Enabling reproducible sentiment analysis: A hybrid domain-portable framework for sentiment classification

Abstract

In this paper a hybrid framework for Sentiment Analysis is presented. In the first part, dictionary based and machine learning based Sentiment Classification are introduced and the two approaches are contrasted. In the second part of the paper, the HSentiR framework, which combines the two approaches, is introduced. Consequently, the framework is evaluated regarding scoring accuracy and practical concerns.