Corpora
Reciprocal instances dataset:
This is a collection of 10,882 verb pairs that have a reciprocal relationship. We first semi-automatically discovered
patterns encoding reciprocity using a set of simple but effective pronoun templates. Then using a set of the most frequently
occuring patterns, we extracted these pairs of reciprocal pattern instances by searching the web.
The set can be downloaded here.
Semantic relations corpora:
This is a text collection of literary works (cluvi) and
European parliament sessions (europarl) which was annotated
with semantic relations. Here, a semantic relation captures the meaning between two nouns in noun phrases of the type "N N" and "N prep. N". For example, faces of children encodes a part-whole relation P-W(faces, children).
The annotated corpora can be downloaded
here!
(New: The corpora are also annotated in the SemEval 2007 - Task 4 annotation style!)