Here's a link to our abstract, "A cross-linguistic analysis of garden-pathing in a statistical dependency parser":
BostonChenHale.pdf
And here's the .tex file if anyone's interested:
BostonChenHale.tex
We appreciate all comments and input!
Marisa
Subscribe to:
Post Comments (Atom)
3 comments:
With further thought, I re-titled the abstract (in the current PDF). Which one's better?
Modeling cross-linguistic garden-pathing in a statistical dependency parser
or
A cross-linguistic analysis of garden-pathing in a statistical dependency parser
the title is tricky. on the one hand the word "modeling" helps
organizers vector the review to the right place.
on the other hand, you want to emphasize with the word cross-linguistic
that it goes beyond the usual.
how about "Garden-pathing in a statistical dependency parser: Chinese, German, English" ? CUNY-goers love subtitles.
or "Garden-pathing in three languages as viewed from the perspective of a statistical dependency parser"
The transition is rather abrupt into paragraph #2
how about "The parser itself builds dependency representations (Hays)
using a transition system defined by Nivre '06. Parser actions are guided
by features. The features are trained...
I think you can safely leave out the description of what an arc is.
"features tested" --avoid passive--> "The parser uses two kinds of features, state based and non-state based....
n.b. Hays is spelled without the e
In the first sentence I was wondering what type of features you talk about. In the second sentence "features are that take parser state into account" is too general. You explain this feature business later, but maybe you could start with the discussion of the types of features and then explain how you use them for deriving predictions with the probabilistic parser.
My impression is that you don't have to explain dependency grammar. But for historical reasons it's better-known in Europe than in the US, so I'm not sure.
The second sentence is long, maybe replace by: The parser correctly reflects human preferences for a variety of ambiguities in Chinese, German, and English when guided by features that take into account intermediate parser-states. These inculde Subject-Object (S0), Main Verb-Reduced Relative, Conjunction, and Prepositional Phrase Attachment.
Post a Comment