Focusing Web Crawls On Location-Specific Content

Retrieving relevant data for location-sensitive keyword queries is a challenging task that has so far been addressed as a problem of automatically determining the geographical orientation of web searches. Unfortunately, identifying localizable queries is not sufficient per se for performing successful location-sensitive searches, unless there exists a geo-referenced index of data sources against which localizable queries are searched. In this paper, we propose a novel approach towards the automatic construction of a geo-referenced search engine index. Our approach relies on a geo-focused crawler that incorporates a structural parser and uses GeoWordNet as a knowledge base in order to automatically deduce the geo-spatial information that is latent in the pages’ contents. Based on location-descriptive elements in the page URLs and anchor text, the crawler directs the pages to a location-sensitive downloader. This downloading module resolves the geographical references of the URL location elements and organizes them into indexable hierarchical structures. The location-aware URL hierarchies are linked to their respective pages, resulting into a georeferenced index against which location-sensitive queries can be answered.