Sentiment Analysis for Polish Using Transfer Learning Approach

Abstract

A method for sentiment polarity assignment for textual content written in Polish using supervised machine learning approach with transfer learning scheme is proposed in the paper. It has been shown that performing simple natural language processing steps prior to classification, provides inspiring results without redundant computation overhead. The documents containing subjective opinions were classified using N-gram and Bi-gram language model that is able to encode some of complex word phrases. The experiments carried out on two real datasets taken from different domains proved that learning on one dataset and testing on another, which is commonly called transfer learning, can be effective and may result in very high classification quality.