Performance Comparison of Traditional and Contextual Representations for Cryptocurrency Sentiment Analysis on Twitter

Melisa Ateş; Muhammet Sinan Başarslan

Article Details

Performance Comparison of Traditional and Contextual Representations for Cryptocurrency Sentiment Analysis on Twitter

Authors:

Melisa Ateş | Muhammet Sinan Başarslan

Publisher:

Duzce University

Main Topics:

Derin Öğrenme | Makine Öğrenmesi Algoritmaları | Sınıflandırma algoritmaları | Makine Öğrenme (Diğer)

Journal:

Düzce Üniversitesi Bilim ve Teknoloji Dergisi Vol.13 Issue 3, pp.1431-1444

Document Type:

Article

Publication Date:

7/2025

Document Language:

English

Document ID (Org No.):

2001355749

pISSN:

2148-2446

Keywords:

Abstract:

 In recent years, discussions about cryptocurrencies, particularly on platforms such as Twitter, have become increasingly prevalent. This study focuses on conducting a sentiment analysis SA of tweets related to cryptocurrencies, applying machine learning ML and deep learning DL methodologies based on natural language processing NLP. This research used a total of 10,000 tweets collected from open sources between 2020 and 2021. Prior to analysis, the dataset underwent detailed pre-processing, during which non-textual elements such as emojis, links, and HTML codes were removed. TF-IDF was initially employed to generate text representations. Various traditional ML models were applied, including Naïve Bayes NB, Decision Tree DT, Support Vector Machine SVM. Advanced DL models were also used, including Bidirectional Long Short-Term Memory BiLSTM and Bidirectional Gated Recurrent Unit BiGRU. To capture contextual relationships more effectively, text embeddings generated by the Bidirectional Encoder Representations from Transformers BERT model were also utilised. When performance was evaluated, the BERT-based BiGRU model achieved the highest Accuracy Acc of 93 and the best F1 score. This demonstrates the effectiveness of combining deep contextual embeddings with models capable of learning from sequential patterns. Overall, the findings suggest that DL approaches, particularly those that incorporate advanced representation methods such as BERT, can significantly outperform traditional models in sentiment classification tasks.

 In recent years, discussions about cryptocurrencies, particularly on platforms such as Twitter, have become increasingly prevalent. This study focuses on conducting a sentiment analysis SA of tweets related to cryptocurrencies, applying machine learning ML and deep learning DL methodologies based on natural language processing NLP. This research used a total of 10,000 tweets collected from open sources between 2020 and 2021. Prior to analysis, the dataset underwent detailed pre-processing, during which non-textual elements such as emojis, links, and HTML codes were removed. TF-IDF was initially employed to generate text representations. Various traditional ML models were applied, including Naïve Bayes NB, Decision Tree DT, Support Vector Machine SVM. Advanced DL models were also used, including Bidirectional Long Short-Term Memory BiLSTM and Bidirectional Gated Recurrent Unit BiGRU. To capture contextual relationships more effectively, text embeddings generated by the Bidirectional Encoder Representations from Transformers BERT model were also utilised. When performance was evaluated, the BERT-based BiGRU model achieved the highest Accuracy Acc of 93 and the best F1 score. This demonstrates the effectiveness of combining deep contextual embeddings with models capable of learning from sequential patterns. Overall, the findings suggest that DL approaches, particularly those that incorporate advanced representation methods such as BERT, can significantly outperform traditional models in sentiment classification tasks.

 In recent years, discussions about cryptocurrencies, particularly on platforms such as Twitter, have become increasingly prevalent. This study focuses on conducting a sentiment analysis SA of tweets related to cryptocurrencies, applying machine learning ML and deep learning DL methodologies based on natural language processing NLP. This research used a total of 10,000 tweets collected from open sources between 2020 and 2021. Prior to analysis, the dataset underwent detailed pre-processing, during which non-textual elements such as emojis, links, and HTML codes were removed. TF-IDF was initially employed to generate text representations. Various traditional ML models were applied, including Naïve Bayes NB, Decision Tree DT, Support Vector Machine SVM. Advanced DL models were also used, including Bidirectional Long Short-Term Memory BiLSTM and Bidirectional Gated Recurrent Unit BiGRU. To capture contextual relationships more effectively, text embeddings generated by the Bidirectional Encoder Representations from Transformers BERT model were also utilised. When performance was evaluated, the BERT-based BiGRU model achieved the highest Accuracy Acc of 93 and the best F1 score. This demonstrates the effectiveness of combining deep contextual embeddings with models capable of learning from sequential patterns. Overall, the findings suggest that DL approaches, particularly those that incorporate advanced representation methods such as BERT, can significantly outperform traditional models in sentiment classification tasks.

 Son yıllarda, özellikle Twitter gibi platformlarda kripto para birimleri hakkındaki tartışmalar giderek yaygınlaşmaktadır. Bu çalışma, doğal dil işleme NLP tabanlı makine öğrenimi ML ve derin öğrenme DL metodolojilerini uygulayarak kripto paralarla ilgili tweetlerin duygu analizini SA yapmaya odaklanmaktadır. Bu araştırmada 2020 ve 2021 yılları arasında açık kaynaklardan toplanan toplam 10.000 tweet kullanılmıştır. Analiz öncesinde veri kümesi, emojiler, bağlantılar ve HTML kodları gibi metin dışı öğelerin kaldırıldığı ayrıntılı bir ön işlemden geçirilmiştir. Metin temsilleri oluşturmak için başlangıçta TF-IDF kullanılmıştır. Naïve Bayes NB, Karar Ağacı DT, Destek Vektör Makinesi SVM dahil olmak üzere çeşitli geleneksel makine öğrenimi modelleri uygulanmıştır. Çift Yönlü Uzun Kısa Süreli Bellek BiLSTM ve Çift Yönlü Geçitli Tekrarlayan Birim BiGRU dahil olmak üzere gelişmiş DL modelleri de kullanılmıştır. Bağlamsal ilişkileri daha etkili bir şekilde yakalamak için, Transformatörlerden Çift Yönlü Kodlayıcı Temsilleri BERT modeli tarafından üretilen kelime gömmede kullanılmıştır. Performans değerlendirildiğinde, BERT tabanlı BiGRU modeli en yüksek doğruluğu 93 ve en iyi F1 puanını elde etmiştir. Bu, derin bağlama dayalı kelime gömmeleri sıralı örüntülerden öğrenme yeteneğine sahip modellerle birleştirmenin etkinliğini göstermektedir. Genel olarak bulgular, DL yaklaşımlarının, özellikle de BERT gibi gelişmiş temsil yöntemlerini içerenlerin, duygu sınıflandırma görevlerinde geleneksel modellerden önemli ölçüde daha iyi performans gösterebileceğini göstermektedir.

Article Details

Performance Comparison of Traditional and Contextual Representations for Cryptocurrency Sentiment Analysis on Twitter

Authors:

Melisa Ateş | Muhammet Sinan Başarslan

Publisher:

Duzce University

Main Topics:

Derin Öğrenme | Makine Öğrenmesi Algoritmaları | Sınıflandırma algoritmaları | Makine Öğrenme (Diğer)

Journal:

Düzce Üniversitesi Bilim ve Teknoloji Dergisi Vol.13 Issue 3, pp.1431-1444

Document Type:

Article

Publication Date:

7/2025

Document Language:

English

Document ID (Org No.):

2001355749

pISSN:

2148-2446

Cryptocurrency

sentiment analysis

text representation

ML

DLKripto para

duygu analizi

kelime temsili

ML

DL

Issues

Article Details

Authors: Melisa Ateş | Muhammet Sinan Başarslan

Publisher: Duzce University

Main Topics: Derin Öğrenme | Makine Öğrenmesi Algoritmaları | Sınıflandırma algoritmaları | Makine Öğrenme (Diğer)

Journal: Düzce Üniversitesi Bilim ve Teknoloji Dergisi Vol.13 Issue 3, pp.1431-1444

Document Type: Article

Publication Date: 7/2025

Document Language: English

Document ID (Org No.): 2001355749

pISSN: 2148-2446

Issues

Authors:

Melisa Ateş | Muhammet Sinan Başarslan

Publisher:

Duzce University

Main Topics:

Derin Öğrenme | Makine Öğrenmesi Algoritmaları | Sınıflandırma algoritmaları | Makine Öğrenme (Diğer)

Journal:

Düzce Üniversitesi Bilim ve Teknoloji Dergisi Vol.13 Issue 3, pp.1431-1444

Document Type:

Article

Publication Date:

7/2025

Document Language:

English

Document ID (Org No.):

2001355749

pISSN:

2148-2446