Bitte haben Sie einen Moment Geduld........
Topic Modeling in the Social Sciences
Dr. Andreas Niekler, Dr. Gregor Wiedemann
Datum: 17.09 - 18.09.2018 ics-Datei
The workshop further extends on aspects of the Big Data Module “Text Mining with R”, especially the application of topic models in social science contexts. Topic models allow for automatic thematic clustering of very large document collections. With the help of complex statistical inference approaches, they observe topics as latent semantic coherences based on shared vocabulary within documents. With such topics, researchers are able to describe contents of a collection or filter a huge collection for specific thematic aspects. The workshop concentrates on applications and best practices for topic modeling in the social science. It covers model selection (choosing best hyper-parameters), quality assurance (validity and reliability) and integration into more complex analysis workflows. Participants take part in theoretical lectures and will be provided with R scripts to compute own models in different exemplary tutorials.
The course is targeted at researchers from social sciences and the humanities who are interested in analyzing large textual data sets.
Participants learn how to compute topic models with R on own data sets and how to choose optimal parameters according to their research requirements. Moreover, they learn how to evaluate the results of a topic model process and utilize them in further steps of an analysis process, e.g. document filtering, visualization or time series analysis.
To take part in the tutorials, participants need to have basic knowledge of R programming. Moreover, we recommend basic knowledge of text mining methods.