Hi all, The key to Linked Data is obviously the links, but I dont think any of us quite know how the links actually work. In theory, I imagine having a record (ie a set of triples), in a local triplestore, which links to records available on the Internet. Periodically, those records are downloaded, and stored in a local triplestore as a cache. Then we somehow follow the links as they are stored in our local triplestore and have a set of string literals which we can show to users on a web page. In practice, I dont know how that would work. I can see us getting a list of record subjects in our local triplestore, and then iterating through each one to download the links in its objects. But I dont know where wed put that content. I also dont know if wed recurse into that content to follow more links or not. Lets say that we store that content in named graph called cache. When it comes time to re-trace the links for our list of record subjects, how do we know what local content to delete? In theory, we could use the URI from the object, but many records contain many subjects, so even if we did delete that subject from cache, we havent deleted the other subjects that it links to internally in the triplestore. Im going to ask around to see if anyone can enlighten me. Im hoping Petter might be able to shed light on how Oslo Public Library makes it work, especially in the context of ls.ext interfacing with Koha! David Cook Systems Librarian Prosentient Systems 72/330 Wattle St Ultimo, NSW 2007 Australia Office: 02 9212 0899 Direct: 02 8005 0595