Cypher Fraud

Inspiration

Our team was motivated to alleviate the suffering of scams and fraud incidents across the globe. Each of us has had their own run-ins with phishing behavior and fraudulent purchases, and we've decided to take matters into our own hands to pursue this injustice.

What it does

Our app records transactions and assigns a risk level according to different triggers. The information is displayed in a data model that details all relevant transaction information, along with any flags that were raised.

How we built it

Our team started off by drawing up pseudocode to support our foundation. We then continued with the backend logic, starting with a model to represent the transactions and accounts in our code, followed by the UI and rulesets that dictated the flag triggers of fraudulent transactions, and then our team wrapped up by establishing an interactable frontend to read in files of bank transaction information.

Challenges we ran into

Our team found difficulty in determining the code logic for the "Exceeding Purchase Limit" ruleset, as we found that differing expenditure among users and context surrounding purchases. We also encountered troubles creating a frontend application for our code since our team were mostly experienced in backend development.

Accomplishments that we're proud of

Our team is proud of the construction of our frontend model, especially since it was the first frontend development that our team has ever created. None of our team was familiar with standard frontend procedures, and we persevered throughout the night by sheer grit and determination.

What we learned

The accomplishments we made have solidified our foundations in Java backend development and even granted us with newfound knowledge on frontend data modeling and data scraping.

What's next for Cypher Fraud Detection

Plans are in motion to establish more rulesets to catch fraudulent activities, along with ideas about expanding our data model implementation. We want to diversify our model to account for contextual purchases, and we want to integrate a machine learning model to differentiate those categorical transactions.