review.ai

Inspiration

All of us have participated in DECA case study competitions in the past, and we are aware of the impact that customer satisfaction has on the success of a business. Therefore, after learning of the theme of UoftHacks IX, we saw how we can use customer reviews to further grow revenue for businesses. Examples of customer reviews include YouTube reviews and tweets on Twitter. Rather than having to sift through all potential customer reviews, we created an application that can automatically determine the satisfaction level of a review.

What it does

The program is a map-user interface that allows users to submit reviews through audio clips, text, and more. From this, the backend straps the database and frontend together to use a machine learning model to determine if a certain review is positive or negative. This certain value is then displayed on the map corresponding to the location of the business.

How we built it

review.ai was built primarily using Visual Studio Code, in the javascript language (with some python). Frameworks that were used include react, node.js, and MongoDB. Additionally, we used Google Maps API and Speech-to-Text API.

Challenges we ran into

The module was not downloaded properly, but after countless hours of troubleshooting (as well as going through Stack Overflow Blogs), we managed to debug the code to help our application run smoother.

Accomplishments that we're proud of

We are proud of the fact that we were able to include Google Maps API and Speech-to-Text API, as these allowed our program to become more interactable and accessible. Furthermore, using these APIs allowed us to learn more about machine learning algorithms (e.g. decision trees, naive bayes, and SVM).

What we learned

We learned how to integrate Google Maps API and Speech-to-Text API into an application. Additionally, we learned how to train test models to perform the certain function that we want.

What's next for review.ai

We hope to expand the machine learning capabilities of review.ai to allow the filtering of more nuanced reviews of the business and be able to further separate the categories of customer satisfaction using keywords. Also, we want to investigate the capabilities of using deep learning for our program, which is a term we learned during the research of this project.