Techadvisor

With personalized recommendations for tech purchases, shopping for technology has never been so easy

Comment 1

Inspiration

We're sure most people can relate to the fact that deciding which new laptop to buy is a real headache. It's difficult to know exactly what one needs, and even harder to translate those needs into specific features that a laptop should have to fulfill them. As such, many tech illiterate people commonly end up either buying laptops and other pieces of technology that do not fit their needs, or spend significant amounts of time deliberating and doing time-consuming research. As such, we wanted to create something to make this process easier for many people.

What it does

This project can be divided into 2 major components: The Voiceflow API, which acts as the frontend, and the backend written in python.

Our backend python scripts are responsible for scraping tech wholesale sites such as newegg, and also for generating recommendations that are relayed to Voiceflow to be communicated to the user.

The Voiceflow API is used here with Amazon Alexa in order to interact with the user and gain unique information about the user which is passed to the backend to be used to calculate recommendations. Users are currently able to give a price range as well as state a main purpose for the laptop, and then recieve a recommendation.

How we built it

Once we had our idea, we first planned out the logic that our Voiceflow interaction would follow. We then put this together using the Voiceflow API. The backend was created in python and used the beautiful soup python package in order to web scrape. The Voiceflow API was able to output user information onto a google sheet, where we then used google drive API in order to connect the same sheet to our python program in order to create the personalized recommendations. These recommendations were then sent back to Voiceflow to be told to the user.

Challenges we ran into

The biggest challenge we ran into was definitely the challenge of finding a way to gather and create our own dataset of laptops, since there is no readily available dataset of laptops and their information. As such, we concluded that the only feasible way was to web scrape. This led to problems with giving recommendations, as we originally thought that we could just give the top 3 items with the highest rating. However, this isn't really accurate as some products might only have a single 5 star rating and be technically rated higher than highly rated laptops with hundreds of ratings. As such, we had to write an algorithm that considered both factors in giving recommendations. Another challenge we ran into was the fact that there was no apparent easy way to link Voiceflow and Python together. Because of this, we ultimately decided to use the google drive API as a way to connect the two.

Accomplishments that we're proud of

We're definitely the most proud of the fact that we were able to take on the seemingly difficult and tedious task of web scraping and come out with not only a deeper understanding of the technologies behind web scraping, but also a deeper understanding of how sites themselves are constructed. We're also proud of the fact that we took the risk of learning how to do an entire new thing - none of us have really worked with websites or any voice - related APIs and it was a really fun experience.

What we learned

We learned a lot about not just web scraping, but websites themselves. We also learned to use the Voiceflow API, which was very intuitive and a pleasure to use. We also picked up some valuable skills in thinking on the spot, as there were a solid two hours when we thought that our project was suddenly made infeasible when we realized that there would be difficulty linking the Voiceflow API and our Python scripts (we're glad we didn't have to scrap everything).

What's next for Techadvisor

We have a lot of friends who are genuine hardware geeks, so we'd genuinely like to improve Techadvisor and see how it matches up to how they would recommend laptops and computers to their friends and family. We'd also possibly like to further Techadvisor to not just work on laptops but other pieces of hardware such as phones or even custom PCs.

Share this project:

Updates