Portlets (i.e. multi-step, user-facing applications to be syndicated within a Portal) are currently supported by most portal frameworks. However, there is not yet a definitive answer to portlet aggregation whereby data flows smoothly from one portlet to a neighbouring one. Both data-based and API-based approaches exhibit some drawbacks in either the limitation of the sharing scope or the standardization effort required. We argue that these limitations can be overcome by using an ontology-based approach. Faced by the difficulty of agreeing on a common API, the portlet can publish its ontology, and markup its outputs accordingly. The portlet consumer (e.g. a Portal) can then use a deep annotation approach whereby the outputs of a source portlet can be deep-annotated with the ontology of the target portlet. This mapping is referred to as a pipe between the two portlets. At querying time, the target portlet can be "fed" from the outcomes of the source portlet. We consider deep annotation as particularly valid for portlet aggregation due to the controlled and cooperative environment where portlets are used.
14th International Conference on World Wide Web, WWW 2005, Pages 372-381, Chiba (Japan)
ONEKIN, UNIVERSITY OF THE BASQUE COUNTRY