A Hybrid Approach for the Recommendation of Scholarly Journals

Abstract

The increasing number of scholarly journals have made it difficult for authors to select the most suitable journal that publishes their research. Existing search systems that recommend journals for manuscript submission are either based on author's profile, bibliographic data or the copublication network. These approaches are not useful for beginner researchers who have no publication records or for those who are interested in new research domains. The present work proposes a hybrid approach that combines clustering and document similarity for the recommendation of scholarly venues. The proposal was evaluated both objectively and subjectively using domain experts. The results of mean average precision (0.84) and normalized discounted cumulative gain (0.89) shows positive recommendations made by the proposed approach.

Bilimsel dergi tavsiyesi için içerik tabanlı bir yaklaşım

“…Yayın geçmişi olmayan ve profil bilgisi tanımlanmamış yeni yazarlar düşünülerek oluşturulan hibrit bir tavsiye modelinde [9], kümeleme ve doküman benzerliği konuları birleştirilmiştir. Kim [10] tarafından ise hem bu problemi hem de içerik tabanlı ve hibrit yaklaşımların getirdiği makalenin yazılmış olma zorunluluğu problemini ortadan kaldıran bir yaklaşım önerilmiştir.…”