Discussion Meeting 3

-- VirginiaDignum - 16 Feb 2007

Back to Semantic Web wiki

• • [Xavier] Ontologies and folksonomies are often placed at opposite sides of an imaginary spectrum. Gruber argues that this is comparing apples with oranges. I still don't think he makes a good point. Aren't they for some part the same? I mean: in his own article he is trying to match them on one another. But of course, regarding structure they might be called opposites. We' re in for discussion here...

• • [Xavier] Shirkey is a great fan of 'wild tagging' – as am I, but when defending a tag named ' to read' I believe he overtaked himself. That's ridiculous! Wouldn't you agree?

• • [Xavier] Many articles say: there's no such thing as a complete ontology. If not for designer mistakes, it's simply because one can only act on the knowledge of the moment, including believes (there was an example of categorizations for religions and countries). Does it matter? If you built a good ontology – also meaning it's easily expandable, as many article argue – would the arise of new knowledge have to be a problem then?

• • [Xavier] Folksonomies and tagging can never work because of the highly personal value laden aspects of words. True or not? Is this then were ontologies come in? Or at the contrary?!

• • [Ruud] Shirkey says some interesting things, even though to_read is indeed ridiculous. But some pages before he also says: "As a long as a lot of people are tagging any given link, the rare tags can be used or ignored, as the user likes." Other users will 'help' ignoring tags like to_read.

• • [Arlette] Vickery says something about the uniqueness of ontologies. Is there or is there not an unique ontology of the world? And is there an unique ontolgy for a specific domain? If there is not even an unique ontology in the specific case, then what is the use of ontologies (if they cannot be exchanged with other systems)? Is an 'uppermost ontology' not an example of an unique ontology?

• • [Arlette] Shirkey claims that "tag overlap is in the system, but tag semantics are in the users". So trying to capture semantics for folksonomy would seem pointless. But are semantics always dependent on human context?

• • [Arlette] Semy states that "portions of the upper ontology can only be annotated and interpreted solely by human beings". Do we agree? Because this is not desirable at all! We want to create an upper ontology to make different systems able to communicate about the same thing, without the intervenience of humans. What will be the use of this upper ontology if it is the case that we still need humans to interpret it?

• • [Wilco] Gruber tries to make a standard for the folksonomy. Isn't this an hopeless effort when users can still do what they want?

• • [Wilco] Semy et al. discuss the use of upper ontologies. In fact upper ontologies are the basic of the semantic web, when they are integrated well with mid-level and domain specific ontologies. Shouldn't it be a task of the w3c to develop a unique upper ontology, so all big organizations can commit their ontologies to this standard?

• • [Wilco] Shirky has a great vision about the bottom-up semantic web. Tagging should be done by the user, and the web will moderate itsself. I only wonder how intelligent agents will handle a semantic web without a modelled ontology.

• • [Jan Jaap] Is it possible to make one unique upper-ontology, which can be used by anyone (so, for any domain)?

• • [Jan Jaap] According to Vickery, ontologies can be used to 'translate between a pair of languages'. Does this meen that, in the end, we will be able to talk in our own language on the internet, but still be understood by anyone, although they speak another language?
  • [Jan Jaap] Everything with a URL can be tagged by anyone, which seems nice because it implies freedom, but isn't that a problem too? Spammers who now use botnets to send spaim mail will be able to (ab)use those botnets later to (mis)tag things in such a way that the 'tag signatures' can get 'polluted'.. The so called 'group value' won't be worth anything anymore..

• • [Swathi] Ontologies do not have a unique upper class. Even the utmost top-level could be named a variety of things within a domain. So the usage of the upper-level domain must have constraints such, that everybody starts of at one certain upper-level ontology. But is this achievable?

• • [Swathi] Folksonomies are difficult to categorize. This leads to eventual chaos, because there is no restriction in the terms used and it does not per se fall within a specific domain. Folksonomies seem only useful on a personalized level and are difficult to place on an abstract level.

• • [Swathi] Ontologies within a domain will always face the interlingual problems. Even within a medical domain for instance, the terminology is different in different languages. Can ontologies be combined to solve this problem?

• • [Melanie] I think Upper Ontologies are very difficult to define, because everyone defines and structures things differently. In that sense, I think they are useless. But can they become useful when we talk about small domains?

• • [Melanie] I think we should both use folksonomies (for large domains) and ontologies (for small domains). Do you agree?

• • [Thijs] According to Shirkey, hyperlinks alone are enough when searching the web as proven by Google. Are they indeed sufficient to find what you want in the chaos that is the Web, or is there still use for categorization?

• • [Matthijs] Would it be a good idea to combine the ideas of ontologies and folksonomies by letting the users freely tag and let a select group of experts define the relations between the different tags?

• • [Ben] The ontology we use to communicate in everyday life seems like a network to me and not a strictly hierarchical structure. Can hierarchical ontologies ever represent the real world?

• • [Ben] Would "wild" and unrestricted tagging and automatic classification (e.g. based on shared properties) be feasable?

• • [Ben] During the last 40 years zillions of euros have been spent on data-modeling for good old-fashioned stand-alone applications. For instance the entity "client" should be mature by now. Can't we use these efforts to structure the semantic web?
• classassignmentgroup3-adjustingontology.doc: The ontology made by group 3 in the class assignment

• groep2-folksonomy.doc: the ontology made by grout 2 during class

• SemanticWeb_folks.doc: the ontology made by group 4 during class