We address the problem of generating a mediated schema from a set of relational data source schemas and conjunctive queries that specify where those schemas overlap. Unlike past approaches that generate only the mediated schema, our algorithm also generates view definitions, i.e., source-to-mediated schema mappings. Our main goal is to understand the requirements that a mediated schema and views should satisfy, such as completeness, preservation of overlapping information, normalization, and minimality. We show how these requirements influence the detailed structure of schemas and view definitions that are produced. We introduce a normal form for mediated schemas and view definitions, show how to generate them, and prove that schemas and views in this normal form satisfy our requirements. The view definitions in our normal form use stylized GLAV mappings, for which query rewriting is easier than general GLAV mappings. We demonstrate the efficiency of query rewriting in a prototype implementation. 1.

Adaptive Hypermedia builds upon the annotation and adaptation of content. As manual annotation has proven to be the main bottleneck, all means for supporting it by reusing automatically generated metadata are helpful. In this paper we discuss two issues. The first is the integration of a generic AH authoring environment MOT into a semantic desktop environment. In this setup, the semantic desktop environment provides a rich source of automatically generated meta-data, whilst MOT provides a convenient way to enhance this meta-data manually, as needed for an adaptive course environment. Secondly, we also consider the issue of source schema heterogeneity, especially during the automatic metadata generation process, as semantic desktop metadata are generated through a lot of different tools and at different times, so that schemas are overlapping and evolving. Explicitly taking into account all versions of these schemas would require a combinatorial explosion of generation rules. This paper investigates a solution to this problem based on malleable schemas, which allow metadata generation rules to flexibly match different versions of schemas, and can thus cope with the heterogeneous and evolving desktop environment.

Schemas are very likely to be changed from time to time because of requirement changes, design revisions, database migration and such scenarios. Especially in a multi-user environment, schemas may be used by different groups or people and are modified to different versions. At some point in time, it is necessary to reintegrate the different versions and have a final unique version of the schema. Basically, the problem of creating the unique version is to merge the modified schemas. Previous works on schema merging describe how to merge two schemas given the mapping between them. In those works, the two schemas need not come from a common source schema. In fact, the original schema is often unavailable. However, in our scenario, we have the original schema, and we use it in decision-making. This simplifies what might otherwise be a complex matching procedure. We attempt to find a generic solution to the schema reintegration problem (i.e., when the original schema is present). In this thesis, we created a framework that implemented the schema reintegration algorithms using generic model management operators. These generic operators have been widely mentioned and explained abstractly in many previous papers. Here we have implemented each operator required, which necessitated formally defining and creating the algorithm for each operator used for this specific schema re-integration purpose. Our contributions are: (a) determining that the generic operators can be used for schema reintegration, and (b) designing, implementing, and analyzing the model management operators used in details. iii Table of Contents Abstract................................. ii

P2P Data Management Systems (PDMSs) allow the efficient sharing of data between peers with overlapping sources of information. These sources share data through semantic mappings between peers. In current systems, queries are asked over each peer’s local schema and then translated using the semantic mappings between peers. In this thesis we propose that a mediated schema can benefit PDMSs by allowing access to more data and by making that access comprehensible. We present our system- MePSys, which uses the mediated schema as a media for query translation in relational PDMSs. We show how to create a mediated schema in PDMSs automatically using the semantics of the existing mappings provided to translate queries. We further discuss how to update the mediated schema in a stable state, i.e., after the system setup period. ii Table of Contents Abstract................................. ii