A Deep Learning Sentiment Analyser for Social Media Comments in Low-Resource Languages

Abstract

During the pandemic, when people needed to physically distance, social media platforms have been one of the outlets where people expressed their opinions, thoughts, sentiments, and emotions regarding the pandemic situation. The core object of this research study is the sentiment analysis of peoples’ opinions expressed on Facebook regarding the current pandemic situation in low-resource languages. To do this, we have created a large-scale dataset comprising of 10,742 manually classified comments in the Albanian language. Furthermore, in this paper we report our efforts on the design and development of a sentiment analyser that relies on deep learning. As a result, we report the experimental findings obtained from our proposed sentiment analyser using various classifier models with static and contextualized word embeddings, that is, fastText and BERT, trained and validated on our collected and curated dataset. Specifically, the findings reveal that combining the BiLSTM with an attention mechanism achieved the highest performance on our sentiment analysis task, with an F1 score of 72.09%.

Human-annotated dataset for social media sentiment analysis for Albanian language

“…The names of all attributes of the dataset and their respective descriptions are presented in Table 1 . The data discussed in this article are related to the research article entitled “A Deep Learning Sentiment Analyser for Social Media Comments in Low-Resource Languages” [16] . The dataset and its supplementary files are hosted in the Mendeley Data repository [17] .…”

Natural language processing for Albanian: a state-of-the-art survey

“…Kastrati et al [15] presented the sentiment analysis of people's opinions expressed on Facebook with regard to the pandemic situation in the Albanian language. Here the authors developed a deep learningbased model to classify people's opinions as neutral, negative, or positive.…”

New Education Policy 2020: A Sentiment Classification

“…To analyse the sentiments of people expressed in social media regarding COVID-19 implemented an attention mechanism, BLSTM. The results revealed that the proposed mechanism performed well and achieved the highest f1-score of 72.09% (7) . The model suggests a better way to pad input sequences.…”