Abstract
Stackoverflow is a platform or forum for people to gain knowledge, get solutions and interact on topics related to different programming concepts. Stackoverflow is one of the biggest Question and Answer forums with more than a million users. There is a wide variety of questions covered on this platform that are segmented with appropriate tags. The user by default enters the tag manually. Due to an enormous number of tags, it becomes difficult to search for the correct tag, many often left unconsidered. An automatic tagging system can be employed that shows likely tags depending on the text entered. The objective dataset chosen by us consists of about 0.5 million such questions, which can be technical or non-technical, collected from various information-dense websites, primarily Stackoverflow. We introduce a tag association scheme (TagAssc) along with code-analysis and strategic tag sampling schemes to get better accuracy scores in predicting tags with limited computing resources.
Subject Classification:
Keywords: