A content-sensitive citation representation approach for citation recommendation

Citation recommendation systems mainly help researchers find the lists of references that related to their interests effectively and automatically. The existing approaches face issues data sparsity high-dimensional in large-scale bibliographic network representation, which hinder citation performance. To address these problems, we proposed a Content-Sensitive representation approach for Recommendation, named CSCR. Firstly, Doc2vec model is used generate paper embedding according content. Then, utilizing similarity between content embeddings select assumed neighbours target paper, append auxiliary links its new network. Thirdly, distributed method implemented on appended obtain node embedding, can learn interpretable lower dimension nodes. Finally, vectors papers be conduct recommendation. Experimental results show significantly outperforms other benchmark methods Normalized Discounted Cumulative Gain (NDCG) positive rate (Recall).