As Large Language Models (LLMs) have become more popular in recent years, the number of survey papers in this field has greatly increased. This large amount of research can be difficult for new researchers to navigate. In this paper, I analyze the metadata of these LLM survey papers, looking at publication trends and grouping them by research focus areas. I build a feature matrix using simple methods like TFIDF vectorization and one-hot encoding. I also use a logistic regression model to predict the category of each paper based on its metadata. To address class imbalance in the data, I apply class weighting to improve performance for less common categories. The results show an improvement in accuracy, from 43% to 47%, showing that class weighting helps. This paper provides insights into research trends in the LLM field and helps new researchers explore the literature and find areas for future work.
Upama Thapa Magar (Wed,) studied this question.