Les,
> -----Original Message-----
> From: Leslie Carr [mailto:[log in to unmask]]
> Sent: 13 March 2008 19:02
> To: John Smith
> Cc: [log in to unmask]
> Subject: Re: Challenging assertions
>
>
> On 13 Mar 2008, at 18:36, John Smith wrote:
>
> > I would be seriously impressed by any algorithm that could tell
> me
> > the core subjects of 'The Wasteland' by T S Eliot and then
> locate
> > other works on the same themes.
> Go on then John, show me what Dewey can describe about the works
> of T
> S Eliot and I'll stand you a beer next month at OR08! (April is
> the
> cruelest month)
We don't use Dewey but if I go to LC classmark PE 4178 in my library I will find nothing but poems by T S Eliot and at PE 4178.W2-W25 nothing but the Wasteland or books about the Wasteland. I won't find books about waste or land or wilderness.
> > Could your algorithms even tell me it was a poem?
> Ahh. What is poetry? My algorithms would have a good chance of
> categorising The Wasteland as poetry because it was (a) published
> in a
> literary magazine and (b) written by a man whose speciality was
> poetry. I think Stevan made a similar point about identifying the
> subject area of an article from an understanding of the journal it
> appeared in. Guilt by association.
Then you are using context and not the content of the document, Stevan's claim concerned analysis of the content of the document. Further how do you know that the journal was a literary magazine or that Eliot wrote poetry? Because a human being asserted it and it was noted somewhere you could find it. No sophisticated computation just a lookup of a human applied subject tag.
Sorry, no cigar (or beer) :-) .
Regards,
John.
|