A Comparative Study of Attention-Based Transformer Networks and Traditional Machine Learning Methods for Toxic Comments Classification

Table 2. Comparative Analysis of Models' Ability to CaptureContextual Dependencies, Handle Long-Range Dependencies, and Learn EffectiveRepresentations for Toxic Language Classification

Model	Ability to Capture Contextual Dependencies	Handling Long-Range Dependencies	Effective Representation for Toxic Language
Logistic Regression	No	No	No
Naive Bayes	No	No	No
Support Vector Machines (SVM)	No	No	No
Decision Trees	Partial	Partial	Partial
Random Forests	Partial	Partial	Partial
Gradient Boosting	Partial	Partial	Partial
Attention-Based Transformer	Yes	Yes	Yes