A Recommendation Technique for Spatial Data

摘要

Recommendation functionalities have been recently considered in traditional database systems as an approach for guaranteeing a satisfactory interaction with the database also to users with a low or moderate technical skill or in presence of huge volumes of, potentially heterogeneous, data. Recommendation is performed by extending query results with additional and potentially interesting items. Among the proposed techniques, current-state approaches exploit the content and the schema of a query result as well as the database instance in order to recommend new items. While some preliminary current-state approaches have been proposed for relational databases, in this paper, we claim that current-state approaches can also be relevant for providing new ways of interactions in spatial databases. To support our claim, we present a current-state recommendation approach for spatial data and topological queries. The proposed approach exploits the principles of locality and similarity between topological predicates to recommend new spatial objects besides those precisely returned by a query. An index-based query processing algorithm for the proposed recommendation operator is also proposed, to guarantee an efficient computation of recommended items.