“StagIt” - Machine Learning based Real Time Multi-Tag Recommendation System for Stack Overflow Questions in the domain of C# technology

Abstract

Community Question Answering platforms have been trendy in recent times. These are web-based information systems that connect users with similar interests. Stack Overflow is one such forum popular among the software development community. The website uses tagging as a strategy to categories question contents and connect questions with the most expert users in particular technological domains. Currently, the tagging feature operates as a manual process. “StagIt” was implemented to eliminate the extra time and ambiguity from the developer activity by automating the tagging process. The application is designed as a Chrome plugin in the Stack Overflow website, and covers selected sub-domains of C# technology. The proposed solution obtains the question title and body from users and predicts the tags on the fly. The system’s core leverages multi-label classification techniques and TF-IDF feature extraction and LinearSVC machine learning algorithm.