[Paper Review] Towards effective research recommender systems for repositories

Petr Knoth, Lucas Anastasiou+5 more|Open Research Online (The Open University)|May 1, 2017

Recommender Systems and Techniques7 references14 citations

TL;DR

This paper presents the CORE Recommender, a state-of-the-art research recommender system integrated into academic repositories and journals, leveraging collaborative filtering and content-based filtering to improve discovery of relevant scholarly articles. The system enhances user experience through personalized recommendations, but faces challenges due to the lack of standardized user interaction logging and global sign-on capabilities in repository infrastructures.

ABSTRACT

In this paper, we argue why and how the integration of recommender systems for research can enhance the functionality and user experience in repositories. We present the latest technical innovations in the CORE Recommender, which provides research article recommendations across the global network of repositories and journals. The CORE Recommender has been recently redeveloped and released into production in the CORE system and has also been deployed in several third-party repositories. We explain the design choices of this unique system and the evaluation processes we have in place to continue raising the quality of the provided recommendations. By drawing on our experience, we discuss the main challenges in offering a state-of-the-art recommender solution for repositories. We highlight two of the key limitations of the current repository infrastructure with respect to developing research recommender systems: 1) the lack of a standardised protocol and capabilities for exposing anonymised user-interaction logs, which represent critically important input data for recommender systems based on collaborative filtering and 2) the lack of a voluntary global sign-on capability in repositories, which would enable the creation of personalised recommendation and notification solutions based on past user interactions.

Motivation & Objective

• To enhance research discovery in academic repositories through an integrated recommender system.
• To address limitations in current repository infrastructures that hinder effective recommendation systems.
• To improve user experience by delivering personalized, high-quality article recommendations across global repositories.
• To identify and overcome key technical and infrastructural barriers in deploying large-scale research recommender systems.
• To evaluate and continuously improve recommendation quality through systematic evaluation processes.

Proposed method

• The CORE Recommender employs a hybrid recommendation approach combining collaborative filtering and content-based filtering using metadata and user interaction data.
• It leverages anonymized user-interaction logs from repositories and journals to model user preferences and item similarities.
• The system is built on a distributed architecture enabling cross-repository recommendations across a global network of repositories.
• It supports integration into third-party repositories through standardized APIs and deployment frameworks.
• Evaluation processes are implemented to monitor and improve recommendation quality over time.
• The system is designed with extensibility in mind to support future enhancements such as personalized notifications and sign-on integration.

Experimental results

Research questions

• RQ1 How can a recommender system effectively enhance research discovery in academic repositories?
• RQ2 What technical and infrastructural challenges impede the development of high-quality research recommender systems in current repository ecosystems?
• RQ3 To what extent can collaborative filtering and content-based filtering be combined to improve recommendation accuracy across diverse repositories?
• RQ4 How can anonymized user interaction logs be effectively utilized when standard protocols for exposing such data are absent?
• RQ5 What role does a global sign-on capability play in enabling personalized recommendation and notification systems?

Key findings

• The CORE Recommender has been successfully rearchitected and deployed in production, demonstrating scalability and cross-repository functionality.
• The system provides personalized recommendations by analyzing user interactions and content metadata across a global network of repositories.
• A major limitation identified is the absence of standardized protocols for exposing anonymized user-interaction logs, which are critical for collaborative filtering.
• Another key challenge is the lack of a voluntary global sign-on system, which hinders persistent user profiling and personalized recommendations.
• Despite these infrastructural constraints, the system achieves high recommendation quality through careful design and evaluation processes.
• The paper highlights the need for community-wide standards in repositories to support next-generation recommendation systems.