepts of pain neuroscience education (e.g., 'Pain is not an accurate marker of tissue state'), and (iv) how engaging they were. Results:We found 106 unique videos that met the inclusion criteria. The videos ranged from having four views to over five million views (Mdn = 1,163 views), with the three most highly viewed videos accounting for 75% of the total views. Animated videos were much more highly viewed than non-animated videos. Only a small number of videos had been posted by a clearly-identifiable reputable source such as an academic or medical institution (10%), although a number of videos were posted by healthcare professionals and professional medical societies. For a small number of videos (7%), the source was unclear. We found 17 videos that addressed at least one target concept of pain neuroscience science education, only nine of which were considered to be at least somewhat engaging. The target concept 'Pain is a brain output' was considered to be well addressed by the most videos (N = 11), followed by 'Pain is a protector' (N = 10). We found only one video that adequately addressed all seven target concepts of pain neuroscience education. Discussion:YouTube contains a variety of videos that practitioners, patients, and families may view to access pain neuroscience education information. A small portion of these videos addressed one or more target concepts of pain neuroscience education in an engaging manner. It is yet to be determined to what extent patients are able to learn information from these videos, to what extent the videos promote behavior change, and thus to what extent the videos may be useful for clinical practice.

Dataset Information

Tubes and bubbles topological confinement of YouTube recommendations.

Publications

Tubes and bubbles topological confinement of YouTube recommendations.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets