Clustering Enhanced Error-tolerant Top-k Spatio-textual Search

There are a large amount of Location-Based Services widely available on a variety of portable electronic devices. It is critical for them to efficiently support top-kquery considering both spatial and textual relevance. Considering both the errors in user input and the spatial databases, it is necessary to support error-tolerant spatio-textual search for end-users. Previous researches mainly focused on set-based textual relevance, which makes it difficult for them to find reasonable results when the input tokens are not exactly matched with those from the records in spatial database. We design a novel framework to support top-kspatio-textual search with fuzzy token matching. A hierarchical index is proposed to capture signatures of both spatial and textual relevance. Based on it, we devise two algorithms to preferentially access the nodes with more similar objects while those with dissimilar ones can be pruned. We further propose a clustering based approach to construct the index by leveraging textual information. We conduct extensive experiments on real world POI datasets, and the results show that our framework outperforms state-of-the-art methods by a significant margin.