robertfeldt/sept repository overview

The main current use case is for doing simple double blind checking/screening. The tool has been used for this in the conferences ICST2018, ICSE2018, ICSA2018, and ASE19. Use at your own risk and remember that you must also check papers manually; sept only has a few, simple heuristics to find the most common/obvious DBR violations.

(Basic) Instructions for use:

1. Install docker, see docker.com

2. Install the sept tool: docker pull robertfeldt/sept

3. To run the tool on a directory of pdf files: docker run -it -v "$PWD":/data robertfeldt/sept:latest julia /usr/src/sept/bin/sept.jl -s icse18 --subdirmove This will create one file per pdf analysed with a file extension of .analysis instead of .pdf. These analysis files are saved in two subdirs named "PASSED" (for pdf files where no DBR violation was found) and "FAILED" (for pdf files where there is something that needs manual check).

After running sept as above you should check all the .analysis files in the FAILED subdir. If it has found any suspicious pieces of text in the paper it will mark the matching parts and give the text/context around the match. Note that especially email addresses people can often blind in many, many different ways so they are often among the false positive matches. Remember that papers that are PASSED by sept may still contain double-blind violations; there are almost an endless number of ways that people can violate the rules and we will not be able to find them all until we have very strong AIs. ;)