Tidying up my Bloglines, came across this story from 2006 in New Scientist:
about a search engine being developed which can distinguish between different instances of the same name, by identifying common keywords to cluster similar results together. In testing, it was between 70 and 95% accurate. Another application of text mining? Will be interesting to see how the Institute of Education project working with the ASSERT project at NaCTeM (National Centre for Text Mining) will get on.