Big Data

New Big Data Sentiment Analysis Show Potential Biden Election Landslide

Pinterest LinkedIn Tumblr

LUX Election 2020, a big data analysis program running on a supercomputer, shows a growing divide in voter sentiment, with Democrat Joe Biden gathering positive sentiment, even among Republicans. Meanwhile, Biden’s opponent, President Donald Trump, is showing significant negative sentiment among voters. The difference in voter sentiment is significant because while a strong negative sentiment may not result in a large shift, it can be a strong indicator that some voters will change sides while others simply don’t vote.

The LUX analysis software uses real-time data from several social media outlets to indicate what voters may really be thinking rather than what they tell pollsters when they call. During the 2016 presidential election, an earlier version of the software, LUX2016, began showing surprising strength for Donald Trump’s candidacy in the days leading up to the election – something the polls didn’t show.

A line graph showing a significant difference between sentiment in favor of Joe Biden and much lower numbers for Donad Trump.

LUX Election 2020 is provided by ICG solutions of Potomac, Maryland. ICG provides data analysis services to the intelligence community as well as to the federal government and industry using real-time streaming data. The election analysis software is available as a service to anyone who wants to use it.

Broad Demographics

The LUX platform allows users to look for a wide variety of data about voter sentiments, including issues such as healthcare, gun rights or race relations. It can sort responses by a wide array of demographic choices. Perhaps most intriguing is that you can watch changes in sentiment in near real time by selecting analyses over a 15 minute period. Of course longer periods of time are probably more useful in determining how voters might act.

Normally, finding voter sentiment in September for an election in November might be interesting for tracking trends, but in the 2020 election, voting is already well underway. Changes in sentiment in September can affect votes directly.

In addition to tracking voter sentiment, the LUX system can track other factors impacting the election, although those data aren’t part of the public website. According to David Waldrop, CEO of ICG Solutions, LUX is also tracking disinformation campaigns and linking them to non-authentic social media users. Those users in turn are being linked to their own interactions to discover networks of disinformation.

“We can use rules to see the age of social media accounts,” Waldrop said. He said LUX uses that information to determine an influence score that looks at reach of a person, their  resonance, meaning when they talk who cares, and recency and frequency. Those are influence scores, which are then paired with a bias score.

“We’re looking for high influence scores and bias score, but in new accounts,” Waldrop said. He said that seeing a Twitter users, for example, that’s only been on the service for a couple of months working hard to build up their influence and also showing strong bias probably indicates an account that’s not authentic.

“We’re now watching explicitly what this group of people is saying to see what words they’re using and any new memes,” Waldrop said. He said that this information can be used to see what the users and others are trying to influence and how.

Waldrop said that he has seen that some group, perhaps the Russians, has created thousands of fake users that appear intended to specifically identify with certain demographic groups, such as suburban women or African American men, apparently with the idea that if their fake users appear to be part of their group they’ll have more influence. He said that they’ve used LUX to identify about 8000 suspect accounts that may be intended to have a role in spreading disinformation.

While sentiment analysis, even using the vast amount of data produced in real-time by social media services, can’t predict the outcome of an election, it can show current trends. Those trends, in turn, can indicate what people are thinking, and those thoughts are what causes them to vote the way they do.