During last weekend I've participated in the Hack4Lem hackathon organized by PKO Bank Polski and Instytut Polska Przyszłości im. Stanisława Lema (Stanislaw Lem's Institute "Polish of the Future"). It lasted 40 hours from June 18th 5PM to June 20th 9AM.
- Deploy working AWS E2 instance
- Use Python and Bash scripting whenever possible
- Don't put 100% effort into the solution
- SwearJar - NLP Processing Ecosystem
- Conclusion
- Additional readings
Because it was my first hackathon ever, I did know that I should focus on more favorable things that winning the prize. So from the two available categories
- Bank of the Future.
- Cybersecurity - Fighting Cyberbullying
I have chosen the latter one and set myself three goals.
Few weeks ago I have created an Amazon Web Services account having the idea of private VPS setup tests, but since then, I didn't do anything on cloud. This was the opportunity for me to push myself a bit and get that experience with AWS Dashboard, SSH access, network hardening and file transfer over rsync.
sudo rsync -e "ssh -i privatekey.pem" -rv {local_folder}/ ubuntu@{IP}:{remote_folder}/
I though that choosing interaction with social media portals would pair nicely with scripting language. Python is my go-to language when I work with hacking challenges and I want to keep better at it. During this hackathon, I wanted to see how I can create Python server and in overall what are the best practiced when writing a standalone application. So far, my experience with Python was some simple request/response actions in a single file script.
I knew I won't join the finalists, neither by delivering the complete solution nor by skipping the night. My idea doesn't feel innovative enough, and I didn't see the point of putting accessibility features by force into it - so it won't fit the constraints of the competition we are challenged with.
What I wanted to have is a solution that will meet my personal goals
SwearJar has a main processing unit that is hosted on the cloud (or other publicly available place) and is exposing its classification powers over the API. Other modules, called Monnies would send the text to it and receive the prediction value - is this an offensive phrase or not.
Now, depending on the Monnie, method of acquisition and action taken are different.
| Monnie | Method of acquisition | Actions taken |
|---|---|---|
| WordPress Doubloon | Comment, before it is submitted | Prevent posting, warn and educate commenter, send to moderation |
| JavaScript Dime | Any defined text field | Prevent submitting, warn and educate user |
| Twitter Penny | Search phrase, profile or any other Twitter query, ex. geolocation | Send a private message or tweet responding to the aggressive tweet |
| YouTube Nickel | Comments under the videos | Send a private message or respond to the aggresive tweet |
This was the main idea - but I know I wouldn't be able to prepare all of it in just 40 hours. So, I focused on the core Python module deployed on AWS (SweatJar) and Twitter Penny Monnie.
Here you can see GitHub repository:
Please don't hesitate to fork and play with it, I don't mind if this idea pays forward.
Did I win something? No.
Actually yes, but only from the Discord giveaways: Lem's book "Niezwyciężony" ("The Invincible") and delivery food service coupons worth of 4 dinners for two.
Am I disappointed? No way! I have completed my goals, I have participated (and survived) my first hackathon, met some insightful people. So, all in all - I'm happy I've subscribed to this (with a little push from my wife).
Are you interested in what it looked from my perspective? See the highlights below my profile:
📌 Follow the
#CyberEthicalhashtag on the social media
👉 Instagram: @cyber.ethical.me
👉 LinkedIn: Kamil Gierach-Pacanek
👉 Twitter: @cyberethical_me
👉 Facebook: @CyberEthicalMe