Spam Detection Using Dense-Layers Deep Learning Model and Latent Semantic Indexing

In the digital age, online shoppers heavily depend on product feedback and reviews available on the corresponding product pages to guide their purchasing decisions. Feedback is used in sentiment analysis, which is helpful for both customers and company management. Spam feedback can have a negative impact on high-quality products or a positive impact on low-quality products. In both cases, the matter is bothersome. Spam detection can be done with supervised or unsupervised learning methods. We suggested two direct methods to detect feedback orientation as ‘spam’ or "not spam", also called "ham," using the deep learning model and the LSI (Latent Semantic Indexing) technique. The first proposed model uses only dense layers to detect the orientation of the text. The second proposed model uses the concept of LSI, an effective information retrieval algorithm that finds the closest text to a provided query, i.e., a list containing spam words. Experimental results of both models using publicly available datasets show the best results (89% accuracy and 89% precision) when compared to their corresponding benchmarks.

@article{Al-Otaibi2025,
  title     = {Spam Detection Using Dense-Layers Deep Learning Model and Latent Semantic Indexing},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {16},
  number    = {1},
  year      = {2025},
  publisher = {The Science and Information Organization},
  author    = {Yasser D. Al-Otaibi and Shakeel Ahmad and Sheikh Muhammad Saqib},
  doi       = {10.14569/IJACSA.2025.0160137},
  url       = {https://doi.org/10.14569/IJACSA.2025.0160137}
}