Automated paper annotation with ReaderBench

Ionut Cristian Paraschiv (1) , Mihai Dascalu (1) , Ștefan Trăușan-Matu (1) , Philippe Dessus (2)

1 Université Politehnica
2 LSE - Laboratoire des Sciences de l'Éducation (Grenoble)

The annotation of articles from a given domain and the generation of semantic metadata can be considered a reliable foundation for creating a paper recommender system. Within this paper, the models from other previous researches are extended with the capability of visualizing articles and the most important concepts from a domain within imposed timeframes. This can be very useful for researchers to check out the most important publications from a given period, to view which are the trends and how a domain has evolved. Our previous analyses used the articles to build a paper graph and to suggest the most relevant articles, given a user defined query in natural language. This research contains a use case and creates visual graph representations to enhance the overall perception of the evolution of a domain.

Mots clés: Scientometrics; paper recommendation system; time analysis; discourse analysis; semantic similarity

Domaines: Education; Environnements Informatiques pour l'Apprentissage Humain

Ionut Cristian Paraschiv, Mihai Dascalu, Ștefan Trăușan-Matu, Philippe Dessus. Automated paper annotation with ReaderBench. 12-a Conferinta Nationala de Interactiune Om-Calculator (RoCHI 2015), Sep 2015, Bucharest, Romania. pp.75-80. ⟨hal-01217024⟩