SecureShopping

Inspiration

We were inspired to do this project by the observation that older adults are more susceptible to scams, including phone calls, emails, and other tactics used by scammers to make money. Therefore, with SecureShopping, we aim to help older adults with one aspect of the internet: online shopping.

What it does

SecureShopping takes in an Amazon link pasted by the user. Info such as product name, description, and customer reviews is then scraped and formatted in an easy way for the user to read and understand. Then, according to reviews, it states whether the product is safe to purchase or possibly a scam, giving the user information that allows them to proceed with caution.

How we built it

Used existing embedding models from hugging face to parse down reviews and then used k-means clustering to find similar talking points for the review to focus on. Used claude code to make a front end.

Challenges we ran into

One challenge that we ran into was the time aspect of running our code. Since Amazon has some systems in place that prevent scripts from being used on its websites, when scraping the data, delays have to be implemented in order for Amazon not to detect it as a script and give the program a Captcha or an IP block. Another focus is that we did not solve the problem of scraping Amazon's data.

Accomplishments that we're proud of

uhhh

What we learned

I learned how to use hugging face. Got better at pandas

What's next for SecureShopping

While SecureShopping is a prototype design, there are various ways to improve the design. First, implementing a better way to comb through customer reviews, as the current methods have some limitations. This can be done by utilizing a better clustering algorithm and adding bot filtering for both message content and contextual data. Another improvement would be implementing RAG, which would allow users to ask more questions about the products and review sets. A third improvement would be adding more accessibility features, such as text-to-speech.

Built With

• claudecode
• python