In many real applications, data are intrinsically uncertain due to measurement errors, interpretability issues, information incompleteness, etc. In those uncertain databases, users usually express quality requirements when the system evaluates their queries. However, as they may not be familiar with the contents of the queried database, their queries may be failing i.e., they may return no results or results that do not satisfy the expected degree of certainty. To provide users with relevant information in order to obtain alternative satisfactory results, we introduce a cooperative approach based on the dualization concept. This approach computes a set of meaningful subqueries (MFSs and XSSs) of the initial failing query, which is of paramount importance for query reformulation and relaxation purposes. The conducted experiments show that our proposition, a Mixed Dualization Matrix-Based approach (MDMB), outperforms existing algorithms, especially for large queries.

A large number of applications manage uncertain data. Usually, users expect high quality results when they pose queries with strict conditions over these data. However, as they may not be clear about the contents of the databases that contain such data, these queries may be failing i.e., they may return no result or results that do not satisfy the expected degree of certainty. In this paper, we deal with this problem, in the field of uncertain databases, by proposing an efficient approach that identifies the parts of the failing query, called Minimal Failing Subqueries (mFSs), that are responsible of its failure. Our approach also computes, in the same time, a set of Maximal Succeeding Subqueries (XSSs) that represent non failing queries with a maximal number of predicates of the initial query. We demonstrate the impact of our proposal with a set of experiments on synthetic and real datasets.

This paper presents a Query-and-Relax System (QaRS) designed to facilitate the exploitation of large knowledge bases. QaRS proposes a graphical interface to construct a SPARQL query and use different cooperative answering techniques. The proposed cooperative techniques help users in finding alternative answers when their queries fail or do not return the expected number of answers. The present demonstration includes three main relaxation strategies: (1)- automatic where the system automatically relaxes the query based on similarity measures, (2)- manual where the user can specify the conditions that can or cannot be relaxed as well as the tolerance values and (3)- interactive where QaRS computes the causes of the query failure as a set of Minimal Failing Subqueries (MFSs )a nd then the user chooses the relaxation operators according to these MFSs.

Recently, ontologies have been widely adopted by small, medium and large companies in various domains. These ontologies may contain redundant concepts (computed from primitive concepts). At the beginning of the development of ontologies, the relationship between them and the database was weakly coupled. With the explosion of semantic data, persistent solutions to ensure a high performance of applications were proposed. As a consequence, a new type of database, called semantic database (SDB) is born. Several types of SDB have been proposed and supported by different DBMS, where each one has its architecture and its storage layouts for ontologies and its instances. At this stage, relationship between databases and ontologies becomes strongly coupled. As a consequence, several research studies were proposed on the physical design phase of SDB. To guarantee the similar success that relational databases got, SDB has to be supported by complete design methodologies and tools including the different steps of the database life cycle. Such methodology should identify the redundancy embedded into ontology. In this paper, we propose a design methodology dedicated to SDB including the main phases of the lifecycle of the database development: conceptual, logical, deployment and physical. The conceptual design of SDB can be easily performed by exploiting the similarities between ontologies and conceptual models. The logical design phase is performed thanks to the incorporation of dependencies between concepts and properties in the ontologies. These dependencies are quite similar to the principle of functional dependencies defined in the traditional databases. Due to the diversity of the SDB architectures and the variety of the used storage layouts (horizontal, vertical, binary) to store and manage ontological data, we propose a SDB deployment à la carte. Finally, a prototype implementing our design approach on Oracle 11g is outlined.

With the increasing needs for the world wide enterprises to integrate, share and visualize data from various heterogeneous, autonomous and distributed sources data and Web data covering a given domain, the development of integration and reconciliation solutions becomes a challenging issue. The existing studies on data integration and reconciliation of results have been developed in an isolated way and did not consider the strong integration between these two processes. On one hand, ontologies were largely used for building automatic integration systems due to their ability to reduce schematic and semantic heterogeneities that may exist among sources. On the other hand, reconciliation of results is performed either by considering that all sources use the same identifier for an instance or by means of statistical methods that identify affinities between concepts. These reconciliation solutions are not usually suitable for real-world sensitive-applications where exact results are required and where each source may use a different identifier for the same concept. In this paper, we propose a methodology that simultaneously integrate source data and reconciliate their instances based on ontologies enriched with functional dependencies (FD) in a mediation architecture. The presence of FD gives more autonomy to sources when choosing their primary keys and facilitates the result reconciliation. This methodology is experimented using the Lehigh University Benchmark (LUBM) dataset to show its scalability and the quality of the reconciliation result phase.

Engineering models are computer-based models that enclose technical data issued from engineering domains. Those models usually implicit many of the details required to understand and interpret the data. In this context, integrating the results of models and querying the heterogeneous information is a challenge.