Developing of Similarity Operators to Support Research in Climate Changes

Abstract

Employing usual operators to compare common data by equality using the total order property to speed up the process is not suitable to the management of complex data,such as multimedia data (images, audio, long texts), time series and genetic sequences.The most suitable property in these cases is the similarity degree between elements. All operations use the similarity measure returnedby a similarity function defined in the domain, leading to build the similarity operators. The similarity query operators can be either unary or binary.The unary operators are used to implement select operations, while binary operators are used to implement join operators. However, the formal definition of these operators is incomplete, because of the lack of definition of set similarity operators, such as similarity union, intersection and subtraction.This project aims at defining these base operators, as well as binary and join operators. One of the main objetives in this project is to provide support to climate changes research, emphasizing Agrometeorology, where the similarity operators will provide support for detection of association patterns between remote sensors and estimated measures extracted from weather forecast simulators. The second objective is to support medical images research, detecting patterns of errors between diagnostics given by specialists and computer-aided diagnosis systems.