Visualisation

We implemented a Flask-based UI with search functionality for Amazon reviews based on product ID, limited to those in the dataset. The product ID can be easily fetched from the Amazon product page URL.

The search page redirects the user to a product landing page which displays a detailed analysis of the aspects and polarity score results from our model. An Ajax request is sent to the database, which returns the results of our models for the searched product ID. We used Flask, jQuery, HTML, and D3.js for development.

Screenshots

The product landing page displays a horizontal bar chart with clusters along with their aggregated polarity scores. The green and red bars denote an overall positive and negative opinion, respectively. We also added a bubble chart with color-coded aspects, where the bubbles are colored according to the cluster and sized according to the frequency of the aspects. Additionally, there is a tool tip that allows the users to click on a bubble and see a review that contains that aspect.