A Novel Automatic Journal Recommender System

Abstract

Recommender Systems (RS) have developed into an important application in several user domains. RS may aid users to discover appropriate items in vast data. The selection of journal publication is generally based on the research domain or topics of document. The traditional method for journal recommendation is carried out by analyzing the document and matching its topics with relavent journal utilizing content-based examination. Though, this approach might create errors because of disparities in manuscript comparisons. In this paper, a novel Jaccard based Journal Finder Neural Network is proposed with pearson correlation coefficient (JJFNN-PC) for journal recommendation. The proposed recommender system allows the researchers to automatically find appropriate publication with journal title and abstract. Similarity coefficient is computed among the journal database and journal title and abstract of user distinctly through Jaccard similarity. The obtained outcome is used for training the novel JFNN that automatically find appropriate publication for user research article. The pearson correlation coefficient is established to validate the correlation between title and abstract of the recommended journal. The experimental result of automatic journal selection process provides the exact journal list and obtained better performance with accuracy of 98.41%.