[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Subject: RE: HM.interactions: initial questions
Hi Niclas, Mark, > -----Original Message----- > From: Niclas Olofsson [mailto:gurun@acc.umu.se] > Mark Brownell wrote: > > So the bottom line for this idea is to provide the information to a commonly > > used rendering machine I am not sure if I understand what you mean by "rendering machine". If I am getting this right though: The SW and HumanML in particular can help managing data and form useful information out of them and I am sure that we will be introduced to more complex, rich GUIs, possibly very dependent on the markup models themselves. Examples may include queries in natural language (both written and spoken), intelligent navigation, complex agent interfaces etc. But one simply cannot predict those. > > Has anyone provided a reasonable selection > > process that determines what will end up being rendered on the user's > > machine? IMHO, that should be handled by the application and has nothing to do with HumanML. > What do you suggest constitutes this selection process? We have had > discussions about approaches where HumanML could provide input to lower > level technologies and middleware (EMOTE, HAnim, etc). Myself I have a > very implementational approach to HumanML. HumanML, no matter how you > described it, must present a use case where the primary focus is to > render a noticable result back to the actors. If this will be a simple > "DOH" or a complex "HMM", I really couldn't say. HumanML will simply be a data/knowledge stream. I used to be trapped in something like: Human-A is a class of person, he currently holds a property "GeneralMoodState" with a literal (Sean will start yelling now) value of "90%" and I will probably do something like <pseudo type="ECMA"> Human-A.faceAppereance.showEmotion(respondTo(ToBoolean(GeneralMoodState) )); </pseudo> and display a smiley or something. But, we could do a lot more and try to resolve more logical functionality for practical apps. Resembling human nature into applications is one thing (that doesn't attract me much), building upon human aspects as a form of data/knowledge is another. So in my future house, I will have an intelligent fridge connected to my house server. The server knows I have a dog as: <daml:Class rdf:ID="pluto"> <rdfs:subClassOf rdf:resource="#pet"/> <rdfs:subClassOf rdf:resource="#dog"/> <!-- the user was stupid enough to provide the system with the dogs preferences --> <humlDesc:likesToEat rdf:resource="/supermarket/goods/forPets#catFood"/> <!-- but the system knows what Pluto needs cause it is a subclass of #dog --> <humlDesc:likesToEat rdf:resource="/supermarket/goods/forPets#dogFood"/> <myRules:budgetPriority rdf:resource="/myRules/budget#highest"/> </daml:Class> So, the system should just order the best kind of dog food available (probably based on the latest medical data about Pluto!) without taking cost under account. If you love dogs, this may be practical! My point is, let's try to get away from rendering and towards knowledge-centric (with the human as a context of course). [anonymous resource ;-)] http://www.daml.org/2001/03/daml+oil-walkthru.html Kindest regards, Manos
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]
Powered by eList eXpress LLC