Early Public Responses to the Zika-Virus on Youtube: Prevalence of and Differences Between Conspiracy Theory and Informational Videos
This paper aimed to analyze the public response to different YouTube videos concerning the Zika virus. Conspiracy theory and informational videos were analyzed. The paper defined a conspiracy theory as “explanations for important events that involve secret plots by powerful and malevolent groups”. Since the spread of conspiracy theory videos is harmful because they contain misleading or untrue information and distract from important health messages, the researchers looked to analyze the sentiment and content of user responses so that implications for online health campaigns and knowledge dissemination are known. False news, of which conspiracy videos fall under, has been shown to spread faster and father than true news. The researchers used 35 of the most popular Zika-virus YouTube videos for their study. They used metrics such as views, comments, replies, likes, dislikes, and shares and compared them for conspiracy and informational videos. They found that there was no statistically significant difference between the two groups of videos. One conclusion was that users respond in similar ways in terms of these metrics to both types of videos. In addition, topic modeling showed that informational videos center around the causes and consequences of the virus while conspiracy videos focus on unfounded theories.
I did not find this paper to provide any surprising results. I think the most significant conclusion was that users responded in similar ways in terms of views, shares, and likes to both types of content. This means that both types of content spread in similar ways. The researchers found that Zika virus video comments were all slightly negative on average, which contradicts prior research that found that false news triggers more negative sentiments than true news. However, they did not suggest why this happened. A takeaway from this study is that health organizations looking to spread helpful health information should give careful thought to how to target audiences and engage them. Future research could include exploring other subjects besides the Zika virus. Since some of the findings contradicted prior work, I think it would be interesting to see if this holds true for other topics. Also, the most effective techniques for spreading true news could be studied. Is it effective to debunk conspiracy videos? What is the best way to engage views with true news?
Automated Hate Speech Detection and the Problem of Offensive Language
Hate speech classifier algorithms have been created before with limited accuracy. Hate speech is particularly challenging to classify because it overlaps with offensive language and context must be take into account when analyzing it. This paper looks to create a hate speech detection model that distinguishes between hate speech and offensive language, with the goal of reaching higher accuracy. They used logistic regression on a sample of about 25,000 tweets to train the model.
I thought the paper had significant implications since many parties are interested in flagging hate speech, such as Twitter, Instagram, or countries with hate speech laws. The researched shared a significant amount of challenges related classifying hate speech. I found it interesting that the definition of hate speech is not well defined and that the researchers had to establish a definition for the premise of this paper. In addition, they found and mentioned several tweets that were misclassified by human coders. I seems difficult to train an algorithm to classify something that is not well defined and is often done with error by even humans. Future work could involve similar research taking into account human biases and using more data. I think that a more established definition of hate speech is necessary before these algorithms can be more accurate.
