Sentimeter-Br dictionary covers unigrams (single element/word), bigrams (two adjacent elements) and stopwords (words in a search can be considered irrelevant) Analysis tools of emotional texts based on word lists, are best known as ANEW, OpinionFinder, Senti-WordNet, WordNet and SentiStrength.
Using social networks to analyze sales and features of smartphones or other objects is justifiable because the amount of information is faster to collect, and more data can be gathered. The contribution of this work is building a dictionary with the use of regional slangs, emotions, negative words and different verb tenses.
The polarity of the dictionary was validated by the machine learning technique. The algorithms used in Weka were Bayesian networks (Naive Bayes and Bayes Multinomial), Decision trees (C4.5) and Sequential Minimal Optimization (SMO) These algorithms were used to train the data and to decide if a sentence has a positive, negative, neutral or spam value. We built a Twitter and Facebook search frameworks that can be accessed by mobile phones.
Auto46
https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.671.2331&rep=rep1&type=pdf