Aspect Based Sentiment Analysis

Jan 12, 2018 | 3234 Views

ABSA is the analysis of a given statement, paragraph, or a huge document for getting insight about what the text or document is talking about.

By seeing the list of all nouns in a sentence or a paragraph, we can get an idea about the document or particular paragraph. If we can get the adjective that describes each noun, we will understand that paragraph more.

If we can run a sentimental analysis on each short sentence in that paragraph, we will get more information. Please don't think we are talking about a text-summarizer. It's more than that.

Suppose we are launching a new laptop. After the product launch, we are monitoring customer feedback. Because we have lots of customers across the globe, we will get lots of feedback.

If we want to analyze all of that feedback, it may take a year or several months. In short, it is impossible now, because the industry is very competitive and growing too fast.

ABSA will be very helpful in this situation. It will extract all of the nouns and adjectives and do a sentimental analysis on each sentence that has a noun.

For example, if the sentence is "Monitor of the lap is good but the battery backup is worse," it contains two nouns, monitor and battery, and each noun also has an adjective.

For monitor, its good and for battery, its worse. Monitor has a positive sentiment and battery has a negative sentiment. If we run our program on a set of feedback, we will be able to find out what features users don't like and the features they do like. We can use that information to redesign the next version of our product.

It is that simple. You can find a sample code for doing ABSA on my GitHub repository.

Source: Codementor