Aspect Based Sentimental Analysis


We have already seen what is sentimental analysis or if you want to read that again click here. So hope you all are clear about sentimental analysis. So now we can move on to Aspect based sentimental analysis(ABSA). ABSA is the analysis of a given statement, paragraph or a huge document for getting insights about what the text or document is talking about. By seeing the list of all nouns in a sentence or a paragraph we will get an idea about the document or that particular paragraph. And if we can get the adjective which describes each noun, we will get more understanding on that paragraph. Plus 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 and after the product launch we are monitoring the customer feedback. As we have lots of customer across the globe we will get lots of feedbacks. If we want to analyze all of those feedback, it may take an year or several months. In short, it is not possible now as the industry is very competitive and growing too fast. So ABSA will be very helpful in such situations. It will extract all the nouns and adjectives, plus do a sentimental analysis on each sentence which have a noun. 

For example if the sentence is " Monitor of the lap is good but the battery backup is worse ", this sentence contains 2 nouns, monitor and battery and each noun has an adjective  also. For monitor its good and for battery its worse. So monitor have a positive sentiment and battery have a negative sentiment. If we run our program on a set of feedback, we will be able to find out what all features which the users are not liking and the features that users like. Then we can use those informations to redesign the next version of our product. 

It is that simple. A sample code for doing ABSA you can find it on my github repository.

Thank you.. 😊


Popular Posts