Automating Search Interfaces"

Don Park donpark at quake.net
Fri Feb 20 09:48:32 GMT 1998


>I agree that this disheartening scenario is quite possible. But what a
>shame! It seems that one of XML's major strength's is its ability to
>search heterogeneous databases. Independent sellers large and small
>would benefit from heterogeneous searches for it would allow super
>accurate marketing. Mom and pop producers should be able to sell their
>boutique goods to the special set of consumers that would be interested.
>A real estate agent in Backwater USA with a unique property should be
>able to sell that product in an industry standard search engine.
>Without accurate, industry specific search interfaces, consumers will
>not easily find these sites. Otherwise we are no better off search wise
>than we are today ó wallowing in inaccurate searches. It would be a real
>shame if the ultimate promises of XML were hindered by lack of
>planning.  Laissez-faire is not always the best way.

As far as I am concerned, the scenario is not only possible, it is
absolutely the only way the history will unfold because content developers
will find it hard to convert non-XML contents into XML using standard DTDs.
Commercial contents are typically composite data which can not be easily
described with a set of standard schemas.  Search services makes it even
worse because their schema requirement will be far less than that of content
providers.

Search across heterogeneous databases can still be achieved without asking
everyone to put on a straightjacket and wiggleahead at manageable speed for
the benefit of mankind.  The key lies in dynamic schema conflict resolution
technologies.  If search service wants the price in pesos and the database
stores  prices in US dollars, price can be converted by an adapter at the
time of demand using currency market datafeed.  Currency conversion can not
be done beforehand nor cached because its shelf-life is basically counted in
minutes.  It is also quite unfriendly to return search results with prices
in ten different currencies.

Also standard DTDs can not adapt to change.  What do you do when the
standard DTD for electronic devices must be changed to include performance
data (i.e. WinMark for Intel machines)?  The problems are simply
mindboggling (well, my mind is easy to boggle).


> It is true that hand crafted search interfaces would be more polished,
> but who should be responsible for their creation. Is there some
> designated Java developer in the hotel industry that will make a search
> engine selflessly for the entire industry. No. If such work is relegated
> to the private companies then such search engines will not represent the
> entire industry in a unbiased way. This leaves nice, but proprietary
> search engines, and we are right back to where we started from; searches
> of privately selected database rather than searches of heterogeneous,
> industry representative databases.

Search companies will attack one industry at a time with the search company
providing the custom user interface and dictating what the DTD should be.
Each attack will be turned into a press event with announcements of support
from major players in that particular industry.  These companies will
announce that they will provide data using the search company's
industry-specific search DTD.  Small companies with less resources will
provide using push model since they do not have the resources for taking
part in distributed search network.  Large companies will place more value
in their data and will provide information on demand, thus taking part in
the search network.

>I do not feel that consumers will not care about search accuracy. When a
>customer is looking for variations of Ginkgo Biloba (an over-the-counter
>drug) they want to see all the sites that sell it and for what price.
>The same is true for travelers looking for room availability at their
>travel destinations. No one wants to wade though a hundred tangentially
>related sites. Without accurate search interfaces, consumers will not
>get this sort of accurate response. The RDF is an important part of
>describing the web, but I have not seen how it would right way to
>address automating search interfaces.

This was an exaggeration on my part.  I appologize.

Have said all this.  I still feel that efforts to standardize DTDs must be
made and must be maintained for the sake of balance and stability.  There
wouldn't be much of a market if everyone used their own currency as their
picture ID.

Prophet for Profit,

Don Park
http://www.quake.net/~donpark/index.html



xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)




More information about the Xml-dev mailing list