Agirre et al. (2000) propose enriching
WordNet with topic signatures using the Web. A
concept in WordNet can contain several senses.
Nevertheless, for each sense a set of cue-words
(hyponyms, hypernyms, etc.) is used to create a
highly specific query that is submitted to the
search engine. The top 100 documents are
retrieved and keywords are extracted. They
experienced formulating the queries as being the
weakest point of their approach. The quality of the
queries highly affected the quality of the retrieved
documents. This is in contrast to our approach that
is not dependent upon a high quality query but uses
clustering and domain identification by utilising
neighbouring entities to find relevant documents
from a set of diverse documents.