Deep context of citations using machine-learning models in scholarly full-text articles

摘要: Information retrieval systems for scholarly literature rely heavily not only on text matching but semantic- and context-based features. Readers nowadays are deeply interested in how important an article is, its purpose influential it is follow-up research work. Numerous techniques to tap the power of machine learning artificial intelligence have been developed enhance most scientific literature. In this paper, we compare improve four existing state-of-the-art designed identify citations. We consider 450 citations from Association Computational Linguistics corpus, classified by experts as either or unimportant, further extract 64 features based methodology techniques. apply Extra-Trees classifier select 29 best Random Forest Support Vector Machine classifiers all selected Using classifier, our supervised model improves method 11.25%, with 89% Precision-Recall area under curve. Finally, present deep-learning model, Long Short-Term Memory network, that uses distinguish unimportant 92.57% accuracy.