I think it's OK to identify the concepts in Shakespeare and use
them to connect documents together. As such Macbeth would have a
place in the 'infosphere' of ghost stories. I agree however that
extracting computer generated summaries of literature rather misses
the point, just as having someone write an essay summarising a
spreadsheet would be distinctly odd. There are some documents that
could be sensibly summarised, but I think the key role that Gerry
McKiernan is suggesting is NOT editing of documents but linking them
together as conceptual classes.
What these super-agents would be doing is identifying a (near
infinite) number of 'dimensions' which represent concepts in a
document: such as death, taxes and aquatic mammals. Any document's
content can then be described by its 'coordinates' in this infospace.
Searching then becomes a mapping process, where you map the (near)
infinite dimensions down to something managable. This role is a
valid one, and would go a long way towards making the Internet a
valuable resource - more like a library than a comic shop. At the
moment there are lots of documents out there, but linking them
together to provide in depth coverage is extremely difficult.
The point is that we do want to read, analyse and interpret
documents as the author originally wrote them. But we also want our
computers to find information for us, preferably in a way that allows
us to make connections of our own. By allowing people to search for
data by concept, rather than keyword, you are letting them find
related items that show the user new connections in the data that
they might never have thought of. How often have you searched for a
book in a library, only to find a better one on the same shelf? That
is possible because libraries sort books by concept, rather than by
the words in the title.
Nick Dearnaley,
AutoNomy.
autonomy@stjohns.co.uk