From: Florent J. <flo...@un...> - 2007-04-16 09:40:10
|
Hi, Leo Sauermann wrote: > Hi Florent, Aperture, > > I added the patch to Aperture-addons. > [snip] Thanks for the pointers for the indexing configuration, I'll look at it. >> For the LuceneSail for querying, I couldn't find the place where the >> "matched" property to be queried using lucene are configured. Could >> you explain me where to change it ? > I dont understand the question, what do you mean by matched property? > I was speaking of luceneSail from Aduna Christiaan Fluit mentioned. I don't know if it works exactly the same way (and I can't use the one from Gnowsis since the API changed...) regards, Florent > best > Leo > >> >> Regards, >> Florent >> >> Leo Sauermann wrote: >>> Hi Florent, Aperture >>> >>> It was Florent Jochaud who said at the right time 04.04.2007 13:07 >>> the following words: >>>> Hi, >>>> >>>> >>>> >>>> I had a first look at your code and the produced index. >>>> >>>> Seems to work quite well (as far as I know Lucene, and at least, >>>> the index >>>> seems to be well populated...) >>>> >>>> I'm not sure how to improve it, I see several way: >>>> >>>> - Should all field be indexed ? or only some specifics >>>> ones ? I >>>> think fulltext, name, keywords, and date should be indexed (not >>>> even sure >>>> for the date) >>>> >>> for now, all fields should be dumped into lucene. we could make some >>> config file, disabling properties. this should be done in rdf, for >>> coolness >>> >>>> - Literals should be indexed without their types definition >>>> >>>> >>> try this: >>> if (node instanceof Literal) >>> { >>> string s = ((Literal)node).getValue(); >>> >>> .... use s... >>> } >>> >>>> - Using URI as field's name make quite longs name, can it >>>> be a >>>> problem for Lucene ? >>>> >>> The fields contain funny character that lucene seems to hate >>> (especially #) which cannot be escaped out of queries so easily. but >>> besides that, I see no problem. >>>> - If date are indexed, seems that it should be first >>>> converted (I >>>> didn't look already the date format in aperture, I'm not quite sure >>>> if it is >>>> necessary) >>>> >>>> >>> for a quick start, try RDFTool.toDate (or however this methid is >>> called, RDFUtil? RDFTool? its a utility class in rdf2go) >>> >>> >>>> - I think other specific lucene's improvement are specific >>>> for each >>>> applications, I'm not sure they are needed in this demo Handler... >>>> >>>> Did I miss something ? >>>> >>>> >>> nope, you got it right! >>> >>> a small search interface with a textfield and a button would rock... >>> >>> best >>> Leo >>>> >>>> >>>> Regards, >>>> >>>> Florent >>>> >>>> >>>> >>>> >>>> >>>> Von: Leo Sauermann [mailto:leo...@df...] Gesendet: >>>> Freitag, 30. März 2007 12:56 >>>> An: Florent Jochaud; Aperture Developers >>>> Betreff: Re: AW: [Aperture-devel] Lucene indexing >>>> >>>> >>>> >>>> Chris Fluit and we are tiddling out the details for a combination >>>> of Lucene >>>> and RDF indexing in one Lucenesail, we have much code and >>>> experience laying >>>> around, and its a nifty problem, so I would not want you to get >>>> dragged into >>>> it too much. >>>> >>>> What we would really need is a "beautifully simple" Lucene >>>> ApertureCrawlerHandler, that is < 500 lines and shows how blissfull >>>> Aperture >>>> is for lucene. >>>> >>>> I started a first version, it compiles, it does soemthing, >>>> though I didn't check what... >>>> http://aperture.cvs.sourceforge.net/aperture/aperture-addons/src/java/org/se >>>> >>>> manticdesktop/aperture/lucenehandler/LuceneHandler.java?view=markup >>>> >>>> >>>> Florent: could you pick up this ball? >>>> >>>> best >>>> Leo >>>> >>>> >>>> It was Florent Jochaud who said at the right time 30.03.2007 10:30 the >>>> following words: >>>> Hi, >>>> >>>> >>>> >>>> I'm not sure to be able to do something in this time, but I will >>>> try :-) I >>>> still need to go in details in lucene usage and how to set up the >>>> index. >>>> >>>> Could you explain me more in details what you mean by simple store ? >>>> Especially where in Aperture you think this storage should be set ? >>>> I will >>>> have a look, but I still don't know aperture in details. >>>> >>>> >>>> >>>> Florent >>>> >>>> >>>> >>>> >>>> >>>> Von: Leo Sauermann [mailto:leo...@df...] Gesendet: >>>> Dienstag, 27. März 2007 17:40 >>>> An: Florent Jochaud >>>> Betreff: Re: [Aperture-devel] Lucene indexing >>>> >>>> >>>> >>>> If you do a simple lucene store within the next week, its in the next >>>> official release of aperture, which will happen sooner or later :-) >>>> >>>> this would fit perfect. >>>> >>>> >>>> best >>>> Leo >>>> >>>> >>>> >>>> It was Florent Jochaud who said at the right time 27.03.2007 08:54 the >>>> following words: >>>> Hi, >>>> >>>> Late and short answer : thanks for all theses information, this >>>> sail seems >>>> really interesting. I'll have a look at the way to use it for my >>>> application >>>> soon. >>>> >>>> Regards, >>>> >>>> Florent >>>> >>>> >>>> -----Ursprüngliche Nachricht----- >>>> Von: ape...@li... [mailto:aperture- >>>> <mailto:dev...@li...> >>>> dev...@li...] Im Auftrag von Christiaan Fluit >>>> Gesendet: Freitag, 16. März 2007 11:50 >>>> An: <mailto:ape...@li...> >>>> ape...@li... >>>> Betreff: Re: [Aperture-devel] Lucene indexing >>>> >>>> Leo Sauermann wrote: >>>> >>>> I saw there is plan for a Lucene Sail architecture for storing the >>>> >>>> full text >>>> >>>> extract of documents. >>>> >>>> >>>> there are two different implementations, one by us for gnowsis, one >>>> >>>> by >>>> >>>> chris fluit >>>> >>>> the one by chris, he knows where it is. >>>> >>>> The most uptodate code is located here: >>>> >>>> https://src.aduna-software.org/svn/info.aduna/projects/aduna- >>>> infosource/src/main/java/info/aduna/infosource/local/ >>>> >>>> This works with Sesame 2 beta 2 and Lucene 2.0. >>>> >>>> Note that this demonstrates a very particular way of combining Lucene >>>> with Sesame. It depends on a virtual "matches" RDF property that will >>>> be >>>> evaluated on a Lucene index by the LuceneSail. This yields document >>>> URIs >>>> and relevancies that are substituted into the query before being >>>> passed >>>> to a wrapped Sail that evaluates it on a "real" RDF store. It does >>>> *not* >>>> perform a full-text search on all RDF literals. Also, the LuceneSail >>>> only handles querying, it does not handle creation and updating of the >>>> Lucene index. >>>> >>>> I plan to do an upgrade to Lucene 2.1 soon, which could make our >>>> LuceneIndex class simpler. >>>> >>>> >>>> We plan to make a better version integrating Lucene and a SAIL. >>>> probably some people from nepomuk.semanticdesktop.org will work >>>> >>>> together >>>> >>>> with chris, the project hasn't started yet, but everyone needs it. >>>> >>>> I think I have gathered enough knowledge of both Lucene and Sesame to >>>> realize the generic full-text indexing and querying Sail that Leo >>>> mentions, I only need to find some time... >>>> >>>> >>>> Also, is this Sail implementation supposed to be integrated directly >>>> >>>> in >>>> >>>> aperture one day ? Or the use of RDF2go means that store-specific >>>> >>>> components >>>> >>>> are "forbidden" ? >>>> >>>> >>>> We could do demos or examples showing how to do it, >>>> but aperture itself should stay clean of these "forbidden planets". >>>> >>>> I think it would be great to have this full-text indexing and >>>> searching >>>> Sail as a contribution in Sesame's SVN (projects/sesame-contrib). It >>>> seems that this is of interest to the Sesame community at large, not >>>> only for Aperture users. >>>> >>>> >>>> Regards, >>>> >>>> Chris >>>> -- >>>> >>>> ----------------------------------------------------------------------- >>>> >>>> -- >>>> Take Surveys. Earn Cash. Influence the Future of IT >>>> Join SourceForge.net's Techsay panel and you'll get the chance to >>>> share >>>> your >>>> opinions on IT & business topics through brief surveys-and earn cash >>>> http://www.techsay.com/default.php?page=join.php >>>> <http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVD> >>>> >>>> &p=sourceforge&CID=DEVD >>>> EV >>>> _______________________________________________ >>>> Aperture-devel mailing list >>>> Ape...@li... >>>> https://lists.sourceforge.net/lists/listinfo/aperture-devel >>>> >>>> >>>> >>>> ------------------------------------------------------------------------- >>>> >>>> Take Surveys. Earn Cash. Influence the Future of IT >>>> Join SourceForge.net's Techsay panel and you'll get the chance to >>>> share your >>>> opinions on IT & business topics through brief surveys-and earn cash >>>> http://www.techsay.com/default.php?page=join.php >>>> <http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV> >>>> >>>> &p=sourceforge&CID=DEVDEV >>>> _______________________________________________ >>>> Aperture-devel mailing list >>>> Ape...@li... >>>> https://lists.sourceforge.net/lists/listinfo/aperture-devel >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> ------------------------------------------------------------------------ >>>> >>>> >>>> ------------------------------------------------------------------------- >>>> >>>> Take Surveys. Earn Cash. Influence the Future of IT >>>> Join SourceForge.net's Techsay panel and you'll get the chance to >>>> share your >>>> opinions on IT & business topics through brief surveys-and earn cash >>>> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV >>>> >>>> ------------------------------------------------------------------------ >>>> >>>> >>>> _______________________________________________ >>>> Aperture-devel mailing list >>>> Ape...@li... >>>> https://lists.sourceforge.net/lists/listinfo/aperture-devel >>>> >> > > > -- > ____________________________________________________ > DI Leo Sauermann http://www.dfki.de/~sauermann > > Deutsches Forschungszentrum fuer > Kuenstliche Intelligenz DFKI GmbH > Trippstadter Strasse 122 > P.O. Box 2080 Fon: +49 631 20575-116 > D-67663 Kaiserslautern Fax: +49 631 20575-102 > Germany Mail: leo...@df... > > Geschaeftsfuehrung: > Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender) > Dr. Walter Olthoff > Vorsitzender des Aufsichtsrats: > Prof. Dr. h.c. Hans A. Aukes > Amtsgericht Kaiserslautern, HRB 2313 > ____________________________________________________ > |