“…Furthermore, text-based approaches are considered superior to citation-based ones for document categorization [3]. The used approaches differ in three aspects: (1) text sections (i.e., abstract, keywords, full text), (2) objective (e.g., classification, recommendation, content extraction, clustering), and (3) used techniques (e.g., bag-of-words, vectorization, Bayesian classifier, topic models, keyword extraction) [1,14,20,41].…”