cSpade + utilities

cSpade is a sequence mining algorithm developed by Mohammed Zaki[1] This is a refactor of the original algorithm aiming at:

• Fix a segmentation error when the data is huge and contraints too relaxed.
• Remove raw pointers and global variables.
• Combine cSpade and its utilities together. Originally, one needs to run makebin, getconf, exttpose before running cSpade
• Remove mmap which is platform specific, clunky and does not improve performance one bit.
• Reroute standard output to separate channels (logging, summary and mined sequences)
• Provide hooks for other languages to wrap. The python-wrapper is here: https://github.com/fzyukio/python-cspade

Code is compatible with Mac/Linux and Windows (requires VC 2015 build tools)

Compile and run

Just type make (works with VC's make.exe on Windows too). The compilation should create 5 files:

• makebin
• getconf
• exttpose
• cspade
• cspade-full

The first 4 files work in the same way as the original code.

cspade-full is an all-in-one solution that calls all of those routines for you. Call:

./cspade-full [arguments]
Mandatory arguments:
    -i --input </path/to/file>
    -s --support <support value>
Optional:
    -u --max-gap <value>
    -m --max-size <value>
    -m --max-len <value>

E.g.
    ./cspade-full -i zaki.txt -s 0.01 
    

cspade-full will print out the list of sequence. To acquire other channels (summary and logger) you will need to modify the code in main.cc. To print out logger, for example, add cout << result.logger;

Test

Run

make runtest
./test

If everything goes right it should say TEST PASSED. The test runs cspade on a pretty big transaction database (2000 lines) using a low support value and compare the output with what was produced by Zaki's unmodified cspade code.

Debug

if you use Clion then it will automatically read CMakeLists.txt and creates 5 targets corresponding to 5 binary files above. You just need to change the working dir and arguments.

Contribute

• PRs welcome & appreciated

Known bugs

• None, but don't set max-size and max-len too big when using a low support value, otherwise it might run out of memory.

Licence

MIT

[1] Mohammed J. Zaki, Sequences Mining in Categorical Domains: Incorporating Constraints. In 9th ACM International Conference on Information and Knowledge Management. Nov 2000.