[code] [blog]

LogClusteringSeptember 14 2015 02:55:02

LogClustering, for lack of a better name, was a project I developed while writing my master's thesis on clustering and invariant detection. It uses the popular invariant detection framework Daikon to generate invariants and MuJava to generate mutants to test the strength of these invariants. This framework allows fully automated generation of test cases, and fully automated testing of clustering algorithms for dtrace files.

The project contains a variety of wrappers around the popular projects Daikon, MuJava, and the java compiler. It contains code for efficiently clustering large sets of daikon log files, and evaluating the results.

Probably more useful to the average Daikon user however, is the fact that it contains a graphical user interface for running Daikon and chicory. It also contains a graphical editor for dtrace files. Below you can find the source code, it is provided in the hope that someone might find it useful, or that someone might continue my research.

File logclustering.tar.xz
Size 30M
Sha512sum 3a31e1a82114defc59202928792dd95bb1356dcf1af88d2d5a89e16ba9456d4b78352