Encontrando os locais de interesse com maior concentração de objetos relevantes para um conjunto de palavras-chave

Abstract

Spatial data is increasingly present in our daily lives. We use various applications that use this data, such as Google Maps and Uber. There are a huge number of interesting questions that can be performed on these data. For example, a tourist maybe interested in hotels that have a lot of restaurants in its vicinity. This project proposes a new query type named Popularity Based Spatio-Textual Preference Query (PSTPQ), whose main contribution, that can select the best spatial objects taking into account the number of relevant spatio-textual objects, for a given set of query keywords, in its neighborhood. We present algorithms to process this query e ciently and evaluate the algorithms proposed in real datasets. Our experiments show that it is more e cient to use spatial indices (e.g. R-Tree) for distances less than 5 km in relation to textual indexes (e.g. Inverted File). In our experiments, the hybrid index processed the PSTPQ query with better performance. The PSTPQ query has as a di erential take into account the number of reference objects in the spatial neighborhood, in addition to selecting the reference objects from the textual description.