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

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%.

[1]  Ana Estela Antunes da Silva,et al.  Analyzing the Brazilian Financial Market through Portuguese Sentiment Analysis in Social Media , 2020, Appl. Artif. Intell..

[2]  Zenun Kastrati,et al.  Aspect-Based Opinion Mining of Students' Reviews on Online Courses , 2020, ICCAI.

[3]  A. Hassanien,et al.  Sentiment Analysis of COVID-19 tweets by Deep Learning Classifiers—A study to show how popularity is affecting accuracy in social media , 2020, Applied Soft Computing.

[4]  Quratulain Rajput,et al.  Lexicon-Based Sentiment Analysis of Teachers' Evaluation , 2016, Appl. Comput. Intell. Soft Comput..

[5]  Ali Shariq Imran,et al.  Cross-Cultural Polarity and Emotion Detection Using Sentiment Analysis and Deep Learning on COVID-19 Related Tweets , 2020, IEEE Access.

[6]  Marenglen Biba,et al.  Classification and Clustering of Emotive Microblogs in Albanian: Two User-Oriented Tasks , 2020, Complex Pattern Mining.

[7]  Marenglen Biba,et al.  An Experimental Evaluation of Algorithms for Opinion Mining in Multi-domain Corpus in Albanian , 2018, ISMIS.

[8]  Wladimir Sidorenko,et al.  Sentiment Analysis of German Twitter , 2019, ArXiv.

[9]  Mona Dadoun,et al.  Sentiment Classification Techniques Applied to Swedish Tweets Investigating the Effects of translation on Sentiments from Swedish into English , 2016 .

[10]  David Vilares,et al.  BabelSenticNet: A Commonsense Reasoning Framework for Multilingual Sentiment Analysis , 2018, 2018 IEEE Symposium Series on Computational Intelligence (SSCI).

[11]  B. Duffy,et al.  Health-protective behaviour, social media usage and conspiracy belief during the COVID-19 public health emergency , 2020, Psychological Medicine.

[12]  Yogesh Kumar Meena,et al.  Aspect-Based Sentiment Analysis of Students’ Feedback to Improve Teaching–Learning Process , 2018, Information and Communication Technology for Intelligent Systems.

[13]  Geeta Sikka,et al.  The emergence of social media data and sentiment analysis in election prediction , 2020, Journal of Ambient Intelligence and Humanized Computing.

[14]  J. Rudolph,et al.  Social media for rapid knowledge dissemination: early experience from the COVID‐19 pandemic , 2020, Anaesthesia.

[15]  Sule Yildirim Yayilgan,et al.  The impact of deep learning on document classification using semantically rich representations , 2019, Inf. Process. Manag..

[16]  Igor Mozetic,et al.  Multilingual Twitter Sentiment Classification: The Role of Human Annotators , 2016, PloS one.

[17]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[18]  COVID-19 Public Sentiment Insights and Machine Learning for Tweets Classification , 2020, Inf..

[19]  Erik Cambria,et al.  Sentiment Analysis and Topic Recognition in Video Transcriptions , 2021, IEEE Intelligent Systems.

[20]  Arun Kumar Sangaiah,et al.  Sentiment analysis: a review and comparative analysis over social media , 2018, Journal of Ambient Intelligence and Humanized Computing.

[21]  Ali Shariq Imran,et al.  Weakly Supervised Framework for Aspect-Based Sentiment Analysis on Students’ Reviews of MOOCs , 2020, IEEE Access.

[22]  Marenglen Biba,et al.  A Thorough Experimental Evaluation of Algorithms for Opinion Mining in Albanian , 2018, EIDWT.

[23]  Aravind Sesagiri Raamkumar,et al.  Measuring the Outreach Efforts of Public Health Authorities and the Public Response on Facebook During the COVID-19 Pandemic in Early 2020: Cross-Country Comparison , 2020, Journal of medical Internet research.

[24]  Marenglen Biba,et al.  Sentiment Analysis through Machine Learning: An Experimental Evaluation for Albanian , 2013, ISI.

[25]  Raymond Chiong,et al.  Multilingual sentiment analysis: from formal to informal and scarce resource languages , 2016, Artificial Intelligence Review.

[26]  Qing Zhu,et al.  COVID-19 Sensing: Negative Sentiment Analysis on Social Media in China via BERT Model , 2020, IEEE Access.

[27]  Mark Kröll,et al.  Sentiment Analysis for German Facebook Pages , 2016, NLDB.