A Review on Personalized Academic Paper Recommendation

Abstract

With the advent of the era of big data, it has become extremely easy for scientific users to have to access academic papers, which has enhanced their efficiency and capacity to search or browse papers. However, it also faces some problems such as the explosion of the literature or information overwhelming. Many researchers focus on academic paper recommendation service, hoping to help scientific users to find documents more efficiently and recommend interested or potentially interested papers which could assist academic users doing research. Through literature review, this paper make a comprehensive summary of the research on personalized academic papers recommendation, presenting the state-of-art of academic paper recommendation methodologies, pointing out its pros and cons and indicating primary evaluation metrics and popular datasets. Finnaly, we outlook the research trend of personalized academic paper recommendation as a reference for interested researchers.

Scientific paper recommendation systems: a literature review of recent publications

“…To the best of our knowledge the literature reviews by Bai et al [ 9 ], Li and Zou [ 58 ] and Shahid et al [ 92 ] are the most recent ones targeting the domain of scientific paper recommendation systems. They were accepted for publication or published in 2019 so they only consider paper recommendation systems up until 2019 at most.…”

“…The already mentioned three most recent [ 9 , 58 , 92 ] and one older but highly influential [ 16 ] literature reviews in scientific paper recommendation utilise different categorisations to group approaches. Beel et al [ 16 ] categorise observed papers by their underlying recommendation principle into stereotyping, content-based filtering, collaborative filtering, co-occurrence, graph-based, global relevance and hybrid models.…”

“…Bai et al [ 9 ] only utilise the classes content-based filtering, collaborative filtering, graph-based methods, hybrid methods and other models. Li and Zou [ 58 ] use the categories content-based recommendation, hybrid recommendation, graph-based recommendation and recommendation based on deep learning. Shahid et al [ 92 ] label approaches by the criterion they identify relevant papers with: content, metadata, collaborative filtering and citations.…”

“…The four predominant categories thus are content-based filtering, collaborative filtering, graph-based and hybrid systems. Most of these categories are defined precisely but graph-based approaches are not always characterised concisely: Content-based filtering (CBF) methods are said to be ones where user interest is inferred by observing their historic interactions with papers [ 9 , 16 , 58 ]. Recommendations are composed by observing features of papers and users [ 5 ].…”

“…Recommendations are composed by observing features of papers and users [ 5 ]. In collaborative filtering (CF) systems the preferences of users similar to a current one are observed to identify likely relevant publications [ 9 , 16 , 58 ]. Current users’ past interactions need to be similar to similar users’ past interactions [ 9 , 16 ].…”