Abstract

Abstract: Microblogging activity as supported by Twitter has rapidly gained a lot of attention within the scientific community. For example, the organizers of scientific conferences started exploiting Twitter for various reasons, e.g., engaging customers via backchannel, or providing awareness support for stakeholders. We assume that there is no equal distribution of Twitter activity over time. Instead we argue that there are particular events or occasions that lead to peaks in the number of tweets. Clear ly distinguishable peaks can be used by conference organizers to promote or announce information. At the Science 2.0 conference 1,879 conference‑related tweets (including retweets) were collected between 14.03.2014 and 14.04.2014. In total 822 tweets ( 68%) came from conference attendees versus 392 unique tweets (32%) from external contributors who were also more likely to retweet (24% vs. 74%). Additionally, we conducted a content analysis of all tweets by using a self‑provided codebook with thre e classes: purpose of tweet, target of web link (if embedded in the tweet), and topical relation to Science 2.0Ž. The purpose of over 80% of the tweets was to share conference content or resources. Pictures and the conference website were the most ofte n tweeted link targets (65%). The top four content categories occurred in 11% to 15% of tweets and were scientific working methods,Ž web topics,Ž projects & research programs,Ž and open science & open dataŽ reflecting what the audience was most inte rested in. These results help to understand Twitter behavior regarding time and content. This study provides a threefold additional value: 1) conference organizers know when to announce important conference‑related information to the audience via Twitter , 2) the first two classes of the validated codebook are transferable to studies in a similar vein and can be easily reused from the community, and 3) supports recording of user feedback to conference topics and highlights.