Modelling Influence and Reach in Sentiment Analysis

Sentiment analysis emerged as an important research topic for its potential in tapping into the large amount of opinions that became a central piece of social networks, providing companies who have any kind of public feedback with new insights on their products and services. Research in this field focuses mainly on the semantic aspects of sentiment recognition and, in more recent years, on the impact of influence in propagation of opinions. Here we present a model that aggregates both these aspects and focuses on providing a framework for calculating global sentiments. The model is based on current literature and provides information regarding not only polarity of opinions, but also on other aspects that affect their diffusion and prevalence, such as influence, reach, ambiguity and relevance. The global sentiment is the result of a weighted evaluation of these aspects and is used both for estimating current sentiment towards an object and for predicting future trends in sentiment.