| Abstract |
We study the information behavior of a particular group of professional social media analysts: online reputation analysts. Reputation analysts determine the reputation of companies based on the reputation polarity of individual media items.
We analyze the annotation process of reputation analysts searching and annotating media expressions to determine the reputation polarity of companies in a large number of economic sectors. Three types of data are analyzed: (i)~questionnaire data, about the process and indicators analysts consider for the annotation; (ii)~log data of the annotation tool; and (iii)~videos of analysts following the thinking-out-loud protocol during the annotation process.
We find that the annotation processes vary greatly over different online media types. Focussing on Twitter, we find that analysts require information about the author, topic, and reach of a tweet. Especially the topical authority of the author, both in online and offline setting, and the number of followers are important factors to determine the influence of a tweet. |