Ensemble learning approach to classify user defined functions in Java programs

Source code of a computer program is vital for any development in information technology industry. Many experienced IT professionals reuse old source code functionalities in order to speed up the development process and utilize the resources. However, there are no any efficient approaches to identify the source code functionality name. Topic modelling is one of the statistical approaches to extract the topic of the semantic function by considering function names, comments and identifiers of a source code. These approaches tend to heavily depend on the programming best practices rather than the source code algorithms. In response to these observations, the author performed research in this work on extracting source code using java parser library and classifying source code functionality names using the Ensemble learning model. To implement the ensemble method, the author merged Decision Tree and Nave Bayes algorithms and use the majority voting mechanism to obtain the result of the ensemble approach. The data was collected from git hub repository and author built an algorithm to extract programming language keywords of the source code in order to feed the model. This is a novel method to topic modeling and the first effort at creating a machine learning model to classify the source code's semantic function names.