Title: Analyzing Right-wing YouTube Channels: Hate, Violence and Discrimination
Brief Summary:
In this paper, by using YouTube, they observed issues related to hate, violence, and discriminatory bias in a dataset containing more than 7,000 videos 17 million comments. They investigate similarities and difference between users comments and video content in a selection of right-wing channels and compare it to a baseline set using a three-layered approach, in which its analyzed with (a) lexicon, (b) topics and (c) implicit biases present in the texts. Among the results, the analyses show that right-wing channels tend to (a) contain a higher degree of words from “negative” semantic fields, (b) raise more topics related to war and terrorism, and (c) demonstrate more discriminatory bias against Muslims (in videos) and towards LGBT people (in comments).
Reflection:
There were some quotes that really stood out to me:
Quote 1: “As stated in a The Guardians article [24], “The Alex Jones Channel, the broadcasting arm of the far-right conspiracy website InfoWars, was one of the most recommended channels in the database of videos” used in a study which showed that YouTube’s recommendation algorithm was not neutral during the presidential election of 2016 in the United State of America [23, 25]”
Reflection 1: I found this very interesting because, despite the mention of the possibility that YouTube showed bias during the 2016 presidential campaign, two YouTube channels were still included in the list of “Baseline channels”, including “YouTube Spotlight”, and “YouTube Spotlight UK”. I found this highly irregular, as I don’t believe these YouTube channels should have been included in the baseline in a political analysis paper, as they mentioned that there was a possibility that YouTube has a past of showing political bias, as it weakens the accreditation of these selected “Baseline channels”.
Quote 2: “From the 194 total Empath categories, we selected the following (a) 15 categories related to hate, violence, discrimination and negative feelings, and (b) 5 categories related to positive matters in general:
negative: aggression, anger, disgust, dominant personality, hate, kill, negative emotion, nervousness, pain, rage, sadness, suffering, swearing terms, terrorism, violence.
positive: joy, love, optimist, politeness, positive emotion”
Reflection 2: I found this information to be misleading and not entirely complete. First, some of the words “selected” don’t necessarily fit the category that they are in, for example, I don’t see how “nervousness” is negative, as it can also be seen as excitement or anxious. Also, I believe that they should have selected the same number of negative words as positive words. With having two data sets used for the same kind of measure, they should try and include the same number of words in these data sets, as it might appear that you’re trying to fit the data to match your hypothesis or in an attempt to show bias towards the results. It is also unclear of how they came up with the number of words to use from the data set, or why they chose the words that they did.
Future Work:
I think that it would be interesting to do an analysis on left-wing YouTube channels. I think it would be interesting comparing the “hateful” vocabulary used on captions and comments on these channels before and after the 2016 presidential election due to the election of President Trump. I would believe that the left-wing YouTube channels would currently show the same results as having a high usage of “hateful” vocabulary.
Also, I agree with the future work that needs to be done in this paper that is mentioned in the conclusion section, with analyzing more of the negations on these so-called “negative” words. For example, a caption like “End of Terrorism?” would probably show up as negative due to the mention of “terrorism”, however, without more research into the content itself, that classification of “negative” might be misleading.
