On Processing Top-k Spatio-Textual Preference Queries

In this paper we propose a novel query type, termed top-k spatio-textual preference query, that retrieves a set of spatiotextual objects ranked by the goodness of the facilities in their neighborhood. Consider for example, a tourist that looks for “hotels that have nearby a highly rated Italian restaurant that serves pizza”. The proposed query type takes into account not only the spatial location and textual description of spatio-textual objects (such as hotels and restaurants), but also additional information such as ratings that describe their quality. Moreover, spatio-textual objects (i.e., hotels) are ranked based on the features of facilities (i.e., restaurants) in their neighborhood. Computing the score of each data object based on the facilities in its neighborhood is costly. To address this limitation, we propose an appropriate indexing technique and develop an efficient algorithm for processing our novel query. Moreover, we extend our algorithm for processing spatio-textual preference queries based on alternative score definitions under a unified framework. Last but not least, we conduct extensive experiments for evaluating the performance of our methods.