Choosing between brands is always a hassle. The decision has to be a calculated one and in this process, lots of factors come in. Public perception or opinion about the brand is one such factor. Several websites show ratings and reviews related to brands. These ratings are, in many cases, biased or skewed. On the other hand, social media has become a hotspot of raw and honest opinions from circles vast and diverse, due to its open and accessible nature. Even with some form of censorship, people are more empathetic towards social media reviews than actual review websites. With one such social media giant, Twitter, as our data source, we present a solution to this dilemma – a brand comparison system, made using sentiment analysis in Python, that uses tweets to compare two (or more) brands, give us insights related to them and make a calculated decision as to which one is better and by how much using an overall rating system.
Keywords: sentiment analysis; emotion analysis; opinion mining; Twitter; Python; Anvil; Colab; machine learning; natural language processing; NLP; digital marketing; brand image; perception; rating; vader; nltk-vader; textblob; flair.
The server needs to be run on Google Colaboratory (or download the module and run on any Python environment, connected to the web). Once it is up and running, we can open up the Anvil client and enter two brands and other details for their comparison.
Steps for running on Google Colab:
- Open the given Colab link
- At the top right, make sure the Colab instance is connected to Google App Engine
- Go to Runtime > Run all
- When the last cell is running, launch the Anvil app
- As long as the server is running on cloud, the app will work
Steps for running on local machine:
- Download the server module (or ipynb from Colab)
- Install all the necessary libraries
- Run the file on local Python environment (or Jupyter)
- Launch the Anvil app and compare!