Understanding and clustering hashtags according to their word distributions

The purpose of our study is to understanding hashtags used in Twitter and clustering hashtags according to their word distribution, so that we can discover the public trend of user's topic in real time and bring benefit to marketing management. We extract the most frequent hashtags from sample tweets and then collect sample tweets related to these hashtags. We create a list of the word with frequent and probability and sort it by most frequent word. When looking through this word list, we can realize what is the meaning represented by each hashtag. Furthermore we cluster hashtags to several groups according to Jensen-Shannon divergence between any two hashtags and represent the results with dendrogram, which provides the hierarchical structure of all hashtags. As a result, it can be expected to deliver advertisements to Twitter users more precisely according to their interests.