Efficient Searching for Essential API Member Sets based on Inclusion Relation Extraction

Abstract

Search tools for Application Programming Interface ( API ) usage patterns extracted from open source repositories could provide useful information for application developers. Unlike ordinary document retrieval, API member sets obtained by mining are often similar to each other and are mixtures of several unimportant and/or irrelevant elements. Thus, an API member set search tool needs to have the ability to extract an essential part of each API member set and to be equipped with an efficient searching interface. We propose a method to improve the searchability of API member sets by utilizing inclusion graphs among API member sets that are automatically extracted from source code. The proposed method incorporates the frequent pattern mining to obtain inclusion graphs and offers the user a way to search appropriate API member sets smoothly and intuitively by using a GUI. In this paper, we describe the details of our method and the design and implementation of the prototype and discuss the usability of the proposed tool.