Project Info

No Scam, Only Ham

This project did not submit a demo video on Devpost.

Inspiration

Considering the rapid growth of technology and the disconnection between many elderly people and their devices, we wanted to use technology for the better and help tackle a prevalent issue that many seniors encounter, for example the "Ore-Ore Sagi" scam in Japan. Our grandparents have fallen victim to scams before, so we felt personally inspired to help senior citizens who are often targeted by scammers for being more trusting and easier to scam.

What it does

The scam detection works by using NLP to identify whether text resembles scam, and it is connected with an interactive quiz with both audio and text in a website.

How we built it

We wanted to build everything onto a website, so we spent a lot of time using HTML and Javascript to create a beautiful website and an interesting quiz. As for the NLP model, we decided to use the BERT Machine Learning Framework due to its efficiency and accuracy when running on ambiguous language in text. With BERT, we worked on collecting a dataset (a combination of personally generated work and work from an existing catalog), preprocessing and tokenizing the data, and designing and testing the model.

Challenges we ran into

We definitely faced a lot of challenges along the way. We decided to dive into a topic and techniques that none of us were extremely familiar with, but we thought that it would be cool. Because of this, we spent a lot of time understanding how frameworks such as BERT work, installing and uninstalling different packages in terminal, and scouring Stack Overflow for ways to debug. We also struggled to find ways to deploy our machine learning model in python to a web app to make them more accessible. While these challenges got infuriating at times, we were able to rise above and create something that we are pretty proud of.

Accomplishments we're proud of

I think our biggest accomplishment was creating a cohesive website that connects the various parts that we have made. As stated above, one of the hardest parts was finding a way to connect our model to a website, and creating that was definitely one of our biggest accomplishments. Beyond that, for the majority of us, this was our first hackathon! Completing something that we all feel proud of is a big accomplishment in it of itself.

What we learned

Learning to integrate different languages across different IDE's and computers with limited experience in a timed environment was stressful, but we enjoyed the learning and collaboration experience. We also learned the importance of time-management.

What's next

Our next steps are including whisper.ai to implement speech-to-text capabilities to increase the impact of our project. We also want to include other languages to make our scam detection more inclusive.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only

CSSIn code
HTMLIn code
JavaScriptIn code
PythonIn code
FlaskClaimed

4 of 5 appear in the indexed code. 1 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

36 KB

Source files

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.