Building annotated resources for automatic text summarisation
Constantin Orasan (Computational Linguistic Group, School of Humanities, Languages and Social Sciences, University of Wolverhampton)
WP5: Components & Systems
Annotated corpora are necessary for automatic summarisation, but given how difficult is to produce them there are only few available. This paper presents an annotation tool which helps the human annotator to select the important units from a text. In addition to the tool, a new annotation scheme is proposed so that phenomena which such as presence of anaphoric expressions and redundancy can be marked. We argue that by annotating these phenomena the results of evaluation can be made more reliable.
Text summarization, Computer-Aided summarisation, Corpus annotation, User interface, Human summarisation