Inspiration
We've all received fake texts from USPS, Amazon, and Facebook. Some of us have even fallen for smishing scams and input our personal information, only recognizing our error after it was too late. Having a tool to categorize smishing messages as spam would help people worldwide protect their privacy.
Our results
- Qualitative analysis of how local users interact with smishing messages
- Program to determine whether a each message in dataset is malignant (smishing) or benign (scam or legit)
Challenges we ran into
- Due to our time restraints, we were limited in how much data we could collect. Given more time, we could've acquired more survey results, resulting in a more representative sample
- We were limited to 500 lookups/day by VirusTotal's free account. If we were to actually build a tool to analyze URLs, we would have to upgrade to a premium account.
What we learned
- Learned about quantitative data analysis, qualitative analysis, user studies, etc
- Learned how to employ new tools - API requests, python packages, Google CoLab
What's next for When You Smish Upon a Star
- Train a machine learning algorithm to identify messages as smishing, scam, or legit
- Qualitative analysis across age groups. Answer questions such as:
- Are certain age groups more targeted?
- Do different age groups receive smishing messages at different times, or with different content?
Built With
- colab
- python
- smishtank
- virustotal
Log in or sign up for Devpost to join the conversation.