What is SentiStrength?
SentiStrength is a lexicon-based sentiment analysis tool developed by researchers at the University of Wolverhampton in the UK. It is designed to estimate the strength of positive and negative sentiment in short, informal text - the kind of text commonly found in social web data like comments, reviews, forum posts, tweets, and more.
The algorithm behind SentiStrenth uses a lexicon of sentiment-related words, combined with a set of grammatical rules, to determine sentiment scores ranging from 1 (not positive/negative) to 5 (extremely positive/negative). It provides separate scores for positive sentiment and negative sentiment, allowing you to detect mixed feelings or sarcasm.
A key advantage of SentiStrength is that because it relies on a lexicon and rules, rather than machine learning, it performs much better on the short, informal, and often ungrammatical text found in social web data. It does not require any training data and works out-of-the-box across a variety of textual datasets.
SentiStrength is available as a free Java library that can be integrated into any Java application. There are also demo applications and APIs available to analyze text through a web interface. It has been used for research and analysis in many domains including social media monitoring, market research, reviewing monitoring, and more.