Sentiment Analysis using Amazon Product Review Data

You can use InfraNodus text analysis platform to perform sentiment analysis of customer product review data. By default, we offer an automatic import function from the Amazon country-specific websites as well as Audible. 

You can also use your own data and the same approach using our CSV import app. You would then need to create a CSV spreadsheet where one column contains product reviews and the other column contains the rating or sentiment.

The approach we propose can be used to effectively:

• understand what the customers are talking about in relation to a product
• discover the positive sides of a product / brand
• reveal the negative aspects of a product / brand
• compare the positive and the negative sentiment 
• compare different products
• study competition
• make informed shopping decisions

Step 1: Import Amazon Data

To get access to the Amazon reviews, go to https://infranodus.com/apps and choose the Amazon import app:

Once you open the page, enter the ASIN of the product you'd like to analyze or copy and paste the link to that product directly.

Choose whether you want to categorize your data by the stars in the review or to simply split them into positive (4- and 5-star reviews) and negative (1- to 3-star reviews). We recommend the former for the ease of analysis and the latter for better granularity. 

Choose the name of the graph you want to create and click Import.

What If You Need Data from Other Sources?

If you need data from other sources, you would need to get access to the reviews and the ratings and save them in a CSV spreadsheet (e.g. with Google Sheets). 

One column should be the customer reviews, the other column can represent the rating, the sentiment, or anything that can be used to evaluate the positive / negative review.

You can use other columns for additional information: each column for every particular dimension you want to study. For example, geographical location, sex, etc. 

Your table would look something like this:

Then you simply use our CSV import procedure and select the "Review" column for analysis and the "Rating" or the "Sentiment" columns for categorization. You will then get the same results as in our Amazon app. 

Step 2: Understand the Sentiment

After about a minute, you will see a graph that shows the main keywords and topics contained in the customer reviews for this product and the connections between them. 

In this example, we're using Zoom H5 sound recorder — one of the most popular devices on the market. 

The reviews are on the left-hand side.

The graph shows the main keywords (bigger) and the relations between them, so you can discover the context.

The keywords that tend to occur together will have the same color on the graph and comprise topical clusters, which can be seen in the Analytics panel on the right.

You can click on the certain nodes on the graph to see the review that's talking about a particular topic:

Step 3: Analyzing the Positive and the Negative Sentiment

You can also filter the graph to only the positive or the negative reviews map.

Use the menu at the top to do that. For instance, you may want to see only the negative reviews to see what they're about:

As you can see, the topical clusters also change and we can see that a lot of criticism is about the product being great but the sound quality — not so good. 

...

Read more about sentiment analysis of product reviews using InfraNodus on our website.