Improved Spatial Keyword Search Based on IDF Approximation

摘要

Spatial keyword search is a widely investigated topic along with the development of geo-positioning techniques. In this paper, we study the problem of top-k spatial keyword search which retrieves the top k objects that are most relevant to query in terms of joint spatial and textual relevance. Existing state-of-the-art methods index data objects in IR-tree which supports textual and spatial pruning simultaneously, and process query by traversing tree nodes and associated inverted files. However, these search methods suffer from vast number of times of accessing inverted files, which results in slow query time and large IO cost. In this paper, we propose a novel approximate IDF-based search algorithm that performs nearly twice better than existing method, which are shown through an extensive set of experiments.