How people talk about health?: Detecting Health Topics from Twitter Streams

The paper proposes an online clustering algorithm for detecting health-related topics. The method extracts from the tweets relevant terms and incrementally groups them by taking into account both term occurrences and tweet age. A detailed experimentation on the tweets posted by users in US shows that the method is capable to group tweets addressing common health issues into the pertinent topic, outperforming traditional topic model approaches, like Doc-p and LDA.