Sentiment Analysis Introduction : part 1

Sentiment analysis has been using as a tool to cassify response from your user/customer as 'positive' or 'negative' or even 'neutral'. It combines two different disciplines : Natural Language Processing and Text Analysis to extract information from text data. In sentiment analysis, algorithm will learn from labelled example data and predict the label of new /unseen data points. This approach is called supervised learning, as we will train our model with corpus of labelled news.

Why sentiment analysis?

People have different ways to express their attitudes or opinions / reviews towards your product /event / movie or even for people. These user reviews have potential to build brand authenticity between customers and even to establish trust in product.Most of the companies are trying to evaluate the brand value of a product based on customer reviews. In this process, company can extract customer behaviors like : which products are popular among users, what percentage of your customer dislike your brand, how many users have postive sentiment on their competitors product? These questions ultimately help companies to focus on different aspects of product like : product revamp , new marketing strategy and even for social media outreach plan.

Before use of sentiment analysis algorithms, people were trying to evaluate those user response based on simple practices like : extracting keywords from the content and restricting their findings only on 'what people are talking about?'. This will never helps you out to answer few important questions 'What people are feeling and thinking about your product?'. That means only the explicit statements/reviews or opinions will have measurable output with your old approach. What would you do with large number of implicit comments?

Even though, it is extremely arduous to determine the actual tone from given text, we can try to develop a more accurate version of our older one.


Comments powered by Disqus