Identification multi-level frequent usage patterns from APIs

Abstract

Software developers increasingly rely on application programming interfaces (APIs) of frameworks to increase productivity. An API method is generally used within code snippets along with other methods of the API of interest. When developers invoke API methods in a framework, they often encounter difficulty to determine which methods to call due to the huge number of included methods in that API. Developers usually exploit a source code search tool searching for code snippets that use the API methods of interest. However, the number of returned code snippets is very large which hinders the developer to locate useful ones. Moreover, co-usage relationships between API methods are often not documented. This article presents an approach to identify multi-level frequent usage patterns (IML-FUP) to help developers understand API usage and facilitate the development tasks when they use new APIs. An identified pattern represents a set of API methods that are frequently called together across interfering usage scenarios. In order to investigate the efficiency of the proposed approach, an experimental evaluation is conducted using four APIs and 89 client programs. For all studied APIs, the experimental results show that the proposed approach identifies usage patterns that are always strongly cohesive and highly consistent.