Ranked Spatial-keyword Search over Web-accessible Geotagged Data: State of the Art

Abstract

Search engines, such as Google and Yahoo!, provide efficient retrieval and ranking of web pages based on queries consisting of a set of given keywords. Recent studies show that 20% of all Web queries also have location constraints, i.e., also refer to the location of a geotagged web page. An increasing number of applications support location based keyword search, including Google Maps, Bing Maps, Yahoo! Local, and Yelp. Such applications depict points of interest on the map and combine their location with the keywords provided by the associated document(s). The posed queries consist of two conditions: a set of keywords and a spatial location. The goal is to find points of interest with these keywords close to the location. We refer to such a query as spatial-keyword query. Moreover, mobile devices nowadays are enhanced with built-in GPS receivers, which permits applications (such as search engines or yellow page services) to acquire the location of the user implicitly, and provide location-based services. For instance, Google Mobile App provides a simple search service for smartphones where the location of the user is automatically captured and employed to retrieve results relevant to her current location. As an example, a search for ”pizza” results in a list of pizza restaurants nearby the user. Given the popularity of spatial-keyword queries and their wide applicability in practical scenarios, it is critical to (i) establish mechanisms for efficient processing of spatial-keyword queries, and (ii) support more expressive query formulation by means of novel 1 query types. Although studies on both keyword search and spatial queries do exist, the problem of combining the search capabilities of both simultaneously has received little attention.