An Effective Spam Message Detection Model using Feature Engineering and Bi-LSTM

Rosewelt, L. Antony; Raju, Naveen D; Ganapathy, Sannasi

doi:10.1109/accai53970.2022.9752652

Cited by 15 publications

(2 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The parameter settings of FT-DBN are provided in this section with the 100-batch size, 200 epochs, and activation functions such as ReLU and Sigmoid with a learning rate of 0.01. The network size is (2,6,8,10) used in this FT-DBN for evaluating the model using the Amazon dataset. This work uses two activation functions and a different number of approaches to train and test the model by considering the Amazon dataset.…”

Section: Classificationmentioning

confidence: 99%

“…The various classification algorithms, including Random Forest, Logistic Regression, Decision Tree, C4.5, Support Vector Machine (SVM), Multiclass SVM, Neural Networks, Deep Belief Networks (DBN), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long-Short Term Memory (LSTM) and Bidirectional LSTM (Bi-LSTM) (2) are used in this field to categorise the user's interests according to their purchase behaviour and predicts the customers purchase behaviour. According to the classification result, the products are recommended to the customers.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Sentiment and Fuzzy Aware Product Recommendation System Using HOA and FT-DBN in E- Commerce

Manikandan¹,

Rämä²,

S³

2023

IJST

View full text Add to dashboard Cite

Objectives: To identify and select the customers' liked products by introducing a new product recommendation system. Methods: This work proposes a new product recommendation system that incorporates a new feature optimization method called Sentiment weighted Horse herd Optimization Algorithm (SHOA) to identify the most suitable words that help perform effective prediction. This work's prediction process is carried out by applying a newly proposed Deep Belief Network incorporating fuzzy temporal features. This work uses two different Amazon datasets. The first dataset contains 51, 00,000 review comments about various products, including books and movies. The second dataset is built with 82,00,000 review comments on Toys and Games. These data sets consider the product id and review rate important features and are used to compare with all other available works through experimental results. Findings:The experiments have been conducted using the Amazon dataset and proved better than other recommendation systems in terms of effectiveness and efficiency through Precision, Recall, Serendipity and nDCG value. Novelty: The introduction of a new DBN with Fuzzy Temporal rules and the newly developed SHOA is novel in this work to recommend suitable products to the customer.

show abstract

Section: Classificationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%