Sentiment Aggregate Functions for Political Opinion Polling using Microblog Streams

摘要

The automatic content analysis of mass media in the social sciences hasbecome necessary and possible with the raise of social media and computationalpower. One particularly promising avenue of research concerns the use ofsentiment analysis in microblog streams. However, one of the main challengesconsists in aggregating sentiment polarity in a timely fashion that can be fedto the prediction method. We investigated a large set of sentiment aggregatefunctions and performed a regression analysis using political opinion polls asgold standard. Our dataset contains nearly 233 000 tweets, classified accordingto their polarity (positive, negative or neutral), regarding the five mainPortuguese political leaders during the Portuguese bailout (2011-2014). Resultsshow that different sentiment aggregate functions exhibit different featureimportance over time while the error keeps almost unchanged.