Discourse Coherence Theories and Modeling (Seminar)

Teachers: Annemarie Friedrich, Alexis Palmer
Location: Building. C7.2, Konferenzraum 2.11
Time: Mon 10-12
First session: 22.04.2013
Appropriate for: M.Sc. LST/LCT
Credit points: 4 CP (only presentation) / 7 CP (presentation & paper)

Coherence: A coherent text is designed around a common topic. In the reading process, the individual units of information enter meaningful relationships, which on the one hand result from coreference between referring expressions, and on the other hand from semantic and pragmatic relations (e.g. Cause) between adjacent units. The text coheres and is not just a sequence of sentences to be interrupted in isolation. [Manfred Stede: Discourse Processing. Synthesis Lectures on Human Language Technologies, Morgan & Claypool. 2012. (page 1)]

Aim of this seminar: We will discuss approaches to modeling discourse coherence, including linguistically motivated theories and recent computational models. The computational modeling of discourse coherence is important for a variety of NLP tasks ranging from text understanding to text generation. When interpreting a text (for purposes such as Question Answering or Information Extraction), it is important to know how the different units of the text relate. When generating text (e.g., summaries), coherence is essential in order to actually convey the information to the reader.

Requirements: see Requirements page