Scientists have developed an online tool 'Tweetcast' that can predict whether citizens will vote for Donald Trump or Hillary Clinton in the upcoming US Presidential elections. <br/><br/>TweetCast uses a machine-learning algorithm to examine words, hashtags, tagged usernames and mentioned websites to uncover which terms are most predictive of voting preference. The tool, developed by researchers of Northwestern University in the US, can also predict which party will dominate a particular state in the November 8 elections.<br/> <br/>As per the algorithm, tweeting the words "lying," "liberal," and "illegal" indicates a vote for Republican candidate Trump.On the other hand, using the words "single," "humanity," and "rights", predicts a vote for Democratic rival Clinton.<br/><br/>The algorithm was trained on Twitter users who have publicly declared support for one of the two candidates. Researchers said that 'TweetCast' has a prediction accuracy of over 80 per cent.