Sentiment and Emotion Based Text Representation for Fake Reviews Detection

Melleng, Alimuddin; Jurek-Loughrey, Anna; Deepak, P

doi:10.26615/978-954-452-056-4_087

Cited by 20 publications

(10 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Here, we were able to classify where words were used that portrayed sadness, fear, joy, disgust, or anger towards the topic of Long Covid. 30 , 31 Whilst it is difficult to accurately measure participants’ true reactions and complex emotions using big data at a more detailed level, our analysis does indicate some preliminary findings that help to establish the basic emotional field of Long Covid sufferers online. Future research could be built upon these findings, using a combination of psychological and textual discourse methods.…”

Section: Methodsmentioning

confidence: 75%

Long Covid: Online patient narratives, public health communication and vaccine hesitancy

Miyake

Martin

2021

DIGITAL HEALTH

View full text Add to dashboard Cite

Introduction This study combines quantitative and qualitative analyses of social media data collected through three key stages of the pandemic, to highlight the following: ‘First wave’ (March to May, 2020): negative consequences arising from a disconnect between official health communications, and unofficial Long Covid sufferers’ narratives online. ‘Second wave’ (October 2020 to January 2021): closing the ‘gap’ between official health communications and unofficial patient narratives, leading to a better integration between patient voice, research and services. ‘Vaccination phase’ (January 2021, early stages of the vaccination programme in the UK): continuing and new emerging concerns. Methods We adopted a mixed methods approach involving quantitative and qualitative analyses of 1.38 million posts mentioning long-term symptoms of Covid-19, gathered across social media and news platforms between 1 January 2020 and 1 January 2021, on Twitter, Facebook, Blogs, and Forums. Our inductive thematic analysis was informed by our discourse analysis of words, and sentiment analysis of hashtags and emojis. Results Results indicate that the negative impacts arise mostly from conflicting definitions of Covid-19 and fears around the Covid-19 vaccine for Long Covid sufferers. Key areas of concern are: time/duration; symptoms/testing; emotional impact; lack of support and resources. Conclusions Whilst Covid-19 is a global issue, specific sociocultural, political and economic contexts mean patients experience Long Covid at a localised level, needing appropriate localised responses. This can only happen if we build a knowledge base that begins with the patient, ultimately informing treatment and rehabilitation strategies for Long Covid.

show abstract

Section: Methodsmentioning

confidence: 75%

Long Covid: Online patient narratives, public health communication and vaccine hesitancy

Miyake

Martin

2021

DIGITAL HEALTH

View full text Add to dashboard Cite

show abstract

“…Wang et al [16] performed a tensor decomposition of 11 relationships that exist between users and products based on reviews and classified them according to the SVM model. Melleng et al [17] combined sentiments and emotions to form review representations, and used these representations to identify fake reviews. In the study of [18], a rule-based featureweighting scheme is proposed that combines various features of reviews, reviewers and products.…”

Section: Related Workmentioning

confidence: 99%

“…SAE [17]: It represents reviews based on a combination of emotion and sentiment, using three sentiment dictionaries and an emotion analysis API that combines sentiments and emotional features to create review representations while using a random forest algorithm for fake review detection.…”

Section: B Baselinesmentioning

confidence: 99%

Detecting Fake Reviews Using Multidimensional Representations With Fine-Grained Aspects Plan

et al. 2021

View full text Add to dashboard Cite

“…Namun, dalam penelitian Elmungrhi & Gherbi memiliki kekurangan karena kurang merepresentasikan dari sentimen yaitu nilai positif atau negatif. Penelitian dari Melleng et. al (2019) melakukan deteksi ulasan palsu dengan pendekatan sentimen dan emosi.…”

unclassified

Klasifikasi Ulasan Palsu Menggunakan Borderline Over Sampling (BOS) dan Support Vector Machine (SVM) (Studi Kasus : Ulasan Tempat Makan)

Awalina

Bachtiar

Indriati

2022

JTIIK

View full text Add to dashboard Cite

Kemudahan memperoleh informasi saat ini telah banyak membantu manusia, salah satu mencari ulasan untuk tempat makan baru. Pencarian ulasan ini dipicu karena pengunjung tidak mengetahui layanan dari tempat tersebut. Ulasan juga dapat menguntungkan penjual, karena mereka mengetahui pengalaman yang dimiliki pengunjungnya. Oleh karena itu, ulasan palsu dimanfaatkan banyak orang untuk membuat ulasan palsu. Ulasan palsu bisa secara efektif dibedakan menggunakan machine learning. Namun, banyak dari dataset ulasan palsu ini tidak seimbang (imbalanced dataset) sehingga dapat mempengaruhi hasil klasifikasi. Oleh karena itu, penelitian ini menggunakan metode BOS untuk mengatasi tidak seimbangnya data dan melakukan klasifikasi dengan metode SVM. Adapun tahapan dari penelitian yaitu preprocessing, lalu pembobotan kata dengan TF-IDF dan fitur sentimen menggunakan lexicon-based features, dilanjutkan proses menyeimbangkan dataset dengan BOS, setelah itu proses klasifikasi oleh SVM. Adapun langkah dalam pengujian BOS dan SVM yaitu pembagian data latih dan uji dengan 80%:20%, setelah itu pencarian parameter terbaik pada data latih dengan 5-fold cross validation, dan dievaluasi dengan data uji. Adapun nilai parameter terbaik pada BOS dan SVM yaitu N dengan nilai 400% dimana hasil evaluasi akurasi dengan nilai 78,6%; precision dengan nilai 19,7%; recall dengan nilai 17,1%; f-measure dengan nilai 14,4%; dan g-mean dengan nilai 32%. Oleh karena itu, penggunaan BOS dapat meningkatkan hasil evaluasi dari terhadap klasifikasi ulasan palsu. AbstractThe convenience of obtaining information nowadays has helped many people such as looking for reviews for new places to eat. The search for reviews was triggered because visitors were not aware of the services of the place. Reviews can also benefit sellers, because they know the experience their visitors have had. Therefore, many people abuse reviews to create spam reviews. Spam reviews can be effectively resolved using machine learning. However, many of these spam review datasets are imbalanced and thus may affect classification results. In this study, BOS algorithm was used to overcome data imbalances, and SVM algorithm for the classification of spam reviews. The stages of the research are preprocessing, then weighting words with TF-IDF and sentiment features using lexicon-based features, followed by the process of balancing the dataset with BOS, and classification process with SVM. Step in testing BOS and SVM are split data of training and test data with 80%:20%, after that the search for the best parameters in the training data with 5-fold cross-validation, and evaluated with test data. The best parameter values for BOS and SVM were N with a value of 400% where the results of the accuracy evaluation were 78.6%; precision with a value of 19.7%; recall with a value of 17.1%; f-measure with a value of 14.4%; and g-mean with a value of 32%. Therefore, use of BOS can improve the evaluation results from the classification of spam reviews.

show abstract

Sentiment and Emotion Based Text Representation for Fake Reviews Detection

Cited by 20 publications

References 21 publications

Long Covid: Online patient narratives, public health communication and vaccine hesitancy

Long Covid: Online patient narratives, public health communication and vaccine hesitancy

Detecting Fake Reviews Using Multidimensional Representations With Fine-Grained Aspects Plan

Klasifikasi Ulasan Palsu Menggunakan Borderline Over Sampling (BOS) dan Support Vector Machine (SVM) (Studi Kasus : Ulasan Tempat Makan)

Contact Info

Product

Resources

About