Skip to content

Latest commit

 

History

History
69 lines (47 loc) · 4.22 KB

README.md

File metadata and controls

69 lines (47 loc) · 4.22 KB
Raw

SecAlertSeqMining

Set of scripts for simplification of using SPMF library for sequential rule and pattern mining in cybersecurity alerts.

Scripts provide two functions.

  • Generate sequential databases in formats required by SPMF. For this purpose there is src/databases/generate-db.py script.
    • Databases can be generated from alerts in IDEA format or alerts in csv with parameters order described in src/databases/alerts/csv.py module.
    • Because SPMF requires numeric representation of items for majority of algorithms, script generates output file with sequences where items are represented as numbers and also csv mapping between numeric representation of items and string representation of items. The mapping is saved in the same directory with .map suffix.
    • For more information run python src/databases/generate-db.py -h.
  • Process outputs of SPMF library (files with rules and patterns). For this purpose there is src/process-outputs.py script.
    • The main purpose of processing outputs is to replace numeric representation of items with their string representation.
    • For more information run python src/process-outputs.py -h.

Example of usage:

(All scripts are written in python 3.6)

First of all install required packages.

$ pip install -r requirements.txt

Generate sequential database.

$ python src/databases/generate_db.py -i alerts.json -o data/db/ --format basic --db-types src src-port

-i alerts.json Input file with alerts is alerts.json.

-o data/db/ Databases will be saved into data/db/ directory.

We specify format as basic which means, that the generated databases will look as described in src/databases/formats/basic/abstract.py module.

The db-types parameter is specified as src src-port which means that the databases defined in src/databases/formats/basic/src.py and src/databases/formats/basic/src-port.py will be generated.

Run SPMF algorithms.

Download spmf.jar from http://www.philippe-fournier-viger.com/spmf/index.php?link=download.php.

$ java -jar spmf.jar run RuleGrowth data/db/basic/src-port data/outputs/src-port.RuleGrowth 0.001 0.1

Now we run RuleGrowth algorithm for the discovery of sequential rules. Rules are stored in data/outputs/src-port.RuleGrowth file.

Process rules of SPMF.

$ python src/process-outputs.py -d data/db/basic/src-port -o data/outputs/src-port.RuleGrowth

Items in the file data/outputs/src-port.RuleGrowth are now replaced with their string representations from data/db/basic/src-port.map file.

Generate custom database

If you want to create your own database, create a module containing class named Database with a constructor and two methods.

  • d = Database(output_dir, file_suffix) The constructor should be able to take two positional parameters. Output directory (where the database should be saved) and database suffix (add this string to output file name).
  • d.read(alert) Read method will be called for each alert in the input file. Alert is an instance of one of the classes from src/databases/alerts/ package (it depends on the type of the input file).
  • d.save() Save method will be called after processing all alerts with read method.

You can extend AbstractDatabase from src/databases/formats/AbstractDatabase.py to make implementation of some stuff easier for you.

Put the module inside one of src/databases/formats/ packages or create your own. Then call generate-db.py script with --format as the package name and --db-types as your module name.

Acknowledgement

This software package is an attachment to the paper "On the sequential pattern mining in the analysis of cyber security alerts" presented at ARES 2017 conference.

HUSÁK, Martin, Jaroslav KAŠPAR, Elias BOU-HARB a Pavel ČELEDA. On the Sequential Pattern and Rule Mining in the Analysis of Cyber Security Alerts. In Proceedings of the 12th International Conference on Availability, Reliability and Security. Reggio Calabria: ACM, 2017. s. "22:1-22:10", 10 s. ISBN 978-1-4503-5257-4. http://doi.acm.org/10.1145/3098954.3098981