Spotlight: Sentiment Analysis

Ivy Apr 17, 2014 No Comments

How do the various political Manifestos compare with respect to important keywords?

What do twitter users have to say about the new product launch?

How is your after-sales support performing against your competitors?

What is the general sentiment about the election manifestos of various political parties?

These are questions one asks to learn the opinion of people, track keywords, analyse opinion and converting text to numerical ratings.

You may ask, why are opinions important?

• “Opinions” are key influencers of behaviors.
• Perceptions of reality are often conditioned on how others see the world.
• Decisions are often based on opinions of others.
• Trends in people sentiments are important to develop or re-define strategies.

While individuals seek opinions from friends and family, organizations use surveys, user groups, opinion polls, consultants and social media.

From where is opinion or sentiment mined?

• Word-of-mouth on the Web
• Personal experiences and opinions in reviews, forums, blogs, Twitter, micro-blogs, etc
• Comments in articles, reviews, etc.
• Postings at social or professional networking sites, e.g. facebook, LinkedIn.
• Organizational internal data – research, surveys, customer feedback from emails, call centers, etc.
• News and reports, opinions in articles and commentaries

Basic search commands used for analysis:

• ‘sentiment’ to predict user sentiment
• ‘language’ to predict what language an event is written in
• ‘tokens’ to make it easier to tokenize terms and phrases for analyzing text
• ‘heat’ to measure how emotionally charged a text is

So how would you define sentiment analysis?

Sentiment analysis is the computational study of opinions, sentiments, evaluations, attitudes, appraisal, views, emotions, and any other content with subjectivity expressed in text form. It looks further the traditional Social KPIs (number of mentions, followers, etc.) to process language and text contained within social data;  reviews, blogs, online discussions, news, comments, feedback, or other documents; using variables such as context, tone and emotion,

How does Sentiment Analysis work?

Sentiment analysis applies certain pre-defined methods to classify the polarity of a post, comment or statement, analysing whether sentiment around a topic is positive, negative or neutral. Proprietary tools are used or in-house models of social media monitoring software are developed – using statistics and natural language processing (NLP) techniques to automate sentiment analysis on vast amounts of social data.

The predictive character of sentiment analysis

As social data is unsolicited and user generated, analysing sentiments in online conversations with respect to specific topics help predicting outcomes, whether a political event, marketing campaign, product launch, etc. For example, sentiment reports pulled from Twitter around the Indian elections helps identify sentiments of the electorate and predict results.While social conversations in digital space generate insights for product positioning.

Application areas of Sentiment Analysis

• Businesses and organizations
• Marketeers – benchmarking of products and services
• Industry and competitor analysis
• Individuals – to facilitate decisions on products or services
• Political candidates – to analyse electorate sentiments for strategy
• Ads placement in the social media , ad mix, etc.
• News Analysis – commentary, predictions, and so on.
• Stock market
• Academia – provide general opinions on studies conducted

Sentiment Analysis and social data enable business decision makers to understand consumer attitudes and behaviors more than ever before. Since social data is unsolicited and user generated, mining this data and categorizing it allows companies to understand intelligence around customers’ feelings towards campaigns, content, or products.

Sentiment analysis = Opinion mining = Sentiment mining  = Subjectivity analysis = Affect analysis = Emotion detection = Opinion spam detection

 

Suggested Reading:

An Introduction to Sentiment Analysis

Sentiment Analysis and Opinion Mining

Sentiment Analysis and Social Media Marketing

More from IVY Blogs:

Social Media Analytics

Text Analytics