This paper discusses the problem of annotation of coreference relations with generic expressions in a large scale corpus. We present and analyze some existing theories of genericity, compare them to the approaches to generics that are used in the state-of-the-art coreference annotation guidelines and discuss how coreference of generic expressions is processed in the manual annotation of the Prague Dependency Treebank.
After analyzing some typical problematic issues we propose some partial solutions that can be used to enhance the quality and consistency of the annotation