Retraction Notice: Evaluating the Enhanced Performance of Tree-Based Models for Natural Language Understanding
Adlin Jebakumari S, Aasheesh Raizada, Rahul Vishnoi · 2024
Tree-based models for natural Language expertise (NLU) have been gaining elevated interest in recent years because of their ability to seize complex relationships among words and phrases in a sentence. In this study, we gift a comprehensive assessment of improved tree-primarily based models for NLU in phrases of accuracy, interpretability, and velocity. Primarily, we evaluate numerous current tree-based models to two newly proposed fashions: a two-Layer Tree (T2T) community and a Weighted Random Tree (WRT) version. We discover that each fashion outperforms the present tree-based models in phrases of accuracy and interpretability without a significant loss in velocity. Normal, our outcomes reveal that the mixture of those more vital tree-based total models offers a practical answer for NLU tasks, specifically for language information applications in sentiment evaluation and question-answering domain names.