The current use cases of TopicExplorer comprises media analyses of blogs, linguistic discourse analysis, analysis of translations of literature, discourse reconstruction from the historic Halle journals during the period of Enlightenment as well as teaching activities in seminars.
Japanese Studies, Media Analysis of Blogs
Mit dem TopicExplorer wurde der zeitliche Diskursverlauf zur radioaktiven Verseuchung von Rindfleisch 2011 anhand von Blog-Texten untersucht. Die automatisch erstellten Topics konnten in vielen Fällen als inhaltlich sinnvoll identifiziert und zu Diskurssträngen gebündelt werden. Diese wurden mit den Ergebnissen einer japanischen Studie einiger hundert einschlägiger Zeitungsartikel zum selben Thema verglichen.
Christian Oberländer: Digitale Japanologie und neue Medien in der Risikogesellschaft: Der Skandal um radioaktiv verseuchtes Rindfleisch als Fallbeispiel, to appear.
Japanese Studies, Linguistic discourse analysis
TopicExplorer is employed to study how the Japanese term individual responsibility (jiko sekinin, 自己責任) is used discourses that appear in internet blog posts. The linguistic analysis of the large corpus of blog posts is expected reveal several new usage scenarios that have not been described before in the related literature. The linguistic analysis is part of a PhD project supervised by Christian Oberländer, Japanese Studies, Martin-Luther-University Halle-Wittenberg.
Japanese Studies, Analysis of translations of literature
Topic models are used to support the analysis of different Japanese
translations of the same work by the German author Thomas Mann. The work
currently analyzed is the novel Tonio Kröger as well as several
translations. Using TopicExplorer it is possible to detect literary
leitmotifs, i. e. frequently recurring metaphorical patterns of speech. Given the fact that those leitmotifs establish the semantical framework of the narrative, the interactive comparison of their translations is expected to reveal hidden properties of the translation process. This analysis is part of a PhD
project supervised by Christian Oberländer, Japanese Studies,
History, Analysis of the Halle journals during the period of Enlightenment (18th century)
The large corpus comprises the bibliographic records of over 30.000 journal articles that were published in Halle / Saale, Germany between 1685 and 1815. TopicExplorer will be used to reconstruct discourses from the titles of the articles and journals. Additionally, the network of authors, editors, printers, publishers and customers, who were involved into the production and dissemination of the articles, will be used to infer meaningful topics. TopicExplorer will help to interactively evaluate the relevance, whether a topic or a combination of several topics constitute a certain discourse. This analysis is part of research activities of the Interdisciplinary Centre for European Enlightenment Studies at Martin-Luther-University Halle-Wittenberg.
TopicExplorer in the class room
TopicExplorer was used in the seminar “Digitales Japan” in the winter terms 2015/16 and 2016/17 that were attended by students from Japanese studies. During the seminar, we discussed non-mathematical basics of topic models and various usage scenarios. With the help of the TopicExplorer, the students where able to conduct several experiments, e.g. evaluating the impact of different numbers of topics, including different part-of-speech word classes into the topic model inference, and checking the effects of the heuristics to prune the vocabulary. The students got first-hand experiences in working with the results of topic models. Several of them will continue to work with TopicExplorer for term papers.