Online media have a huge impact on public opinion, economics and politics. Every day, billions of posts are created and comments are written, covering a broad range of topics. Especially the format of hashtags, as a discrete and condensed version of online content, is a promising entry point for in-depth investigations. In this work we provide a set of methods from static community detection as well as novel approaches for tracing the dynamics of topics in time dependent data. We build temporal and weighted co-occurence networks from hashtags. On static snapshots we infer the community structure using customized methods. We solve the resulting bipartite matching problem between adjacent timesteps, by taking into account higher order memory. This results in a matching that is robust to temporal fluctuations and instabilities of the static community detection. The proposed methodology, tailored to uncover the detailed dynamics of groups of hashtags is adjustable and by that broadly applicable to reveal the temporal behavior of various online topics.