A Bisaya language model for a neural-network based sentiment analyzer

Sentiments are insights. It paints a distinct picture of one’s perception of subjects. In Natural Language Processing (NLP), text classification is one of the most useful tasks to gain essential and valuable information through contextual mining of the source material. One predominant text classification application used in most social media analyses is sentiment analysis, a classifier type aimed at digging deep into the text and extracting subjective information to support organizations\u27 understanding of social sentiments. This research proposes a neural-network-based language model for the task of classifying whether the statement expressed a positive or negative polarity. The contributions of this work are the following: (1) collection of sentiment annotated Bisaya news articles, tagged and valuated by Bisaya linguistic experts, (2) word embedding learned from Bisaya text which addresses the lack of comprehensive semantic resources, (3) the Bidirectional Long Short Term Memory (BiLSTM) with Attention, neural network sentiment analyzer trained on the supervised Bisaya dataset, and (4) a Bisaya language model, capable of analyzing text data useful for different NLP applications