A Survey On Data Interlinking Methods by Stephan Wölger, Katharina Siorpaes, Tobias Bürger, Elena Simperl, Stefan Thaler, and, Christian Hofer.
From the introduction:
In 2007 the Linking Open Data (LOD) community project started an initiative which aims at increased use of Semantic Web applications. Such applications on the one hand provide new means to enrich a user’s web experience but on the other hand also require certain standards to be adhered to. Two important requirements when it comes to Semantic Web applications are the availability of RDF datasets on the web and having typed links between these datasets in order to be able to browse the data and to jump between them in various directions.
While there exist tools that create RDF output automatically from the application level and tools that create RDF from web sites, interlinking the resulting datasets is still a task that can be cumbersome for humans (either because there is a lack of insentives or due the non-availability of user friendly tools) or not doable for machines (due to the manifoldness of domains). Despite the fact that there are more and more interlinking tools available, those either can be applied only for certain domains of the real world (e.g. publications) or they can be used just for interlinking a specific type of data (e.g. multimedia data).
Another interesting survey article from the Semantic Technology Institute (STI) Innsbruck, University of Innsbruck.
I like the phrase “…manifoldness of domains.” RDF output is useful information about data. The problem I foresee is that the semantics it represents are local, hence the “manifoldness of domains.” Not always, there are some domains that are so close as to not be distinguishable, one from the other, and linking RDF will work quite well.
One imagines that RDF based interlinking OfficeDepot, Staples and OfficeMax should not be difficult. Tiresome, not terribly interesting, but not difficult. And that could prove to be useful for personal and corporate buyers seeking price breaks or competitors trying to decide on loss leaders. Not a lot of reasoning to be done except by the buyers and sellers.
I am sure there would still be some domain differences between those vendors but having a common mapping from one vendor number to all three vendor numbers could prove to be very useful for customers and distributors alike.
For more complex/abstract domains, where “…manifoldness of domains.” is an issue, you can use topic maps.