At present a large amount of research exists into the design and implementation of adaptive systems. However, not many target the complex task of authoring in such systems, or their evaluation. In order to tackle these problems, we have looked into the causes of the complexity. Manual annotation has proven to be a bottleneck for authoring of adaptive hypermedia. One such solution is the reuse of automatically generated metadata. In our previous work we have proposed the integration of the generic Adaptive Hypermedia authoring environment, MOT (My Online Teacher), and a semantic desktop environment, indexed by Beagle++. A prototype, Sesame2MOT Enricher v1, was built based upon this integration approach and evaluated. After the initial evaluations, a web-based prototype was built (web-based Sesame2MOT Enricher v2 application) and integrated in MOT v2, conforming with the findings of the first set of evaluations. This new prototype underwent another evaluation. This paper thus does a synthesis of the approach in general, the initial prototype, with its first evaluations, the improved prototype and the first results from the most recent evaluation round, following the next implementation cycle of the spiral model [Boehm, 88].