Some time ago I sent a request for references that describe the way in which parsing accuracy varies with the length or complexity of input sentences. I summarise responses to this quest below, for those who may be interested.
I am grateful to Marta Recasens for sending me details of the paper: McDonald, Ryan and Joakim Nivre. 2007. Characterizing the Errors of Data-Driven Dependency Parsing Model. Proceedings of EMNLP-CoNLL 2007, pp. 122–131. http://acl.ldc.upenn.edu/D/D07/D07-1013.pdf
and to Daniel Zamen for highlighting his PhD thesis "Parsing with a Statistical Dependency Model" (p.73) which also contains information relevant to the request. The thesis is available at http://ufal.mff.cuni.cz/~zeman/publikace/disertace/thesis.pdf
Both of these works involve dependency parsing. It would be interesting to find similar studies investigating the accuracy of alternate parsing models, if anyone is aware of such work.
________________________________________ From: corpora-bounces at uib.no [corpora-bounces at uib.no] on behalf of Evans, Richard J [R.J.Evans at wlv.ac.uk] Sent: 30 April 2012 10:07 To: corpora at uib.no Subject: [Corpora-List] Variation of parsing performance with sentence length
Dear corpora subscribers,
I am currently working on a project involving syntactic simplification of text in Bulgarian, English, and Spanish. With regard to syntactic parsing, is anyone aware of research reporting on the way in which parsing performance varies with the length/complexity of input sentences? Any pointers would be gratefully received. I will post a summary of responses.
Richard-- Scanned by iCritical.
_______________________________________________ UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora Corpora mailing list Corpora at uib.no http://mailman.uib.no/listinfo/corpora-- Scanned by iCritical.