A Supervised Approach for Sentiment Analysis using Skipgrams
Javi Fernández, Jose Manuel Gomez, Patricio Martínez-Barco · 2014
We present a supervised hybrid approach for Sentiment Analysis in Twitter.A sentiment lexicon is built from a dataset, where each tweet is labelled with its overall polarity.In this work, skipgrams are used as information units (in addition to words and n-grams) to enrich the sentiment lexicon with combinations of words that are not adjacent in the text.This lexicon is employed in conjunction with machine learning techniques to create a polarity classifier.The evaluation was carried out against different datasets in English and Spanish, showing an improvement with the usage of skipgrams.