COSTER: A Tool for Finding Fully Qualified Names of API Elements in Online Code Snippets

Abstract

Code snippets available on question answering sites (e.g., Stack Overflow) are a great source of information for learning how to use APIs. However, it is difficult to determine which APIs are discussed in those code snippets because they often suffer from declaration ambiguities and missing external references. In this paper, we introduce COSTER, a context-sensitive type solver that can determine the fully qualified names (FQNs) of API elements in those code snippets. The tool uses three different similarity measures to rank potential FQNs of a query API element. Results from our quantitative evaluation and user study demonstrate that the proposed tool can not only recommend FQNs of API elements with great accuracy but can also help developers to reuse online code snippets by suggesting the required import statements.