Team Ditto was inspired by the movement to support local businesses during the pandemic, especially the movement to support local restaurants. We understood this shift wasn't a one-size-fits-all for restaurants. Many restaurants had to suddenly depend on technology to sell their food when the best way to actually experience it was to dine in person so a customer could truly understand the vibe, energy, and soul of the food. Online reviews, especially in this digital-dependent space, can be difficult to analyze for restaurant owners. Just finding the time to be able to read what your customers are saying about you can be a challenge. That is why our team wanted to create a platform that would help small to medium-sized restaurants gain insights from online food reviews more efficiently using the example from right in Sacramento's back yard.
At its core, Sift helps restaurant owners gather and visualize high-level insights from large amounts of data from online reviews. Sift can do this by utilizing machine learning to analyze reviews and perform sentiment analysis to determine whether the entity is spoken of positively or negatively. We believe, if given the time and resources, this ability could've been scaled to analyze if positive entities were spoken in relation to certain topics that determine a restaurant's success (price, taste, atmosphere, cleanliness, customer service).
We used IBM Cloud’s Natural Language Understanding API to build this model and used a data set provided by Yelp.
Challenges we ran into Our team ran into many challenges, but one of the most prominent being the technical aspects of using a data set so large (many GBs), instead of downloading the data file locally we used an API pull from Kaggle to feed the data into the model. We also ran into time challenges and paywalls of additional API pulls.
We are proud of the time and energy each team member committed to finishing this project.
Each member of Team Ditto learned a bit more about machine learning, design, and hackathons in general that we will take with us into our professional careers.
In the future, it would be intriguing for the team to continue to push the limits of the chosen model as well as developing the actual web platform.
- IBM Watson Natural Language Understanding: https://cloud.google.com/https://www.ibm.com/cloud/watson-natural-language-understanding
- Google Cloud Platform: https://cloud.google.com/
- Adobe XD: https://www.adobe.com/products/xd.html
- Programming Language: Python, JSON, HTML, CSS
- Developers: Isabel Abonitalla & Lam Nguyen
- UI/UX Designer: Taylor Hines & Thuy Le
- UI: https://xd.adobe.com/view/18ab8b1a-c404-4251-82cc-45d4cc97a39b-4df8/?fullscreen
- Presentation: https://docs.google.com/presentation/d/1kryjPRuL4EiRepL9_cQdc1CDhfAl7s3n29HMuPkdCBY/edit
- Dataset (Yelp Review - Partially): http://34.66.185.168/yelp_review_small.json