Busca por grupos de pontos de interesse usando processamento qualitativo de regiões espaciais.

Abstract

In order to decrease mobility difficulties, create travel itineraries or save time, people usually face the need to find Points of Interest (POI) that share the same spatial extent or are located in interconnected regions. POI searches using web tools focus exclusively on queries for a single type of establishment (e.g., restaurant or hotel) or for keywords referring to a place’s name (e.g., Starbucks or Subway). Retrieving a group of places by using keywords and con- nectivity relationships between their regions is a current challenge for search tools, as they do not consider POI’s representation as a region, but as a point in space. The main exist- ing solutions are based only on the distance’s calculation between these points. However, few tools are able to assess the connectivity relationships between POIs’ spatial extensions. In this context, the present study proposes a textual search technique for a group of POIs, based on the qualitative relationships between spatial regions. With the technique, it is pos- sible, for example, to find different types of establishments that are neighbors or are located in the same building. The solution, named Topo-MSJ, defines a pattern of qualitative spa- tial queries by using the combination of a state-of-the-art algorithm, the “Multi-Star-Join” (MSJ), along with a spatial model of qualitative relationships, entitled “Region Connection Calculus” (RCC). Topo-MSJ, in a single query, retrieves up to four different types of spatial connectivity relationships and is particularly suited to the Big Spatial Data scenario. The algorithm’s efficiency is evaluated through the proposed solution’s comparison with other works that use qualitative indexing solutions, in addition to a comparative evaluation of the queries in SQL format. The databases used in the experimental evaluation include approx- imately 900,000 POIs from the American states of California and New York, as well as textual and geographic databases from the European Environment Agency (EEA), which are used by the qualitative indexing works compared to this research. The experimental results indicate that the proposed algorithm is more efficient (in terms of execution time) than SQL queries performed on spatial databases. Furthermore, it is shown that even allowing the ex- ecution of more complex queries, it is possible to achieve similar execution times compared to other existing qualitative indexing solutions.