Topic Identification of Science and Mathematics Literature Using Latent Dirichlet Allocation

Amidst the ever-expanding realm of scientific and mathematical literature, distilling valuable insights from a multitude of articles holds paramount importance. This study introduces a fresh perspective on identifying topics within science and mathematics articles, employing Latent Dirichlet Allocation (LDA) as its cornerstone. LDA, serving as a probabilistic generative model, is adeptly harnessed to unveil latent topics nestled within a corpus of articles, thereby illuminating intrinsic thematic architectures. Our LDA model, meticulously crafted, has the capability to discern five distinct categorizations of topics, offering a nuanced understanding of the diverse array of subjects encapsulated within the domain of science and mathematics. This approach not only facilitates the systematic organization of voluminous literature but also fosters a deeper comprehension of prevalent themes and emerging trends across these disciplines. Evaluation result, based on coherence scores value, in modelling topics related to science and mathematics, with a coherence score of 0,47508 as the highest using alpha is 0.01 and beta is 0.91.