20 Years of Statistical Parsing for German: What Have we Learned?

old_uid11919
title20 Years of Statistical Parsing for German: What Have we Learned?
start_date2012/12/07
schedule11h-12h30
onlineno
summaryIn this talk, I will look at the development of statistical approaches to parsing German from the beginnings by Dubey and Keller in 2003. German is interesting for parsing research since it counts as a morphologically rich language. But on the richness scale, it is fairly low, and it is still close to English. German is also an interesting case because there are two large-size treebanks available with two different annotation schemes. This puts us in a unique position to look into the influence of decisions in the annotation schemes, independent of the language. However, as the results so far show, there are more questions than answers. I will discuss results on using different parsers as well as different treebanks.
responsiblesCandito