Recommending High Impact Scholarly Articles using Context-Based Network Analysis

Recommending similar articles is one of the essential tools in social network analysis, used to identify relationships that are not else perceived. Commonly, such a system relies on similarity metrics, with the notion that similar pairs are likely to have common preferences. However, traditional methods neither consider imprecise data, binary citation frequencies nor the citation context. The present method proposes a supervised model with 3 features. Three features in specific, link-based doc2vec for assessing authors’ co-cited sentence similarity. Author fetched preprints in the condensed matter under the computer science category based on the category called autism. Our model performs well, with precision, recall values of between 0.78 and 0.67 as this is a real time dataset, and our model provides the top 5 recommendations of articles to the authors.