Grubsy

Inspiration

Perplexity AI

What it does

The user goes to a url of a restaurant, they could be a Owner or customer, they get presented with a sentiment analysis done by our LLM based on the reviews of that place. It shows the benefits, avg rating, etc.. If you are an owner it shows what you need to work on and how to improve on it.

How we built it

We used a webscraper to gather information on the reviews of a particular place, and fed it to an LLM and sent it to an api so the front end receives the data. We used flutter for the front end to present it to the user.

Challenges we ran into

Scraping reviews from multiple sources, Analyzing text data to determine sentiment can be complex due to language nuances.

Accomplishments that we're proud of

Seamless Review Aggregation: Successfully integrated data from multiple platforms, ensuring accurate and consistent reviews across different sources.

Real-Time Ranking: Developed an efficient real-time ranking algorithm that factors in quality, speed, and satisfaction, providing users with up-to-date recommendations.

Advanced Sentiment Analysis: Leveraged powerful NLP models to analyze customer sentiments, allowing us to extract meaningful insights from complex, unstructured data.

User-Centric Design: Delivered a simple, intuitive user interface that makes it easy for food seekers to find the best restaurants near them without overwhelming them with information.

Scalability and Performance: Optimized the app to handle large volumes of data and users, ensuring a smooth experience even during peak usage times.

Transparent and Ethical Practices: Built a platform that operates transparently and ethically, respecting user data and complying with platform terms of service

What we learned

Data Handling is Critical: Collecting, cleaning, and standardizing data from diverse sources is more complex than anticipated. Effective data processing pipelines are essential to ensure accuracy and consistency.

NLP Challenges: Sentiment analysis and understanding user reviews at scale require advanced NLP techniques. We learned the importance of choosing the right models and fine-tuning them for our specific domain.

Algorithm Optimization: Developing efficient algorithms for real-time ranking involved careful consideration of performance, accuracy, and scalability. It's crucial to balance speed with data quality.

User Feedback is Vital: Creating an intuitive UI/UX requires constant iteration based on user feedback. Our early designs evolved significantly to meet user expectations and improve engagement.