Suchergebnisse

A good deal of Twitter research focuses on event-detection using algorithms that rely on keywords and tweet density. We present an alternative analysis of tweets, filtering by hashtags related to the 2012 Superbowl and validated against the 2013 baseball World Series. We analyze low-volume, topically similar tweets which reference specific plays (sub-contexts) within the game at the time they occur. These communications are not explicitly linked; they pivot on keywords and do not correlate with spikes in tweets-per-minute. Such phenomena are not readily identified by current event-detection algorithms, which rely on volume to drive the analytic engine. We propose to demonstrate the effectiveness of empirically and theoretically informed approaches and use qualitative analysis and theory to inform the design of future event-detection algorithms. Specifically, we propose theories of Information Grounds and "third places" to explain sub-contexts that emerge. Conceptualizing sub-contexts as a socio-technical place advances the framing of Twitter event-detection from principally computational to deeply contextual.