D2.3: Models

 

Summary

In the previous section, the different facets of representation of a person have been examined, and for each of them the application domains that are the most likely to utilise each of these facets has been identified. To date and to our knowledge, a similar work had not been conducted on a large scale, although some tentative attempts, specific to some domains (in the domain of e-learning, for instance, Diogene (2002) tries to identify other user models in different domains in order to provide guidance in the definition of the student model), are worthy of mentioning. 

 

This document has revealed some notable findings. Firstly, it has been observed that several application domains (human resource, e-learning, justice) have already started to create quite complete and holistic models of the person that fulfil the needs of this domain (see the example of HR-XML, IMS/LIP, or JXDM in the annex). In some cases, domains have independently developed different representations (in some cases very elaborated) for representing the same objective (such as for the representation of the address of a person), raising the question of the duplication of work. 

In other cases, attempts to reuse some of the existing specifications have been observed. For instance there is a clear similarity between the specifications developed for e-learning in IMS/LIP and in Human resources with HR-XML that indicates some clear interdependence between the two specifications. Also, the Liberty ID-SIS Service Specification explicitly mentions the use of another specification (vCard) as part of the specification. 

Currently, the modelling of identities appears to be composed of dispersed initiatives, is still very much under construction (and incomplete), and very complex for the most advanced initiatives. Such initiatives would probably benefit from additional cross-disciplinary studies, such as this, even if the benefit of a holistic approach still needs to be demonstrated (because of its complexity). It has been noted in particular that there is relatively limited interaction between different “Universes” (such as identity management and intelligent adaptive systems) for which representation of persons is very important. 

Finally, and not surprisingly, it can be noted that some aspects of the person modelling, such as the behavioural or social characteristics, are still largely underdeveloped, especially considering that they will be essential in the new technologies of tomorrow, such as intelligent mobile and ubiquitous applications. 

 

 

27 / 53