Identifying refactoring operations in source code changes is valuable to understand software evolution. Therefore, several tools have been proposed to automatically detect refactorings applied in a system by comparing source code between revisions. The availability of such infrastructure has enabled researchers to study refactoring practice in large scale, leading to important advances on refactoring knowledge. However, although a plethora of programming languages are used in practice, the vast majority of existing studies are restricted to the Java language due to limitations of the underlying tools. This fact poses an important threat to external validity. Thus, to overcome such limitation, in this paper we propose RefDiff 2.0, a multi-language refactoring detection tool. Our approach leverages techniques proposed in our previous work and introduces a novel refactoring detection algorithm that relies on the Code Structure Tree (CST), a simple yet powerful representation of the source code that abstracts away the specificities of particular programming languages. Despite its language-agnostic design, our evaluation shows that RefDiff's precision (96 percent) and recall (80 percent) are on par with state-of-the-art refactoring detection approaches specialized in the Java language. Our modular architecture also enables one to seamlessly extend RefDiff to support other languages via a plugin system. As a proof of this, we implemented plugins to support two other popular programming languages: JavaScript and C. Our evaluation in these languages reveals that precision and recall ranges from 88 to 91 percent. With these results, we envision RefDiff as a viable alternative for breaking the single-language barrier in refactoring research and in practical applications of refactoring detection.

Acquiring meaningful data to be employed in building training sets for classification models is a costly task, both in terms of difficult to find suitable samples as well as their quantity. In this sense, Active Learning (AL) improves the training set building by providing an efficient way to select only essential data to be attached to the training set, consequently reducing its size and even enhancing model's accuracy, when compared to random sample selection. In this paper, we proposed a framework for time series classification in order to monitor sugarcane area in São Paulo, Brazil. The AL approach consisted of selecting seasonal time series information from less than 1 percent of each class' pixels to build the training set and evaluate this selection by an expert user supported by distance measurements, repeating this process until both distance measurement thresholds were satisfied. In most years, the classification results presented about 90 percent of correlation with official estimates based on both traditional and satellite image analysis methods. This framework can then help Land Use Change (LUC) monitoring as it produced similar results compared to other methods that demands more human and financial resources to be adopted.