From: harald k. <sc...@ma...> - 2004-09-02 11:43:58
|
Hi Christoph, just one comment. >Hi Bernard > >that sounds pretty interesting and I think it's closely related to the >task that Bret is facing currently: How could the visualisation of a >given TopicMapObject or a bunch of topicmap objects be more easily >customized. > >I think this alludes to several topics: > >1. Panckouckes abstraction process should be conceptually divided into >at least two chunks:=20 >The first (S1) is to "extract" the relevant TopicMapObjects from the >Topicmap.=20 >=20 >The second (S2) is to transform the extracted Objects into a Model that >is suitable to be displayed by renderers. >=20 >Both stages should be customizable. A TologAbstractor for example may >always perform a TologQuery to extract objects (S1), but how they are >transformed into a model could be up to a ModelBuilder that is passed >into the abstraction process.=20 > > > >2. The renderers must be openend for customisation as well. Some simple >solution would assign icons to Gestalten via a HashMap and pass this map >into the Graph-Renderers. Just for the beginning > =20 > We already have something like that, the Representations and=20 RepresentationRegistry which were part of the TopicMapViewer (the web based panckoucke application) , the only thing we need to do is=20 implement it in the other Renderers. I will have a Look at this as soon a= s i find some time (expect somebody else volunteers ? :) ). > >3. TMNav needs elements (both conceptual and gui) to assemble a bunch of >settings that influence the abstraction and the rendering.=20 > >3a. Theese 'assemblies' should either be applied automatically to an >abstraction demand or they should be chooseable by the user. > >3b. In order to apply them automatically we need some general >rule-mechanics that enable us to express directives like 'a topic of >type X' or 'all topics that has an occurrence of type Y'. Maybe we could >use a tolog query here... > > >Sounds really interesting and it would be certainly great to have it.=20 >But we have to concretise it a bit more... > >Anayway, thanks for the suggestion. If you already thought further, I >would be glad to hear it > > >Bye >c > > > > > > > > >Am Mo, den 30.08.2004 schrieb vt...@ne... um 11:04: > =20 > >>Hello, >> >>Fist of all, I've just download the latest version of TM Nav : >>Whaooooo !=20 >>Real improvement against the 0.2.5 one : I'm really impress by the new >>"abstracting" and "rendering" features, which are really very useful. >> >>I mail you to know if you plan to include some *cross* TM navigation >>features in complement to the *direct* TM navigation feature you've >>implemented in the latest release ... >> >>I mean :=20 >> by *direct* TM navigation : navigation allowing you to know >>*everything* about *a unique topic* (id, all name(s), all PSI(s), all >>occurrence(s) and all r=F4le(s) player, ...) >> by *cross* TM navigation : navigation allowing you to know *only few >>things*, but about *all the topics of one given topic type* (some >>name, some occurrence(s) and some r=F4le(s) player), ...) >> >> >>Imagine something which allow you to define sorts of extraction >>"patterns" : >> >>One "pattern" define : >> - 1 : a given topic type to begin the "cross TM navigation" >> - 2 : the occurrence you want to see via the definition of "given >>occurrence types [of the given topic type]) >> - 3 : the topics you want to see via the definition of "given role >>types involved in a given association type where the given topic play >>a given role type" >> - and for each topic define again point 2 and point 3 >> - and so on ... >> >>For example : I want to "extract" :=20 >> - all topics of the *people* topic type=20 >> - with *e-mail* "occurrence type"=20 >> - with *telephon* "occurrence type"=20 >> - ... And perhaps other kind of "occurrence type"=20 >> - and all the topic of "company type" playing the role of "employer" >>in a *employ* "association type" where the given people play a >>*employe* "role type" >> - with *home page* occurrence type"=20 >> - ... And perhaps other kind of "occurrence type"=20 >> - and for each *company* : all the topic of "company type" playing >>the role of "partner" ... >> - and for each *company* : all the topic of "company type" playing >>the role of "subsidiary" ... >> - ... >> >> The idea is to be able to use patterns to navigate TMs (with sort >>pr=E9-define differents perspectives) ... For one given topic type you >>have one or more pattern defined, if there is only one pattern you >>take this pattern otherwise you give the choice to the user (by >>contextual menu for exemple) ... And when the "view" is extract you >>can apply to the extracted "output" topics the available pattern(s) >>according to them type by the same processus I describe for the >>"input" topic type ... >> >>A pattern should be valid for a set of TMs conform to a given >>ontology. >>I think if the topic map is driven by a ontology it is more powerful, >>because topic type are declare "by intention" in the ontology and not >>"by extention" in the KB ... And in this case you can associate a set >>of "pattern" to a given ontology and use it for each map conform to >>this ontology ... But it's not mandatory to have the feature I >>describe above >> >>Hope I was clear in my explanations >> >>Regards and thank a lot for such wonderful tool >> >>Bernard >> >>------------------------------------------------------------- >>NetCourrier, votre bureau virtuel sur Internet : Mail, Agenda, Clubs, >>Toolbar... >>Web/Wap : www.netcourrier.com >>T=E9l=E9phone/Fax : 08 92 69 00 21 (0,34 TTC/min) >>Minitel: 3615 NETCOURRIER (0,16 TTC/min) >> >> >> >>------------------------------------------------------- >>This SF.Net email is sponsored by BEA Weblogic Workshop >>FREE Java Enterprise J2EE developer tools! >>Get your free copy of BEA WebLogic Workshop 8.1 today. >>http://ads.osdn.com/?ad_idP47&alloc_id=10808&op=3Dclick >>_______________________________________________ >>Tm4j-users mailing list >>Tm4...@li... >>https://lists.sourceforge.net/lists/listinfo/tm4j-users >> =20 >> |