R. Siva Sankari, J. Dhilipan
Citation recommendations for scholarly articles are a significant priority which can assist researchers in improving work quality by recommending suitable related stuff. Furthermore, present automatic recommendation systems have flaws such as citing irrelevant papers, detecting necessary paperwork promptly, and providing accurate and high-quality recommendations with excellent results. In order to overcome these issues modified deep neural network is proposed to provide high-quality citations to the researchers. Initially, the raw data are collected and preprocessed using three approaches which are tokenization, stemming, and stop word removal. After that extracting the keywords from preprocessed data using co-occurrence statistical information. This approach searches for exact concepts focused on the co-occurrence of terms in the same phrases and total issues. Then converting, the keywords into vector form using TF-IDF vectorizer. After that, the data are grouped using affinity propagation clustering then the extracted data are given as an input for the modified deep neural network. Finally, the classifier predicts whether the recommendation is a global recommendation or a local recommendation. After classification, cosine similarity is used to provide similar citation recommendations to the researchers based on the given query. The simulation analysis shows that the proposed method obtains 99% accuracy, 0.01% error, precision is 99 %, specificity is 99% so on. This demonstrates that the proposed strategy outperforms previous methods currently in use. Assumed from this proposed classification, citation paper quality can be improved and provide high-quality citations to researchers.
Citation recommendation; TF-IDF vectorizer; Affinity Propagation Clustering; DNN; Coot optimization; Cosine similarity
Cite this paper
R. Siva Sankari, J. Dhilipan. (2022) Automatic Citation Recommendation System Based on Modified Deep Neural Network with Coot Optimization. International Journal of Computers, 7, 111-124