- From: Kevin Smathers <kevin.smathers@hp.com>
- Date: Wed, 10 Sep 2003 09:34:48 -0700
- To: "Seaborne, Andy" <Andy_Seaborne@hplb.hpl.hp.com>
- Cc: "'karger@theory.lcs.mit.edu'" <karger@theory.lcs.mit.edu>, "'www-rdf-dspace@w3.org'" <www-rdf-dspace@w3.org>
Hi David, I've just gotten a new box that is capable of running Haystack on my desk, and made the mistake of telling it where my IMAP server is located; I have little over 3000 messages in my Inbox ;). Oops, looks like the system died. There are a couple of very high level issues that I suspect need to be resolved for Haystack/Jena integration. Based on the requirements research I did earlier this year, if Jena is to replace Cholesterol/Adenine then Jena will need to support event triggers and probably incorporate the S, P, O and O, P, S indices used by Haystack, and Haystack would have to translate the Adenine inferences into one of the inferencing languages that Jena supports. However, as I understand it, Haystack is normally configured to query its models in memory with currently unused graphs swapped out to disk. If integrated at the pager level then Haystack could continue using Cholesterol and Adenine, and use e.g Josecki simply as a file transfer utility for paged subgraphs. Although the second approach has the effect of sharing information between Haystack and Jena, it is really more like import/export than actual integration. The huge number of triples created by Adenine will probably overwhelm Jena's storage capabilities, and doesn't make good use of Jena's built in inferencing. Arguably this would mean that Jena should only be used to export DSpace data to Haystack through this interface, and not for storage of Haystack triples as a whole. I can copy out the relevant parts of our requirements doc if you are interested. Cheers, -kls Seaborne, Andy wrote: >David, > >In the first instance, that would be me. > >There are two possible scenarios: using a Joseki server and remotely >connecting Haystack to that Joseki server (for example, the new history >store) or connecting Haystack to a external but local Jena model (for >example, one held in a database). > >Both these are interesting; both can be done with the same query approach >although the tighter local binding may also be capable of finer grain >access. I can help with either. > >There is also the open mailing list mailto:jena-dev@goups.yahoo.com for >developers using Jena which everyone should feel free to use. This is the >place for Jena and Joseki questions. > > > >>I think this should be relatively straightforward once the >>information starts to flow. >> >> > >I hope so as well. > > Andy > >-----Original Message----- >From: David R. Karger [mailto:karger@theory.lcs.mit.edu] >Sent: 10 September 2003 15:46 >To: www-rdf-dspace@w3.org >Subject: haystack + jena > > > > >Something we've been discussing for a while (and an important step for >simile) is to get haystack working as front end against a jena back >end. I've identified a haystacker to work on this from our end. Is >there an HPer who can similarly accept responsiblity for the jena side >of things? I think this should be relatively straightforward once the >information starts to flow. > >thanks >David > > -- ======================================================== Kevin Smathers kevin.smathers@hp.com Hewlett-Packard kevin@ank.com Palo Alto Research Lab 1501 Page Mill Rd. 650-857-4477 work M/S 1135 650-852-8186 fax Palo Alto, CA 94304 510-247-1031 home ======================================================== use "Standard::Disclaimer"; carp("This message was printed on 100% recycled bits.");
Received on Wednesday, 10 September 2003 12:35:44 UTC