Reverse spatial and textual k nearest neighbor search

Abstract

Geographic objects associated with descriptive texts are becoming prevalent. This gives prominence to spatial keyword queries that take into account both the locations and textual descriptions of content. Specifically, the relevance of an object to a query is measured by spatial-textual similarity that is based on both spatial proximity and textual similarity. In this paper, we define Reverse Spatial Textual k Nearest Neighbor (RSTk NN) query, i.e., finding objects that take the query object as one of their k most spatial-textual similar objects. Existing works on reverse kNN queries focus solely on spatial locations but ignore text relevance.To answer RSTk NN queries efficiently, we propose a hybrid index tree called IUR-tree (Intersection-Union R-Tree) that effectively combines location proximity with textual similarity. Based on the IUR-tree, we design a branch-andbound search algorithm. To further accelerate the query processing, we propose an enhanced variant of the IUR-tree called clustered IUR-tree and two corresponding optimization algorithms. Empirical studies show that the proposed algorithms offer scalability and are capable of excellent performance.

Learning to Un-Rank

“…For example, for the models with the Euclidean distance or cosine similarity, many ltering methods have been suggested based on di erent geometric properties [20,26,30]. However, for the models with textual queries only a few works exist, which cannot be fully applied to our problem as either the se ing is monochromatic [21], the indexing tree of the queries should be built which is not scalable to very high dimensions [39], the storage of k nearest neighbors for all the queries is required with a xed k [9], or only the conjunctive queries are considered [1]. In the remainder of this section, we present our algorithm for the dynamic generation of the RkNNs for the textual data.…”

Hybrid-LSH for Spatio-Textual Similarity Queries

“…Related Work. There are many studies on spatial textual similarity query processing [2,3,9]. A good survey of techniques can be found in [10].…”

“…[11] proposes a new indexing framework for processing the location-aware text retrieval query. [3] proposes a hybrid indexing structures called Intersection-Union-R tree (IUR-tree) and an efficient approach that take into account the fusion of location proximity and document similarity. For grid style, this category of indices combines a grid index with a text index (e.g., the inverted file).…”

“…One obvious way [1] to combine spatial distance and textual similarity is to use a spatial index to a spatial partition of the large dataset, which is most relevant to the spatial location of the query issuer, and for the given partition a string index is used to filter out those irrelevant objects by keywords matching and then rank the results using a hybrid ranking function that can combine spatial distance and textual similarity [2][3][4]. Spatiotextual data ordinarily are high-dimensional.…”

Efficient Algorithms and Cost Models for Reverse Spatial-Keyword k -Nearest Neighbor Search

“…Finally, a preliminary version of this work appears in [Lu et al 2011]. But in this journal article, we first propose a new cost model to theoretically analyze the performance of algorithms dealing with both location proximity and textual similarities, which is not discussed in any previous literature (to our best knowledge).…”