Building Semantic Relationships Incrementally in Dataspace

Recently, the notion of Dataspace has been introduced as a virtual space where many data sources are managed regardless of their structures and locations; this, with the aim to return best efforts answers to user´s queries. In fact, a dataspace is a set of participants which are data sources connected to the dataspace and a set of relationships between those participants. A data source, a participant, could be a file, a relational database, XML repository, web pages and so on. However, no formal definitions about relationships which could be built among the data sources have been proposed. Therefore, one of the challenges occurring in this new abstraction of data management is to incrementally build meaningful relationships between sources. In this paper, we introduce SEM-HDM, a SEMantics-based Heterogeneous Data Management which aims at constructing semantic relationships between heterogeneous sources of data based on a defined syntax and semantics. In fact, SEM-HDM first constructs a Semantic dictionary and then a reasoning dictionary which is improved incrementally by analyzing user´s activities on a given dataspace. We show on an example of three heterogeneous sources of data how SEM-HDM builds his semantic and reasoning dictionary.