So, you've probably heard about the US Presidential Election Campaign....
Election campaigns in general, and especially the US Presidential Campaign, generate gigantic datasets which can be used by scientist and engineers globally. Just imagine all the tweets, Facebook posts, election statistics, new articles and blogs, advertising, etc being generated over the entire span of the campaign. They all provide a rich and valuable perspective in the lives of millions of people and can be used by companies and governments for a number of tasks such as generate more targeted and appealing adverts for you, or predicting how voters may react to certain events or news.
Here, we use this data to give voters and followers of the election campaign a view on how the Twitter community views the two remaining candidates: Hillary Clinton and Donald Trump
Twitter is one of the few social media sites which empower users to speak their mind in a concise and straightforward manner. (sometimes blunt, and sometimes well versed). Twitter provides live updates on the lives and views of both candidates and the latest news regarding their campaigns. Unlike blogs or Facebook posts, Twitter users can create tweets (sometimes referred to as micro blogs) much quicker and to a much larger audience, thus creating a faster way to spread their content. Twitter also allows the public to freely interact with their favourite Twitter users (such as celebrities and athletes) in a quick informal manner, making it easier for the "famous" to interact with their followers.
There are two key components in the back end that allow this website to generate its data: Spark Streaming and IBM Watson
Spark Streaming allows us to grab live tweets from Twitter and perform processing of data which includes deciphering which tweets have which candidate as the main subject, and creating the aggregation of data. Spark Streaming allows us to save the data for future use, and streaming the data into our front end.
IBM Watson's Tone Analyzer allows us to determine the tone and sentiment behind every tweet. Once Spark Streaming has cleaned tweets, we can then run the tone analyzer which provides 13 different tone attributes of the given tweet. These tones to determine if a tweet was in support or against a candidate. The 13 tones are broken down into 3 categories: Emotional Tones, which refer to the emotions and feeling in the language of the text; Social Tones, which measures social tendencies in the text; and Language Tones, which describe the writing style of text.
All of the data is stored in accessible formats which can become available upon request. Possible uses could be for academic research and market research. Please contact me directly.
Hi! Hope you like the website :)
My name is Saif Charaniya, and I am a Masters in Big Data student as Simon Fraser University in Vancouver, Canada. I enjoy collecting and analyzing datasets and presenting visualizations for public interest.
Following the US Election gave me the idea to build this website. I spent five weeks of my free time (which there is sadly little of) to create the front and back ends which support the site.
If you would like to reach out to me, you can follow me on Twitter: @saif__c or email me at: saifc[at]sfu.ca or visit my personal website at: saifcharaniya.com
I would really appreciate if you could spread the word about the site through your own social media presence. Thanks!