Toward Altmetric-Driven Research-Paper Recommender System Framework

The volume of literature and more particularly research-oriented publications is growing at an exponential rate, and better tools and methodologies are required to efficiently and effectively retrieve desired documents. The development of academic search engines, digital libraries and archives has led to better information filtering mechanisms that has resulted to improved search results. However, the state-of-the art research-paper recommender systems are still retrieving research articles without explicitly defining the domain of interest of the researchers. Also, a rich set of research output (research objects) and their associated metrics are also not being utilized in the process of searching, querying, retrieving and recommending articles. Consequently, a lot of irrelevant and unrelated information is being presented to the user. Then again, the use of citation counts to rank and recommend research-paper to users is still disputed. Recommendation metrics like citation counts, ratings in collaborative filtering, and keyword analysis' cannot be fully relied on as the only techniques through which similarity between documents can be computed, and this is because recommendations based on such metrics are not accurate and have lots of biasness. Henceforth, altmetric-based techniques and methodologies are expected to give better recommendations of research papers since the circumstances surrounding a research papers are taken into consideration. This paper proposes a research paper recommender system framework that utilizes paper ontology and Altmetric from research papers, to enhance the performance of research paper recommender systems.