Thursday, August 7, 2008

Authoring of Adaptive and Adaptable Hypermedia (A3H) Workshop

Attended and presented at the A3H workshop in Hannover last week, have been meaning to outline the talks that were there so here goes:

First of all the link for the workshop is here

The first talk was by Fawaz Ghali, and the work he has been doing in Warwick trying to connect the social web in the form of folksonomies with the semantic web in particular ontologies. Advantages of this would include semantic relations between folksonomy tags, enabling of reasoning on the social web and augmenting the authoring process of adaptive hypermedia by providing rich free heirarchically structured data. Ghali process consists of three main phases, firstly mispelt tags are filtered, tags are grouped according to co-occourance value, and finally the groups are mapped to matching elements of ontologies.

Maurice Hendrix then presented his paper which outlines how a meta level can be added to the LAG language which allows for a fuzziness of values in the LAG rules. This solves the following limitations identified by Hendrix:
-adaptation is on a per concept basis
- adaptation engine does not allow for non-instansiated variables
- cant combine multiple strategies.
I understood how you could combine multiple strategies with this approach, as you dont have to strictly specify a leaners model value so beg_engineers and beg_cs can be treated as beg students, but didnt really get how the other limitation are solved

Sergio Gutierrez looked at the formalization of exercises, which reminded of the work we did on the ontaware project looking at the same thing which we published at SWEL workshop at AIED05. This work is based on a parameter approach and is interesting, as it shows how many questions can be asked of the same root question.

The work of Ortigosa et al. looked at what was the least amount of questions you need to ask in order to understand the learner enough to provide adaptation

Another paper from Warwick looked at the use of collaborative authoring in the development of AEH for the MOT tool. To do this the CAF was extended to include collaborative elements, such as tags and opinions. This allowed for feedback on material in AEH - which could then be used in the adaptive process.

Eric Ras presented a really interesting paper which outlined the use of semantic wikis to capture knowledge, which are then used as the basis of the creation of learning spaces.

After the papers were presented there was a discussion panel, where interesting points were raised about the lowering of the barriers for authoring in developing AH. Most people agreed that the cost-benefit ration has to be addressed, in that we must outline how beneficial it is for an author to create AH and how we can lower the cost by providing better authoring tools. This discussion made me think that the application of AEH in the academic scenerio (the most common test bed) perhaps doesnt make sense anymore. Academics do not have the time to spend on authoring and do not see much in the way of recognition when they do develop good AEH. I think perhaps the focus of the AEH community should be on the corporate sector as this sector spends a massive amount of money in getting employees up to speed at the start of their career and also during their career with a particular company. This represents not just a massive cost in terms of the salaries paid to the employee and to the trainer, but also in terms of opportunity cost, lost while the employee is training. Anything to cut the time employees spend training will represent a massive cost saving for business. This is where we should be aiming our AEH apps as the cost - benefit ratio is significantly more in our favour here. Then when we have matured our authoring tools here, perhaps then we could start to look at the academic scenerio which will be much harder to break into.

Overall a great workshop, where real debate took place about the future of authoring in this area.

No comments: