Real-Time Classification of Twitter Trends

The community of users participating in social media tends to share about common interests at the same time, giving rise to what are known as social trends. A social trend reflects the voice of a large number of users which, for some reason, becomes popular in a specific moment. Through social trends, users therefore suggest that some occurrence of wide interest is taking place and subsequently triggering the trend. In this work, we explore the types of triggers that spark trends on the microblogging site Twitter, and introduce a typology that includes the following four types: news, ongoing events, memes, and commemoratives. While previous research has analyzed the characteristics of trending topics in a long term, we look instead at the earliest tweets that produce the trend, with the aim of categorizing trends early on and providing a filtered subset of trends to end users. We propose, analyze and experiment with a set of straightforward languageindependent features that rely on the social spread of the trends to discriminate among those types of trending topics. Our method provides an efficient way to immediately and accurately categorize trending topics without need of external data, enabling news organizations to track and discover breaking news in real-time, or quickly identify viral memes that might enrich marketing decisions, among others. The analysis of social features as observed in social trends also reveals social patterns associated with each type of trend, such as tweets related to ongoing events being shorter as many of the tweets were likely sent from mobile devices, or memes having more retweets originating from fewer users than for other kinds of