Calling the Election: Sentiment Analysis for Election Campaign Monitoring

Entity Extraction, Intelligence Analysis, Sentiment Analysis, Social Media Analysis

In recent years the explosion of social media has made available an unprecedented amount of real-time data that is of great value when trying to measure public opinion. According to the New York Times, more than one billion election-related tweets were posted on Twitter during the last presidential election, from the first presidential debate until election day.

Befittingly, there has been increasing attention to analyzing social media to monitor election campaigns, gauge political polarization and negative campaigning, and even forecast election results. When combined with traditional off-line election polling, interested parties can produce a more accurate picture of the electorate at any point in time.  According to TechCrunch, social media did a better job at predicting the 2016 election outcome than traditional polls alone.

What issues are getting more attention from the public? What candidates are garnering more negative (and positive) sentiment? How are people reacting to events during the election campaign in real time? What are swing state voters saying? These are all critical questions that can inform not only the campaigns themselves, but the public at large.

Sentiment Analysis, an increasingly popular AI-based Text Analytics technology, can help answer these questions. Sentiment Analysis can automatically detect sentiment in large text corpora. It identifies likes, dislikes, opinions, and intent in text.

NetOwl’s Sentiment Analysis for Election Campaign Monitoring

There are a number of ways in which NetOwl provides superior Sentiment Analysis for election campaign monitoring and forecasting:

  1. NetOwl’s Sentiment Analysis goes beyond positive and negative sentiments. Your average sentiment analysis software may detect negative sentiment around a political leader or policy but may not be able to determine what specific aspects the negative sentiment is related to. Through its Entity- and Aspect-based sentiment analysis, NetOwl is able to detect the objects of sentiments (e.g., a politician, a political party, a policy) and the specific aspects of those objects (e.g., a politician’s character, a political party’s stance on an issue) that the sentiments are about.
  2. NetOwl provides fine-grained sentiment detection distinguishing for instance complaints, threats to boycott, and dissuasion among others.
  3. NetOwl is also fast, scalable, and consistent making it possible to process massive amounts of data in real time without getting “tired” and making different judgements for the same text.
  4. NetOwl’s intelligent normalization of identified sentiments and target entities enables data aggregation and quantification to provide the overall view over a large collection of content. With NetOwl’s dashboard, sentiment data can be sliced and diced as desired, for instance by candidates or themes to show a deeper analysis of voters’ emotions and intent. For any entity of interest (e.g., a presidential candidate), the user can see the breakdown of public opinion, the top positive and negative sentiments, the specific aspects that are the object of those sentiments (e.g., a politician’s character, his/her track record), and, if desired, drill down to the source text for inspection and further analysis. Other useful charts show sentiment evolution over time. Using location metadata and automated geocoding of mentioned place entities, sentiment can be plotted on a map to show, for instance, hot spots of support or disapproval.

NetOwl’s Sentiment Analysis is best suited for public opinion mining and election campaign monitoring to gauge candidate support and opposition as well as citizens’ reaction to issues as they arise during an election campaign helping to provide a more accurate representation of the state of the race.