Warning: This tool or project is no longer maintained and kept available only for archival purposes. Since GoodRelations and schema.org have evolved significantly in the past years, the current status available on this page is unlikely to function as expected. We take no responsibility for any damage caused by the use of this outdated work, to the extent legally possible.

Due to a lack of resources, we are unable to provide support for this project outside of consulting projects or sponsored research. Please contact us if you can contribute resources to update and enhance these resources.

GoodRelations - The Web Vocabulary for E-Commerce

This is the archive of the goodrelations dicussion list

GoodRelations is a standardized vocabulary for product, price, and company data that can (1) be embedded into existing static and dynamic Web pages and that (2) can be processed by other computers. This increases the visibility of your products and services in the latest generation of search engines, recommender systems, and other novel applications.

[goodrelations] Product models URIs in goodrelations ontology

Martin Hepp (UniBW) hepp at ebusiness-unibw.org
Tue Oct 6 12:05:10 CEST 2009


-- apologies for resending, the original message was not delivered --

Hi Eugenio:

Since I assume that your questions is of a more general interest, I am
copying the GoodRelations mailing list in my reply:

Eugenio Tacchini wrote:
> I'm reading your papers about it and at the moment it's not very clear 
> to me how exactly product models are identified.
>
> Imagine a scenario in which several e-commerce web sites sell (among 
> others) the same product (e.g. a Macbook air) and a user ask to a 
> goodrelations-aware search enginge which are the 3 cheaper offfers for 
> a Macbook air. Do all the web sites need to refer to the same URI to 
> identify the product "Macbook air" (provided by who?
> A (goodrelations) apple Web site?) or "Macbook air" is just a literal 
> assigned to a property? In the second case how the linkining process 
> is performed? Just by using a simple string comparison?
>
In an ideal world, Apple as the manufacturer would / will publish the
data for the make and model "Macbook Air" in its own domain name space,
including all distinct features etc.
Example

<http://www.apple.com/products/macbook-air#makeAndModel> a
gr:ProductOrServiceModel.

Retailers could then announce offer for macbook-air computers by using
gr:ProductOrServicesSomeInstancesPlaceholder and using Apple's
authoritative URI when linking to the make and model:

foo:myBunchOfMacBooks a gr:ProductOrServicesSomeInstancesPlaceholder;
     gr:hasMakeAndModel
<http://www.apple.com/products/macbook-air#makeAndModel>.

(Side remark: This would allow inferring product feature details from
the model specification, as sketched in

http://www.ebusiness-unibw.org/wiki/GoodRelations_Ruleset_for_Product_Model_Properties).

Now, what will happen (is already happening) in practice is that may
people will define a make and model for the MacBook Air, simply because
there either is no "authoritative" definition (e.g. by Apple) or they
have no easy way of getting the respective URI (it may not be included
in their current product database and they have no resource searching
for those URIs).

So BestBuy and Amazon and other will all define multiple
gr:ProductOrServiceModel instances in their own name space and provide
links only to their local model definition.

The challenge for a GoodRelations commerce dataspace will then be to
consolidate the multiple URIs describing the same make and model -
entity consolidation as known from distributed databases.

Fortunately, already a few simple heuristics will get you far, since you
are operating on structured data, not on plain text.

For example, you could compute string distance metrics for all pairs of
gr:ProductOrServiceModel instances in the space for their rdfs:label
literals.

Even more powerful will be searching for matching gr:hasEAN_UCC-13
properties. If the EAN or UCC13 code of two gr:ProductOrServiceModel
instances is nonempty and matches, you can be almost certain that they
reflect the same model.

Hope that clarifies the issue :-)

By the way, you face the same problem of entity consolidation anywhere
on the Semantic Web - dozens of foaf profiles reflecting the same human etc.

Note that this is a strong argument for NATIVE RDFa / RDF meta-data,
because all RDF derived from microformats or proprietary structures will
lack authoritative URIs.


Best

Martin

-- 
--------------------------------------------------------------
martin hepp
e-business & web science research group
universitaet der bundeswehr muenchen

e-mail:  mhepp at computer.org
phone:   +49-(0)89-6004-4217
fax:     +49-(0)89-6004-4620
www:     http://www.unibw.de/ebusiness/ (group)
          http://www.heppnetz.de/ (personal)
skype:   mfhepp
twitter: mfhepp

Check out GoodRelations for E-Commerce on the Web of Linked Data!
=================================================================

Webcast:
http://www.heppnetz.de/projects/goodrelations/webcast/

Recipe for Yahoo SearchMonkey:
http://tr.im/rAbN

Talk at the Semantic Technology Conference 2009:
"Semantic Web-based E-Commerce: The GoodRelations Ontology"
http://tinyurl.com/semtech-hepp

Talk at

Overview article on Semantic Universe:
http://tinyurl.com/goodrelations-universe

Project page:
http://purl.org/goodrelations/

Resources for developers:
http://www.ebusiness-unibw.org/wiki/GoodRelations

Tutorial materials:
CEC'09 2009 Tutorial: The Web of Data for E-Commerce: A Hands-on 
Introduction to the GoodRelations Ontology, RDFa, and Yahoo! SearchMonkey
http://tr.im/grcec09







More information about the goodrelations mailing list