Results from "Binary Reduction of Dependency Graphs"

<p>The raw data of the reduction of the&nbsp;238 bugs as extracted from the runs of the four algorithms ddmin, verify, closure, binary. An extra column also exists that describes the run of Binary Reduction on the list of classes directly.</p> <p>The columns in `deliverable.csv` are predicate (the name of the decompiler), name (the name of the project we ran on), size (number of classes), scc&nbsp;(number of strongly connected components). Then for each tool we have&nbsp;size, scc&nbsp;(final size and scc), iters (iterations to last success), total-iters (iterations before finishing),&nbsp;time (time to last success (s)), total-time (time before finishing), timeout (did the predicate timeout, or succeed), check (did the bug still exist after reduction)</p> <p>The `benchmarks.csv` covers basic statistics about the programs used in the results. Lib is the number of library classes, LOC is lines of source code in the program, classes are the number of classes, edges are the edges in the dependency graph, degree is the average in and out degree in the graph, scc&nbsp;is the number of strongly connected components, out_degree and in_degree is the median in and out degree in the graph.</p>