Who Makes Trends? Understanding Demographic Biases In
Who Makes Trends? Understanding Demographic Biases in Crowdsourced Recommendations Abhijnan Chakraborty∗#, Johnnatan Messiaso#, Fabricio Benevenutoo, Saptarshi Ghosh∗, Niloy Ganguly∗, Krishna P. Gummadi# #Max Planck Institute for Software Systems, Germany ∗Indian Institute of Technology Kharagpur, India oUniversidade Federal de Minas Gerais, Brazil Abstract A large number of prior works on trending topics have fo- what Users of social media sites like Facebook and Twitter rely on cused on the trends are (e.g., classifying the trends into crowdsourced content recommendation systems (e.g., Trend- different categories (Naaman, Becker, and Gravano 2011)), ing Topics) to retrieve important and useful information. Con- or how the trends are selected (e.g., proposing new algo- tents selected for recommendation indirectly give the ini- rithms to identify trends from the content stream (Benhardus tial users who promoted (by liking or posting) the con- and Kalita 2013)). Complementary to the earlier works, our tent an opportunity to propagate their messages to a wider focus in this paper is on the users who make different topics audience. Hence, it is important to understand the demo- worthy of being recommended as trending. Specifically, we graphics of people who make a content worthy of recom- attempt to analyze the demographics of crowds promoting mendation, and explore whether they are representative of different topics on the social media sites. By promoters of the media site’s overall population. In this work, using ex- a topic, we refer to the users who posted on the topic be- tensive data collected from Twitter, we make the first at- fore tempt to quantify and explore the demographic biases in it became trending, thereby contributing to the topic’s the crowdsourced recommendations.
[Show full text]