Contextual Revelation of Intentions in Machine Learning Articles on arXiv: A New Dimension in Academic Content Analysis

Abstract

Our study focuses on developing an innovative system for classifying intentions in machine learning articles on arXiv. This research aims to bring a new dimension to the analysis of academic content by identifying and classifying the underlying intentions of authors, a dimension not covered by traditional search systems. Our method, which enriches content with contextual classifications, transcends simple domain categorization by incorporating specific intention identifications such as proposing new methods, analyzing existing data, or presenting experimental results. To achieve this, we have implemented natural language processing models based on transformers, which have proven to be extremely effective for this task, offering high accuracy in intention classification. The methods used and the results obtained demonstrate the effectiveness of these advanced models in enriching academic articles, thus contributing to the creation of a valuable and contextual tool for the scientific community, particularly in the field of machine learning.