PULS
Foto: Matthias Friel
"Discourse Parsing" is the task of ascribing structure to a text document. We look at a number of different perspectives from which this structure can be determined, and then focus on the automatic identification of coherence relations and an associated tree structure for the text. The underlying approaches are the "Penn Discourse Treebank" annotation scheme (Webber et al.) and "Rhetorical Structure Theory" (Mann/Thompson). Practical applications, or "downstream tasks", of determining these structures include text summarization, sentiment and argument analysis; the last part of the course will cover them.
To obtain credit, participants have the choice between (i) giving an oral presentation and writing a term paper on a theoretical or empirical investigation, and (ii) doing a software project involving a subtask of discourse parsing. This term, one option in this regard will be participation in a hackathon in early June (organized by UP Digital Humanities network).
M. Stede: Discourse Processing. Morgan & Claypool, 2012.
© Copyright HISHochschul-Informations-System eG