Understanding Your Restaurant Reviews
Imagine having a personal guide through your restaurant reviews – I created a proof of concept does just that. It's like having a friend who decodes customer sentiments, helping you understand exactly how people feel about your restaurant. Transform feedback into a roadmap for success and make your customers even happier with my user-friendly and insightful dashboard.
Problem
Running a restaurant is no small feat, and amidst the daily hustle, owners strive to connect with their customers through reviews. However, they encounter some tough hurdles:
Time Crunch: The demanding nature of restaurant management leaves little breathing room. Squeezing in the time to thoroughly understand customer sentiments from reviews becomes a real challenge.
Analytical Gaps: Not everyone is born with a knack for deciphering the subtle nuances in customer feedback. Some restaurant owners may find it challenging to extract meaningful insights from reviews, affecting their ability to make informed decisions.
Solution
How did I do it?
At a high level, I leveraged Amazon AWS to store, analyze, and visualize the restults.
All the reviews were exported into a CSV and were stored in S3 bucket.
Create a job that will run Amazon Comprehend on your file. This will create a new file of the sentiment analysis, which you will create a database from via Amazon Glue and Amazon Athena.
Once you have your database, leverage Amazon Quicksight to visualize your results
Amazon Comprehend is a natural-language processing (NLP) service that uses machine learning to uncover valuable insights and connections in text.
What did I learn?
This was my first time using Amazon AWS.
I had a blast figuring out how to set up my first AWS bucket and a task that could run on the files inside the bucket. These are the kinds of things developers often talk about when they're explaining how they bring a feature to life. It was really satisfying to piece everything together and understand the process.
Amazon Comprehend returns the sentiment of the text and the score for each of the sentiment
The score represents the likelihood that the sentiment was correctly detected. For example, online 25 below it is 99 percent likely that the text has a Positive sentiment. There is a less than 1 percent likelihood that the text has a Negative sentiment.
Final Thoughts
In the future, I’d like to add the ability to extract and present reoccuring “tags” or ”phrases” that people have mentioned. It may be a good way to present a highlights of what people like or dislike.
In addition, I would love to try to take feedback from our call centre system to understand our customer’s sentiment about the products that I oversee. Alternatively, I can feed Amazon Comprehend from our NPS survey feedback.