Inspiration
Our inspiration for this project came from the realization that customers often spend hours searching through countless reviews before making a purchase. We wanted to create a tool that could automate this process and present the most useful information in an easy-to-understand format. We researched various machine-learning techniques and settled on using natural language processing algorithms to analyze customer reviews.
What it does
Our ML app uses natural language processing to analyze e-commerce reviews for most positive/negative comments. Our Python web-scraper gathers all the customer reviews in one place to which we apply sentiment analysis.
How we built it
To start, we utilized the Beautiful Soup web scraping library to extract customer reviews from e-commerce websites such as Amazon and Best Buy. We then learned how to interact with the Cohere API, which provided a powerful and easy-to-use sentiment analysis tool.
We served our backend API endpoint with FastAPI, a Python framework, because of its low latency and ease of use.
Our team opted to use React for the user interface due to its ease of use and powerful capabilities. We developed a simple and intuitive interface that allows users to enter a product name and view top words, average rating, the most positive and negative review, and so on.
Challenges we ran into
One of the main challenges we faced was ensuring the accuracy of our sentiment analysis algorithm. We spent a significant amount of time fine-tuning our algorithm and testing it on a variety of different reviews. Another challenge was ensuring that our web scraping was robust enough to handle changes in website layouts and HTML structure.
Overall, building our machine-learning application was a challenging but rewarding experience. We're proud of the final product we've created and are excited to see how it can help users save time and make more informed purchase decisions.
Accomplishments that we're proud of
- Building a full-stack application and integrating our frontend app and backend API smoothly
- Creating a more powerful React app with type checking using TypeScript
- Adding subtle animations and features to elevate the user experience
- Trying out a new backend framework: FastAPI and NLP API: Cohere
- Implementing a robust web scraper that could handle a variety of inputs
What we learned
- How to use Cohere's sentiment analysis API
- How to utilize FastAPI to quickly develop and deploy web applications
- Effective methods for debugging TypeScript and frontend errors
- Applications of natural language processing and machine learning
What's next for Savvant
Looking to the future, we aim to continue improving the accuracy and efficiency of our sentiment analysis algorithms, as well as expanding the scope of our web-scraping capabilities to cover a wider range of e-commerce websites.
In addition, we plan to incorporate more advanced natural language processing techniques, such as topic modelling and entity recognition, to provide users with even more valuable insights from customer reviews.
We also aim to explore the possibility of integrating our app with popular e-commerce platforms, enabling users to access our insights directly from their favourite online stores. This will further streamline the product research process, making it even easier and more convenient for users to make informed purchase decisions.
Overall, the future of our machine-learning application is bright, and we're excited to continue pushing the boundaries of what's possible with this cutting-edge technology.
Log in or sign up for Devpost to join the conversation.