There are some important remarks which can be observed from the depicted architecture:

• It should be noted that the concept of “reusable learning object” encompasses not only contents, but also packaged activities of various granularities, with embedded pedagogical methods (e.g. as described by IMS LD), which results in a considerable richness in the underlying ontologies describing goals and capabilities.
• Repositories in the architecture as well as service requests and responses are built on top of WSMO and WSML frameworks.
• Content providers and registries can be distributed in several layers, thus enabling intermediation for additional services. For example, metadata regarding quality of resources can be added to the architecture as a mediation layer, thus being able to exploit the diverse approaches to quality of learning objects as are being realized in repositories like MERLOT or CAREO.
• Discovery (i.e. asking for providers with certain capabilities), selection (i.e. asking for learning objects complying with certain learner and situation characteristics) and composition (i.e. putting together several learning objects into a higher level unit) are the key processes that will be addressed.
• Composition itself can be distributed to content providers or simply intermediaries, since the process of composition can be expressed in terms of other discovery and selection phases.
• Negotiation can be implemented as a stepwise process of discovery and selection, in which the learning object criteria become more specific in each round.