Creating a Large Topic Map...

Advertise your Wandora project here...

Creating a Large Topic Map...

Postby akivela » Fri Feb 15, 2013 3:28 pm

Yejun Wu, David J. Dunaway, (2013) "Creating a Large Topic Map by Integrating Wandora and Ontopia", Library Hi Tech, Vol. 31 Iss: 1

http://www.emeraldinsight.com/journals. ... w=abstract

I haven't seen/read article itself. If you spot a free copy of it somewhere, please do drop a line.
akivela
Site Admin
 
Posts: 257
Joined: Tue Sep 18, 2007 10:20 am
Location: Helsinki, Finland

Re: Creating a Large Topic Map...

Postby aki » Sun May 26, 2013 10:14 pm

I finally bought the article 'Creating a large topic map by integrating Wandora and Ontopia' written by Yejun Wu and David J. Dunaway. After reading the article I felt a bit annoyed. The article writers didn't understood the hyperbolic projection of Wandora's default graph panel and they have somehow misinterpreted several topic map related terms used in Wandora. This post tries to clear up these weaknesses. I emphasize the overall article is good in showing an example of integrating two different Topic Maps tools.

Article's figure 2 shows an image of a simple graph. The image is captured from Wandora's default graph panel and writers say they didn't figure out how to make graph image larger. I suspect the problem was not about print quality images but the projection. Wandora's graph panel uses hyperbolic projection. This means that graph renders nodes and edges larger near center. Wandora user can change the curvature of the hyperbolic projection with a mouse wheel, keeping a shift key down while the wheel is rolled. When the curvature is small, the projection is almost plane and resembles the projection used in Wizigator. Using near plane projection it is easy to scale graph larger with mouse wheel. Also, opening connected nodes is possible using context menu options. I emphasize that I am claiming Wandora's graph visualization is better than Wizigator in Ontopia. It really depends on your taste. I would also prefer Wizigator for small scale and controlled graph browsing.

Another annoying part of the article relates to the table which compares topic map terms used in Ontopia, Wandora and Entity-Relationship Model. It looks to me that the article writers have misunderstood the topic map terminology used in Wandora. I suspect the varying quality of Wandora's documentation has facilitated the misunderstanding.

First misunderstanding relates to the concept of subject identifier. Article claims that Wandora's subject locator relates to topic map standard's subject identifier. This is not true. Wandora uses subject identifiers too. In Wandora, each topic may contain one or more subject identifiers. Wandora supports subject locators too and the concept of Wandora's subject locator is equal to the subject locator of topic map standard.

Article claims that Wandora's player is equal to a topic. Wandora uses term player to refer a topic that plays a role in an association. Thus, player is a name of a topic used in context of associations. Speaking about topics generally, Wandora also uses term topic.

Third, Article writes Wandora's role is equal to a topic type. Wandora uses term role to refer a topic that describes a player topic in an association. In other words, a role is a name of a topic that is part of an association, and describes a player. Speaking about topic types generally, Wandora calls a topic type as a topic class. It is a class-instance relation between two topics. More over, Wandora's (default) implementation of class-instance relation doesn't use associations.

Next misunderstandings relate to topic names. Wandora uses two level topic naming. In Wandora, a topic may contain single untyped base name and several typed variant names. A variant name is not a name type as the article claims. A variant name has to have a type which is essentially a topic.

It is somehow sad the article has so many misunderstandings related to Wandora specific topic map terms while it would have been very easy to ask us do we really call topic name types as variant names, for example. I can't imagine the problem was a giant typo in manuscript. I believe there is a deeper problem hidden here.

As a footnote, I have to confess that I would not be as annoyed if the publication Library Hi Tech wouldn't be as influential. I even found a copy in our public library.

Aki / Wandora Team
aki
 
Posts: 12
Joined: Mon Jun 28, 2010 2:40 pm


Return to Use cases

Who is online

Users browsing this forum: No registered users and 1 guest