Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

A Hybrid Model for spam review detection using fine-tuned BERT and DistilBERT

Xue, Chao LU (2023) DABN01 20231
Department of Statistics
Department of Economics
Abstract (Swedish)
Spam review detection has gathered significant attention within the Natural Language Processing (NLP) academic community, especially in recent years due to the growing reliance on online reviews for purchases and the increased uti- lization of Artificial Intelligence. The prevailing methods for identifying fraudulent reviews often concentrate on text-based features employing diverse techniques. Re- cent strides have been made with BERT’s involvement to enhance performance. However, a limited number of studies incorporated the process of fine-tuning BERT through addressing the difficulties presented by class imbalance. This paper intro- duces a hybrid model featuring fine-tuned BERT for the purpose of detecting fake reviews. The model... (More)
Spam review detection has gathered significant attention within the Natural Language Processing (NLP) academic community, especially in recent years due to the growing reliance on online reviews for purchases and the increased uti- lization of Artificial Intelligence. The prevailing methods for identifying fraudulent reviews often concentrate on text-based features employing diverse techniques. Re- cent strides have been made with BERT’s involvement to enhance performance. However, a limited number of studies incorporated the process of fine-tuning BERT through addressing the difficulties presented by class imbalance. This paper intro- duces a hybrid model featuring fine-tuned BERT for the purpose of detecting fake reviews. The model combines textual-based features with numerical behavior fea- tures in a concatenation layer, and compares results with text-only baseline model. Experimental results reveal that the fusion technique does not significantly outper- form the baseline. (Less)
Please use this url to cite or link to this publication:
author
Xue, Chao LU
supervisor
organization
course
DABN01 20231
year
type
H1 - Master's Degree (One Year)
subject
keywords
Spam review detection, BERT, imbalanced class, feature fusion, DistilBERT
language
English
id
9135175
date added to LUP
2023-11-21 12:54:55
date last changed
2023-11-21 12:54:55
@misc{9135175,
  abstract     = {{Spam review detection has gathered significant attention within the Natural Language Processing (NLP) academic community, especially in recent years due to the growing reliance on online reviews for purchases and the increased uti- lization of Artificial Intelligence. The prevailing methods for identifying fraudulent reviews often concentrate on text-based features employing diverse techniques. Re- cent strides have been made with BERT’s involvement to enhance performance. However, a limited number of studies incorporated the process of fine-tuning BERT through addressing the difficulties presented by class imbalance. This paper intro- duces a hybrid model featuring fine-tuned BERT for the purpose of detecting fake reviews. The model combines textual-based features with numerical behavior fea- tures in a concatenation layer, and compares results with text-only baseline model. Experimental results reveal that the fusion technique does not significantly outper- form the baseline.}},
  author       = {{Xue, Chao}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{A Hybrid Model for spam review detection using fine-tuned BERT and DistilBERT}},
  year         = {{2023}},
}