Eyesover Technologies measures public opinion by continuously analyzing mentions of candidates or parties from online media platforms (Twitter, Facebook, Reddit, and various online news sites).

Each mention is analyzed for sentiment, which provides a measurement of how positively or negatively the topics or subjects are discussed and provides a strong indication of the author’s support of a topic or subject.

A mention is defined as an original post, comment, or tweet. Retweets or likes are not considered mentions. Multiple mentions about a candidate originating from the same account in a 24-hour period will be aggregated to count as one.

We use the self-reported location in an account owner’s online profile to determine if a mention has been published by an eligible voter in the jurisdiction in question. Mentions from individuals with no location information in their online bio are disregarded.

Based on its mentions, an account is analyzed to determine the candidate the account owner is most likely to support. The number of accounts supporting each party or candidate is totaled, and the support score represents each candidate’s share of the total number of accounts the system has identified to be supporting one of the candidates.

The system’s methodology is largely dependent on sentiment and the volume of mentions. Due to our real-time analysis, candidate mentions are prone to significant fluctuations, which will be immediately reflected in daily support scores. Because of these fluctuations, we use a four-day rolling average support score.

Our system uses a number of bot and fake account detection processes to minimize the number of these accounts that are considered in our scoring.