Exploring Large Language Model survey papers via Machine and Ensemble Learning

Abstract

Nowadays, there is an influx of researchers emphasizing Large Language Models (LLMs). While the field is broadening, it becomes difficult to keep up all the models, and techniques associated with the novel idea. To tackle this problem, a study has been conducted for assigning survey papers to taxonomy in an automated way. In this assignment, I am using their dataset for the task of exploration, manipulation, and evaluation. After finishing the instructed part, I did further exploration by using a cross tab between taxonomy and date, representing different visualizations for survey papers by taxonomy over time, and plotting the box of release day by taxonomy title. The experimental analysis indicates that Logistic Regression (LR) outperformed all the 8 Classifiers in terms of accuracy score, while GaussianNB (GNB) shows the most commendable precision score. For weighted recall and f1 score, LR shows the highest performance in text classification data.