Utilizando técnicas de aprendizado de máquina para classificar o tempo de resposta de perguntas no stack overflow

Abstract

The internet has changed the way knowledge is shared. Nowadays it has become common for people to search their queries in search engines, thus finding a wide range of pages, associated with the searched subject. Question and answer sites appear as a quick and effective way of sharing knowledge. On these sites, users post their queries and expect other users to provide a solution to their question. One of the problems with having questions answered by other users is that sometimes the answer can take a while to be provided. This work aims to extract features from the questions posted on Stack Overflow and train a model capable of predicting whether a question will have an answer within a day. To train the model, the extra tree classifier algorithm was used, the SMOTE technique was applied to the dataset for resampling the data. The results obtained show that the model had a precision of 72% .