Political analysis may once have depended entirely on subjective human evaluation of candidates and their spoken and written words, and on non‐scientific measurements of the electorate. With the increasing availability of data on current and prior political events, it is possible to add meaningful data-driven attributes to political analysis and forecasting. This type of analysis is used by political entities to understand their electorate, and by the electorate to understand and evaluate their political entities. Big Data collected from internet-of-things devices, online social networks, large‐scale surveys, search engine queries, historical events, news archives, and other sources can materially improve the quality of this analysis. This applies to fomenting and forecasting political unrest, automating or assisting in the fact-checking of political news and speech, predicting election outcomes, as recent work on empirically determining tipping points in influencing public opinion has shown.
Marketing companies and election consultants have long used sophisticated polling techniques in order to determine and shape public opinion so that candidates can use their findings to their advantage. In the last decade, however, we have seen well‐known applications of large-scale data analysis in politics, with mixed success. In 2008, President Obama’s campaign effectively monitored and leveraged social media as an important part of his campaign strategy. In 2016, Donald Trump used micro-targeting to identify and influence small groups of voters with tailored content. Meanwhile, many of the data-driven election forecasts fell short in predicting the outcome of the election. Subsequent to the 2016 US election, interest in automated or assisted fact-checking has grown to combat what is perceived as a problem with factually incorrect news and political speech.