SE researchers often mine only a few local repositories. However, there may be too few relevant data points in these repositories to support the mining of desirable patterns, such as ones among API methods of interest. One way to address this problem is to mine Internet-scale software repositories—for example, via a code search engine. Mining can then be applied to the entire open source world or to many software repositories within an organization or across organizations. Further, execution traces collected from even an average-sized program can be very long, and dynamically or statically extracted call graphs can be enormous. Analyzing such large-scale data poses a challenge to existing mining algorithms.