API linking in stack overflow
Ang, Wei Loon
Date of Issue2017-04-12
School of Computer Science and Engineering
Singapore Membrane Technology Centre
Stack Overflow is one of the most popular and active online forum with over 100 million views every single month. Stack Overflow provides up-to-date answers to almost any API query of any library and verified by reputable Stack Overflow members themselves. This makes Stack Overflow an extremely reliable source to clarify doubts and find answers. However, within the context of Stack Overflow, there exists many coding terminologies which may belong to multiple classes or libraries. Most these terms do not have any hyperlink to its official API documentation. This in turn leads to context ambiguity making it hard for users to do further reading. In addition to that, manually linking each and every term can be time consuming and tedious. Hence, there is a need to create a way to automatically detect the coding terminologies and link them to the correct API documentation. The challenge however, is that API terms found in natural language texts are rarely fully qualified where simple matching of names may lead to many potential code elements declared by different classes or libraries. This report will describe a previous approach done by the previous FYP student followed by an adaptation of a machine learning approach  which will be used to automatically disambiguate the terms found in Stack Overflow.
Final Year Project (FYP)
Nanyang Technological University