In this thesis we deal with the problems outlined above and we focus on the problem of semantically accessing distributed information concerning e-service management and Information System interoperability. Specifically, as for the first research theme, we describe AESM (Adaptive E-Service Manager), a platform based on both Intelligent Agent technology and XML whose goal is to adapt service delivery to both exigencies of users who are benefiting from the services and the technical features of the devices they are exploiting. In AESM, each user is associated with a profile describing his past behaviour and his preferences; in addition, each device is described by means of a suitable profile registering its technical features. AESM analyzes the information stored in both user and device profiles and, then, is capable of both selecting and suitably presenting the most relevant information contents for the user. AESM has been specialized in various application contexts, like e-commerce, e-government, e-recruitment and e-learning; several experimental tests point out its effectiveness.

As for the second research theme, in this thesis, we illustrate XIKE (XML Intensional Knowledge Extractor); this system proposes a suitable formalism for describing the contents of an XML Schema and exploits suitable graph-matching algorithms for detecting a wide range of interschema properties (i.e., synonymies, homonymies, hyponimies, overlappings and sub-schema similarities) existing between two (or more) XML Schemas. After this, XIKE exploits the derived interschema properties for carrying out both the integration and the clustering of involved XML Schemas. The effectiveness of XIKE is confirmed by a large number of experimental tests.