Indexing Structures for Geographic Web Retrieval Leonardo Andrade 1 and Mário J. Silva 1 University of Lisboa, Faculty of Sciences Abstract. Context-aware search in mobile web environments demands new re- trieval methods that rank web resources based on the proximity to users’ loca- tions. This paper presents the indexing and ranking architecture of a new geo- graphic web retrieval system that can accept the user location as input and ranks searched items based on the estimated distances between the users and the re- sources. We describe the criteria to be considered by a geographic similarity metric and the indexing data structures that we created for fast selection of web resouces based on proximity. 1 Introduction In ubiquitous computing environments, the number for geographic information retrieval applications is quickly growing. With the massification of third generation mobile tech- nologies and wi-fi networks new web retrieval applications are no longer confined to the home desktop. For example, travelers who search for “restaurants” in a web search engine are not in general interested in the list of the most popular restaurants in the world - but in those closest to their present location. A significant portion of documents have a local scope, understood as the geographic region covered by the document. It is estimated that one fifth of the queries submitted to search engines have a geographic context [1]. This context can also be seen as the scope of the query. However, classic systems designed for text retrieval perform poorly when a query related to a location is submitted. To achieve good results, a natural solution is to build a system with geographic rea- soning capabilities. It is necessary to build structures that index spatial information re- lated to the documents and develop algorithms for geographic ranking. Previous works supported geographic indexing and searching under a framework based on classic GIS (Geographic Information Systems) [1, 2]. However, we do not rely on classic spatial in- dexing structures. Our geographic ranking algorithm combines spatial and non-spatial features, but the index structures are similar to those used in classic text retrieval. This work is part of the GREASE project which researches on methods, algorithms and software architecture to develop geographic-aware IR systems [3]. GREASE de- veloped and published an ontology of Portuguese geographic names and has been re- searching methods for classification of documents by geographic scope. Disambigua- tion of geographic names is performed while assigning scopes to documents. This paper discusses access methods to retrieve documents classified with scopes.