Frank Nack and Jacco van Ossenbruggen
Date: 23 Feb 2000
Within the project goals, i.e. to increase the level of automated adaption of varying user and system characteristics during the process of creating hypermedia presentations, a number of research topics need to be worked on. This paper reflects the ideas of the INS2 group of CWI on the different research topics of interest and the fields we feel that the group can best contribute to within the Dynamo project.
The first general remark is about the needs to decide upon a target application domain for the project. It seems not practical to develop a 'generic' system, since such a thing does not exist for presentation. It is of interest to identify particular representational structures for different media, presentational strategies, or user aspects which might be generally applicable. However, it is fair to say that particular domains will require special solutions. We are happy to work in particular on the aspect of picking a feasible domain and to sketch out the application within it.
The second general point we would like to stress is concerned with the underlying methodology for material retrieval and resulting presentation. The two potential ways of doing this are either bottom-up or top-down. Depending on the chosen application one of the two might be preferable - yet we think it would be good to provide both strategies. This would allow us to get a better understanding of representational issues (depicted vs. propositional), query issues, and design issues.
The different parts of importance within a an 'automated' hypermedia system are at least:
We will now look at the tasks of each part and try to shape out were we see potential to contribute to the development.
From our side it is assumed that the database will be there. The database contains not only the media items but also relevant meta information about the items. We do not care which type of database is picked but we are very much interested in the conceptual modeling of the meta-data. Meta-data is here understood not only as information about relevant presentational items but also as additional 'content' description. These might be helpful for the use of an ontology describing the domain. Another use might be the ability to generate new media items if the required information is not available among the existing media items - but for this project media generation is not of paramount importance, though repurposing or restructuring might be. All aspects within the representation of content/meta-data with regards to platform optimisation are of great interest to us.
A question that needs perhaps special attention is: what is required in situations where not only the data but also the meta-data needs to be streamed (run time delivery of meta-data).
Though an important aspects of the project, we are not interested in query optimisation. However, since we see the need to investigate the different approaches on using retrieval results, we are very much interested in the structure of query results.
As far as the user profile is concerned, this might be a research topic other groups are more interested in.
This is actually the part we are most interested in and, due to our research history, we are strong in supporting the project here. In particular the aspect of hypermedia composition rules are of paramount interest for us. This aspect covers questions, such as what is the interaction between space, time, link; what type of design rules are required (e.g. fonts, emphasis, etc); and which 'conceptual dependencies (rhetorics, narrative structure, summarization,)' need to be modeled and how should that be done.
Note, that the interrelationship between the different parts is obvious and none of them can be looked at exclusively. The above statements should just support the discussion on work tasks and schedule.