Menu
▾
▴
Re: [Jonto-users] [JeromeDL-devel] R: R: Jeromedl customization: localization of taxonomies, wordnet vocabulary, and indexing properties
|
From: Mateusz K. <mat...@de...> - 2007-12-18 21:17:23
|
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <meta content="text/html;charset=UTF-8" http-equiv="Content-Type"> <title></title> </head> <body bgcolor="#ffffff" text="#000000"> Giuseppe,<br> <br> sorry you have to wait so long.<br> So these are my first thoughts, what can be wrong:<br> - most of all - what did you write in dcont.properties file? You have to put there some information concerning the structure of yours taxonomy RDF file (e.g. its name, the root nodes and the namespace, just like in the already implemented taxonomies)<br> - secondly - RDF file itself. Frankly I have only used N-triples standard as they are most intuitive for me (though pretty large) and don't quite know, how JOnto (or rather its Sesame wrapper) would work with other types of RDFs. <br> - generally speaking - your step-by-step algorithm of building new project is correct. I have just done it quickly myself and everything is OK (I changed ACM module to Dcont).<br> - don't pay to much attention to XxxParser.java files, as these were only created to format text files with taxonomies to correct RDF N-Triples files.<br> - what is exactly wrong? I mean - do you see only the header of Dcont taxonomy but without any contents? Or just standard JOnto taxonomies? Or whole JOnto is broken?<br> <br> To better understand and help you I could make good use of any error message you get in Tomcat console while you are running JOnto. By the way - how do you test it if it works? There is a simple webapplication in JOnto_webapp which allows you to see a test page with all taxonomies, just like in JeromeDL, but without other unnecessary things - only the "accordion" GUI. It is more handy than testing JOnto in JeromeDL.<br> <br> I hope we will together resolve your problem.<br> <br> Regards,<br> Mateusz<br> <blockquote cite="mid:001e01c840a1$e9bb7cf0$bd3276d0$@it" type="cite"> <meta http-equiv="Content-Type" content="text/html; "> <meta name="Generator" content="MS Exchange Server version 6.5.7638.1"> <title>[JeromeDL-devel] R: R: Jeromedl customization: localization of taxonomies, wordnet vocabulary, and indexing properties</title> <!-- Converted from text/plain format --> <p><font size="2">Hi Matheus!<br> as from the below mailing with Sebastian I would customize the jonto taxonomies in Jeromedl by adding a new domain taxonomy. So, after the svn of the whole Jonto project, I built a new Jonto taxonomy project, as from the below Sebastian's suggestions. I named my JOnto taxonomy project "JOnto_dcont". The building process seemed to be successful, but at the launching of Jeromedl as well of the http interface of the underlying sesame platform, the new dcont taxonomy is not loaded. I realized my dcont taxonomy building by a step by step process as follows. Please verify it and let me know about what is wrong or missing.<br> <br> Thanks in advance,<br> Giuseppe<br> <br> Step by step building process:<br> 1) Creation of a new project folder ( "JOnto_dcont") by copying the whole folder of a previous Jonto taxonomy folder (ex. the JOnto_acm);<br> 2) Customization of the new folder by:<br> 2.1 renaming the *Context.java and *Parser.java respectively to DcontContext.java and DcontParser.java, and adaptation of the internal code by string find and replace actions respectively on the pre-existent taxonomy name and the new taxonomy one.<br> 2.2 adaptation of the build.xml and of the .project files by string find and replace actions as above;<br> 2.3 ant generation of the jonto_dcont.jar, and its storage into the JOnto_dcont/lib folder;<br> 3) Updating of the JOnto_base module by:<br> 3.1 including reference of the new Context getting method in JOnto_base/src/context.properties;<br> 3.2 adaptation of the build.xml file, with reference to the new taxonomy in the various sections (i.e: property name, target name, copy todir);<br> 3.3 generation of the updated jonto_base.jar;<br> 4) Updating of the JOnto_webapp module by:<br> 4.1 adaptation of the build.xml file as for Jonto_base;<br> 4.2 generation of the updated jonto_webapp.jar;<br> 5) copy of jonto_user.jar, jonto_base.jar in the Tomcat lib;<br> 6) copy of the JOnto_webapp.jar in the Jeromedl_webapp/WEB-IF/lib.<br> <br> <br> -----Messaggio originale-----<br> Da: Sebastian Kruk [<a moz-do-not-send="true" href="mailto:seb...@de...">mailto:seb...@de...</a>]<br> Inviato: venerdì 14 dicembre 2007 11.01<br> A: Giuseppe Bux<br> Cc: Devel JeromeDL; <a class="moz-txt-link-abbreviated" href="mailto:jon...@li...">jon...@li...</a>; Mateusz Kaczmarek; <a class="moz-txt-link-abbreviated" href="mailto:mar...@li...">mar...@li...</a><br> Oggetto: Re: R: Jeromedl customization: localization of taxonomies, wordnet vocabulary, and indexing properties<br> <br> Giuseppe,<br> <br> just a short sum up:<br> <br> 1) yes this is the way to do it - just copy the structure for yourself <br> - and please include reference in JOnto_base/src/context.properties. <br> The format is -> full Java path to the Context class = name of the <br> Context getting method (singleton implementation)<br> <br> 2) We will need to wait for Matuesz<br> <br> 3) yes, I clearly understand you. I was hoping to do such a thing with <br> Polish a long time ago. I just have never find enough time to proceed. <br> Together with Mateusz we have initiated a research that can be extend <br> to cover this topic [<a moz-do-not-send="true" href="http://library.deri.ie/resource/YuynUqdU">http://library.deri.ie/resource/YuynUqdU</a>]. The <br> idea is to mine the translations based on the co-occurrence of various <br> tags used. That is just one way to handle that - and do some cool <br> research at the same time :)<br> <br> Cheers,<br> <br> S.<br> <br> <br> On 14 Dec 2007, at 09:43, Giuseppe Bux wrote:<br> <br> > Hi Sebastian!<br> > Thank you for your quick answer.<br> > Please find inline my additional comments and questions.<br> ><br> > Cheers,<br> ><br> > Giuseppe<br> ><br> > -----Messaggio originale-----<br> > Da: Sebastian Kruk [<a moz-do-not-send="true" href="mailto:seb...@de...">mailto:seb...@de...</a>]<br> > Inviato: giovedì 13 dicembre 2007 23.55<br> > A: Giuseppe Bux<br> > Cc: Mariusz Cygan; Devel JeromeDL; <a class="moz-txt-link-abbreviated" href="mailto:jon...@li...">jon...@li...</a>; Mateusz <br> > Kaczmarek; Maciej Dąbrowski; <a class="moz-txt-link-abbreviated" href="mailto:mar...@li...">mar...@li...</a><br> > Oggetto: Re: Jeromedl customization: localization of taxonomies, <br> > wordnet vocabulary, and indexing properties<br> ><br> > Giuseppe,<br> ><br> > a quick answer to your questions (although might not be complete):<br> ><br> > 1) JeromeDL uses JOnto project [<a moz-do-not-send="true" href="http://sf.net/projects/jonto">http://sf.net/projects/jonto</a>] (yet<br> > another from my Corrib cluster) to handle controlled vocabularies.<br> > Hence to deliver new taxonomy you just need to have it in a properties<br> > file or RDF file (e.g. SKOS compliant), and you simply need to<br> > implement two interfaces in java and package it. The whole process<br> > does not take much effort. @Mateusz - could you provide a step by step<br> > description on a wiki?<br> ><br> > Well! I did the svn checkout of the project, so I have available the <br> > various jonto projects on controlled vocabularies.<br> > For each *_vocabulary project I view a common src module structure, <br> > where are included the *.ntriples and *.properties files, as well <br> > the *_Context.java and the *_Parser.java. Are these the properties/<br> > rdf files and the two java interfaces you mention above? So, if it <br> > is right, should I replicate a similar project for my own <br> > vocabulary? A step by step description of the process, by @Mateusz, <br> > could effectively help me.<br> ><br> > 2) with wordnet it is more tricky - JOnto uses JWNL library to handle<br> > WordNet dictionary. I was always hoping to be able to use other<br> > national dictionaries - but we never had time to test it. However, the<br> > infrastructure is there - and what I would suggest to do is to split<br> > current wordnet module into an abstract/base dictionary module and<br> > implementation modules, e.g., for wordnet. Depending on how much time<br> > Mateusz can contribute to the either we can do that or your team.<br> ><br> > An initial contribution from Matzeus could be the provision of some <br> > guidelines for splitting the wordnet module.<br> ><br> > 3) Answer to this question I leave to Mariusz, Maciej, and MarcOnt<br> > team :)<br> ><br> > Hope it was useful,<br> ><br> > Yes, it was useful enough!<br> > However my issue about metadata localization would extend the <br> > problem of the simple addition of new user dictionaries, towards a <br> > more large problem of managing simultaneously, as synonyms, words <br> > with the same concept but in different language (ex. the English <br> > word "apple", and the correspondent Italian word "mela"). I think <br> > this is one of the main problems of effective social inter-language <br> > indexing/bookmarking in shared digital libraries.<br> ><br> > Giuseppe<br> ><br> > Cheers,<br> ><br> > S.<br> ><br> ><br> > On 13 Dec 2007, at 16:50, Giuseppe Bux wrote:<br> ><br> >> Hi Sebastian, Mariusz!<br> >> I would start customization of my local instance of Jeromedl, by<br> >> localizing taxonomies, wordnet vocabulary, and indexing properties.<br> >> Localization will be targeted to an Italian audience. Here in the<br> >> following my envisaged approach and questions about. Please give me<br> >> your kind comments and suggestions.<br> >><br> >> 1) Localizing taxonomies: localization should be done by<br> >> enriching some selected current taxonomy dictionary (ex. DDC, Dmoz,<br> >> ACM, …) with synonyms in local language (i.e. in Italian). This in<br> >> order to manage domain indexing and related resource searching, by<br> >> using either English or local (i.e. Italian) terms defined in each<br> >> selected dictionary. To do that, I would follow steps should be as<br> >> follows:<br> >><br> >> a. To extract the RDF description of each dictionary from the<br> >> Sesame-Jonto repository, by a query template as follows:<br> >> Select distinct *<br> >> from {uri} p {value}<br> >> where p= <<a moz-do-not-send="true" href="http://www.w3.org/2004/02/skos/core#prefLabel">http://www.w3.org/2004/02/skos/core#prefLabel</a>> AND uri<br> >> like "*<a moz-do-not-send="true" href="http://www.corrib.org/jonto/acm/*">http://www.corrib.org/jonto/acm/*</a>"<br> >><br> >> obviously, for each query, the “like” value of the “where”<br> >> construct must be adapted to the uri of the given taxonomy.<br> >><br> >> I have some troubles with the sesame response, i.e.: the query<br> >> response in html format gives the expected sesame tabular form,<br> >> while the response in rdf format gives me a plain list such as:<br> >> uri p value uri p value General Literature uri p value General <br> >> ……..<br> >><br> >> I need instead a rdf/XML description: How can obain it? Could you,<br> >> eventually, provide me with the download site of RDF/XML<br> >> descriptions of the dictionaries stored in the sesame-Jonto<br> >> repository?<br> >><br> >> b. To initiate a dictionary customization project in an RDF<br> >> editor, such as Marcont portal or Protègè, by instantiating it with<br> >> the extracted rdf description of the given dictionary and with the<br> >> definition of synonyms of its terms, in local language.<br> >> c. To update the sesame-jonto repository with the new rdf<br> >> description of the given dictionary.<br> >><br> >> The “modify” actions of sesame should support the updating <br> >> process.<br> >><br> >> 2) Localizing wordnet vocabulary: it should be done simply by<br> >> replacing the current wordnet English vocabulary with a<br> >> correspondent commercial bi-language wordnet vocabulary (in my<br> >> case: Ital-Wordnet): does jeromedl implementation gives specific<br> >> constraints to do that?<br> >><br> >> 3) Localizing indexing properties: this should be done by<br> >> predefining additional marcont properties and make them available,<br> >> in the resource uploading form, for selection and instantiation<br> >> as xmarcont properties. Really a best approach should be to manage<br> >> the original marcont ontology for its localization as above for the<br> >> dictionaries: what about this topic in the Marcont project?<br> >><br> >> Not more to describe.<br> >><br> >> Looking for your kind comments and suggestions,<br> >><br> >> Cheers,<br> >> Giuseppe<br> >><br> >><br> >><br> >><br> >><br> ><br> > --------------------------------------------<br> > -- Sebastian Ryszard Kruk<br> > -- Lead Researcher, Project Manager<br> > -- Semantic Infrastructure Lab, eLearning Cluster<br> > -- Digital Enterprise Research Institute<br> > -- National University of Ireland, Galway<br> > -- mailto: <a class="moz-txt-link-abbreviated" href="mailto:seb...@de...">seb...@de...</a><br> > -- GG: 335067, Jabber: <a class="moz-txt-link-abbreviated" href="mailto:s_...@ch...">s_...@ch...</a><br> > -- Skype: sebastiankruk<br> > -- WWW: <a moz-do-not-send="true" href="http://www.sebastiankruk.com/">http://www.sebastiankruk.com/</a><br> > -- mobile (IRL): +353 85 7126591<br> > -- VoIP (PL): +48 52 5110114<br> > --------------------------------------------<br> ><br> ><br> <br> --------------------------------------------<br> -- Sebastian Ryszard Kruk<br> -- Lead Researcher, Project Manager<br> -- Semantic Infrastructure Lab, eLearning Cluster<br> -- Digital Enterprise Research Institute<br> -- National University of Ireland, Galway<br> -- mailto: <a class="moz-txt-link-abbreviated" href="mailto:seb...@de...">seb...@de...</a><br> -- GG: 335067, Jabber: <a class="moz-txt-link-abbreviated" href="mailto:s_...@ch...">s_...@ch...</a><br> -- Skype: sebastiankruk<br> -- WWW: <a moz-do-not-send="true" href="http://www.sebastiankruk.com/">http://www.sebastiankruk.com/</a><br> -- mobile (IRL): +353 85 7126591<br> -- VoIP (PL): +48 52 5110114<br> --------------------------------------------<br> <br> <br> <br> <br> -------------------------------------------------------------------------<br> SF.Net email is sponsored by:<br> Check out the new SourceForge.net Marketplace.<br> It's the best place to buy or sell services<br> for just about anything Open Source.<br> <a moz-do-not-send="true" href="http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace">http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace</a><br> _______________________________________________<br> JeromeDL-devel mailing list<br> <a class="moz-txt-link-abbreviated" href="mailto:Jer...@li...">Jer...@li...</a><br> <a moz-do-not-send="true" href="https://lists.sourceforge.net/lists/listinfo/jeromedl-devel">https://lists.sourceforge.net/lists/listinfo/jeromedl-devel</a><br> </font> </p> </blockquote> <br> </body> </html> |
