The scientific world is witnessing an unprecedented expansion in the volume of knowledge production and an increasing complexity in the interdisciplinary overlap, which calls for the development of intelligent systems capable of accurately and efficiently classifying scientific terms. This study aims to propose an advanced model for the automatic classification of scientific terms using deep learning techniques, based on a massive database from the arXiv platform comprising over 136,000 research documents covering nine major scientific fields. The research methodology relied on precise text processing, including linguistic cleaning and morphological analysis, along with representing terms in a numerical space using Word2Vec technology, followed by building a classification model based on Long Short-Term Memory (LSTM) neural networks, enhancing the ability to understand semantic contexts and complex conceptual relationships. The model's results demonstrated promising effectiveness, achieving an overall classification accuracy of 72%, with outstanding performance in fields such as computer vision at 89% and natural language processing at 87%. The model also showed a remarkable ability to distinguish between different contexts of multi-use terms, although some challenges persisted in classifying underrepresented categories. The study illustrates that deep learning techniques offer effective potential for managing and organizing scientific terminology, while also highlighting the importance of addressing challenges related to the representation of rare categories and the need for more advanced models to understand fine grained cognitive structures. The study recommends several future directions for model development, including expanding the range of covered scientific fields, and emphasizes the necessity of developing multilingual solutions that consider the diverse cultural and cognitive contexts of scientific terms. This research constitutes a qualitative addition to the field of intelligent knowledge organization and paves the way for building more advanced tools to support scientific research and improve the management of academic content in digital environments................. Keywords:............ scientific term classification; deep neural networks; text mining; LSTM; natural language processing.