ABSTRACT
Recommendation systems offer a method to assist users in their choice of interest. It is supportive in recommending things from numerous domains. Researchers express their concepts and knowledge in the form of research articles for the outside community. However, they have plenty of options when they start doing groundwork (related work). At times, they end up with an inaccurate selection of research work, resulting in a waste of time and effort. Selecting the most influential research article has been a very mind-numbing task for novice authors. In this work, we introduce the compendium network, a semi-automated curated networked database of Portable Document Formats (PDFs), citations, collaborations, and summaries. This compendium network is built by scraping, integrating, and pre-processing domain-specific research articles in the field of computer science. The authors use this network knowledge to recommend the most influential domain specific research articles to the researcher working in a particular domain. This work suggests an article founded on a content-based approach, by selecting and ranking nearest neighbors of a seed document (query) that are embedded in a linear space and by using a model that discriminates between cited and uncited research articles. Empirical outcomes show that the proposed method attains an average accuracy of 93%, which is a 10–20% improvement in comparison with the existing techniques from the literature.
