PropBank

From Cohen Courses
Revision as of 22:15, 30 November 2010 by PastStudents (talk | contribs)
Jump to navigationJump to search

The PropBank dataset (or, Proposition Bank) [1] is annotated with Semantic Role Label.

The dataset is available via Language Data Consortium (LDC) [2].

Semantic role label of the PropBank is predicate-argument relation. Each sentence in the text of the original Penn Treebank is annotated with a set of semantic roles defined for each verb. For example,

Frameset '''accept.01''' "take willingly"
Arg0: Acceptor
Arg1: Thing accepted
Arg2: Accepted-from
Arg3: Attribute
Ex: [<sub>Arg0</sub> He][<sub>ArgM-MOD</sub> would][<sub>ArgM-NEG</sub> n't]''accept'' [<sub>Arg1</sub> anything of value][<sub>Arg2</sub> from those he was writing about]. (wsj_0186)

Hand-checked syntactic parse trees are available since annotation was performed on the Penn Treebank corpus.