TWEETDICT : Identification of Topically Related Twitter Hashtags
Fabian Dreer, Eduard Saller, Patrick Elsässer, Desislava Zhekova · HilDok – Institutional Repository (Universität Hildesheim) · 2014
This paper presents the TWEETDICT system prototype, which uses co-occurrence and frequency distributions of Twitter hashtags to generate clusters of keywords that could be used for topic summarization/identification. They also contain mentions referring to the same entity, which is a valuable resource for coreference resolution. We provide a web interface to the co-occurrence counts where an interactive search through the dataset collected from Twitter can be started. Additionally, the used data is also made freely available.