With Google TextAds, I've noticed on my blog that the text ads take a
little while to get updated to correspond to the new content, which
seems to imply that they take note of the new blog content and decide
which text ads to serve up. The question is, if this is true, how can
it done quickly and efficiently for new pages and updated pages.
The best answer is a Bayesian classifier, which is trained to recognise
that pages containing a certain pattern of word occurrences fits into a
particular classification. If you then associate text ads with a
classification, e.g., "restaurants", and the Bayesian classifier
recognises that a web page fits into the "restaurant" classification
because of a pattern of word occurrences, then you serve up text ads for
that classification to that page.
Bayesian classifers need training, so for each classification, you give
it a set of pages that you think typically describe pages of that class
- it then automatically creates the pattern descriptors it will use
against other pages.
Hope this helps,
Simon
Sam Lee wrote:
Hi,
I am implementing a Google Adwords-like Text Ad thing.
In Adwords, advertisers enter keywords and phases in
their ads. When visitor visits a webpage with
potential Google text ads, I want to know how they
link the webpage to the actual text ads? Linking those
text ads to the webpage is easy, they can just use the
keywords and phases advertisers enters, and then match
them against their index of webpages. But what about
the other way around, webpage to the text ads? I can
associate the webpage address with the text ads easily
by storing them in database, but it will be a problem
when new webpages join the ad network. You can't run
all the keywords and phases of the ads against the new
webpage everytime a new webpage is added, too much
resource.
How will you do it?
Many thanks.
__________________________________
Yahoo! Music Unlimited
Access over 1 million songs. Try it free.
http://music.yahoo.com/unlimited/
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
--
Dr. Simon Courtenage
Software Systems Engineering Research Group
Dept. of Software Engineering, Cavendish School of Computer Science
University of Westminster, London, UK
Email: [EMAIL PROTECTED] Web: http://users.cscs.wmin.ac.uk/~courtes |
http://www.sse.wmin.ac.uk
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
