Big Data and Social Media Sentiment Analysis

Big data

"We live in an age of unprecedented amounts of information, generated in every sector, including business, government and health care, delivered at high speed and available in a wide variety of forms and formats. This data may come from many different sources, such as social media posts, digital pictures and videos, cell phone GPS, purchase transaction records and signals sensors used to gather climate information. This information - high volume, diverse and fast - is what is called Big Data."

Luciana Dalla Valle

Social media are amongst the most prolific generators of big data and allow billions of people all around the world to daily interact, post and share contents and give spontaneous feedback on specific topics. As opposed to traditional media such as newspapers, books and television, social media is freely accessible, which means everyone can publish content and control how the information is generated and shared. Through social media, people express their opinions and sentiments towards specific topics, products and services. The emergent research field of sentiment analysis allows us to detect, extract, analyse and classify the opinions and sentiments of people, expressed as textual input.

At the University of Plymouth, data scientists are working to develop innovative sentiment analysis techniques to understand complex social phenomena and perform predictions based on sentiments extracted from social media. From Facebook, for example, a range of information is available for each post to public pages, including its message, when it was created, and the number of likes, comments and shares it has received. Dr Julian Stander and Dr Luciana Dalla Valle extracted Facebook data to analyse sentiment scores and voting patterns for the June 2016 EU referendum in the UK. Results shown that there was increasing online activity as the referendum date approached and that the average number of likes and shares of 'leavers' was considerably higher than 'remainers'.

Future research will aim at overcoming one of the main challenges related to sentiment analysis: the quality of information extracted from social media. In particular, new methodologies, able to identify and deal with fake news, will be developed.