A study on algorithms for maximizing the influence score of spatio-textual objects
Μελέτη αλγορίθμων για τη μεγιστοποίηση βαθμού επιρροής σε spatio-textual αντικείμενα
Nowadays, more and more applications are used that manage spatial objects annotated with textual descriptions. Advanced query operators and data indexes have become, not just useful, but indispensable, in order to help users handle the huge amount of available data by answering efficiently in their queries. With these data, users are offered the opportunity to pose spatio-textual queries with their preferences. The results of such a query consists of spatio-textual objects ranked according to their distance from a desired location and to their textual relevance to the query. A problem that arises from this context is how to select a set of at most b keywords to enhance the description of a spatial object, in order to make the object appear in the TOPk results of as many users as possible. This problem is referred in later work as Best Term and it is proven that it is NP-hard. In this thesis we study the design and development of an algorithm that approximately solves this problem. The presented algorithm focuses on using efficiently the data structure of an IR-tree index, that is build over the spatio-textual data, in order to compute the b keywords needed. An extended number of experiments will be demonstrated that will show the effectiveness of the proposed algorithm. A comparative performance analysis will be provided for this algorithm and the already introduced algorithms as baselines. As it will be shown by the experimental studies, this algorithm is an efficient solution for the Best Term problem.