An online demo of our knowledge poor anaphora resolution system.

A multipurpose annotation tool which can be used for a variety of annotation tasks.

For an input noun, the program retrieves the 30 most similar ones based on distributional data extracted from BNC (predicate-object co-occurrence pairs). At present 10 different distributional similarity measures are implemented.

These demos allow users to run a term-based summariser in different settings, compare the output of term-based summarisers or run the term-based summariser on an annotated corpus.

A corpus of newswire texts extracted from Reuters corpus annotated with NP coreference and event coreference.

Corpus annotated with information about the important sentences.

Two sets of annotation guidelines developed in our British Academy funded project on coreference annotation.

A set of annotation guidelines developed in the CAST project in order to annotate the CAST corpus.