A Textual History of Mozilla: Topic Data and Graph Gallery

All images were generated by Michael L. Black using data mined from 60 versions of Netscape Navigator, Mozilla Suite, and Mozilla Firefox. Data table with version labels is available following paragraph 14 in the write-up of this study, published by Digital Humanities Quarterly.

All source code was pre-processed in Python according to the recommendations in Kuhn, et al (2007). The topic models were produced using a customized workflow incorporating the LDA from the MAchine Learning for LanguagE Toolkit (MALLET) Java library and post-processed using a custom Python implementation of the DiffLDA model described in Thomas, et al (2011). All galleries were generated using a combination of Python and R.

Gallery of individual topic graphs:
60 topics, represented as membership (% of all tokens assigned to topic over time)