A RESPONSE SURFACE MODELING METHOD FOR SPAM CLASSIFICATION
Keywords:
Response surface methodology, Spam classification, Non-linear modeling, Feature transformation, Deep neural networksAbstract
Spam identification is an important task in Natural Language Processing. To improve the limited nonlinear representation of traditional linear classifiers, this paper proposes a spam classification model based on Response Surface Methodology. The model constructs a dual response surface on Bag-of-Words features: feature-level quadratic transformations learn nonlinear keyword effects, while a classifier-level quadratic correction adjusts the linear decision output. The proposed structure preserves interpretability through explicit polynomial parameters and improves boundary fitting without relying on large-scale deep models. Experiments on public spam datasets show that the model achieves higher accuracy than Logistic Regression and LSTM, while maintaining much lower computational cost than BERT. The results indicate that response surface modeling provides a lightweight, interpretable, and efficient solution for spam identification. This method is especially suitable for resource-constrained scenarios requiring fast training, low inference cost, and transparent feature-level explanation, and it offers a practical balance between model complexity, classification accuracy, and deployment efficiency.References
[1] Asim S, Mohd N N, Zubair M R, et al. An Improved Framework for Content-and-Link-Based Web-Spam Detection: A Combined Approach. Complexity, 2021. DOI: 10.1155/2021/6625739.
[2] Shan C L, Zhang X Y, Xing H L, et al. A Chinese Spam Recognition Method Based on Content and ERNIE3.0-CapsNet. Journal of Information Security Research, 2024, 10(03): 233-240.
[3] Gui J, Zhou Y, Yu K, et al. PSC-BERT: A spam identification and classification algorithm via prompt learning and spell check. Knowledge-Based Systems, 2024, 301: 112266.
[4] Dawei Z, Yanyu L. Identification and Filtering of Web Spams Using a Machine Learning Method. International Journal of Computational Intelligence and Applications, 2022, 21(04).
[5] Zengle G. A fusion algorithm model based on KNN-SVM to classify and recognize spam. Journal of Physics: Conference Series, 2021, 1982(1).
[6] Zhang Y Q, Wang W. Spam Classification Using Support Vector Machine Optimized by Genetic Algorithm. Journal of Computer Applications, 2009, 29(10): 2755-2757.
[7] Sharma P K, Lal G, Shukla M, et al. Quantum behaved binary gravitational search algorithm with random forest for twitter spammer detection. Results in Engineering, 2025, 25: 103993.
[8] Ömer A, Serkant S A, Merve O, et al. A Comprehensive Review of Cyber Security Vulnerabilities, Threats, Attacks, and Solutions. Electronics, 2023, 12(6): 1333.
[9] Faris H, Al-Zoubi M A, Heidari A A, et al. An Intelligent System for Spam Detection and Identification of the Most Relevant Features Based on Evolutionary Random Weight Networks. Information Fusion, 2018, 48: 67-83.
[10] Wang Z Y, Xu W R, Guo J. New Technology for Intelligent Text Search. CAAI Transactions on Intelligent Systems, 2012, 7(01): 40-49.
[11] Shang T, Guo Z Y, Wang Y S. A High-Accuracy Spam Recognition Method. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2011, 39(S2): 287-290.
[12] Xu Y B, Li Z, Dong Y. Fast Spam Filtering Based on Social Computing and Machine Learning. Systems Engineering — Theory & Practice, 2014, 34(S1): 179-186.
[13] Chen J Y, Zhou S F, Min H Q. A Hybrid "Word Frequency-Filter" Feature Selection Method for Spam Recognition. Journal of South China University of Technology (Natural Science Edition), 2017, 45(03): 82-88.
[14] Zengle G. A fusion algorithm model based on KNN-SVM to classify and recognize spam. Journal of Physics: Conference Series, 2021, 1982(1).
[15] Dawei Z, Yanyu L. Identification and Filtering of Web Spams Using a Machine Learning Method. International Journal of Computational Intelligence and Applications, 2022, 21(04).