The Potsdam branch of the project is in charge of
applying computational linguistics methods to the
analysis of individual texts, and thereby contributing
to the precision and coverage of the overall analysis
of the discourse (the network of contributions on a
particular topic). One central goal is to classify
expressions of subjectivity, so that the sentiment
encoded in a text - and possibly in the relation to
other texts that are being referred to - can be
quantified. The other, more exploratory, goal is to
devise methods for determining the "quality" of a
social media contribution: Does it make a point, does
it carry the discussion forward, or is it (for
example) just a quick statement of discontent? - For
both issues, the idea is to support the human analyst
in order to make her or his work on exploring a large
corpus of texts more effective.
