Realizing a Scalable and History-Aware Literature Broker Service for OpenAIRE

The OpenAIRE infrastructure is the point of reference for Open Science in Europe. Its services populate and provide access to a graph of objects relative to publications, datasets, people, organizations, projects, and funders aggregated from a variety of data sources, such as institutional repositories, data archives, journals, and CRIS systems. Not only, objects in the graph are harmonized to achieve semantic homogeneity, de-duplicated and merged, and enriched by inference with missing properties and/or relationships. The OpenAIRE Literature Broker Service is designed to offer subscription and notification functionalities for institutional repositories to: (i) learn about publication objects in OpenAIRE that do not appear in their collection but may be pertinent to it, and (ii) learn about extra properties or relationships relative to publication objects in their collection. Due to the high variability of the information space the following problems may arise: (i) subscriptions may vary over time to adapt to information space evolution, (ii) repository managers need to be able to quickly test their configurations before activating them, (iii) notifications may be redundant, and (iv) notifications may be very large over time. This paper presents the data model and software architecture of the OLBS, specifically designed to address these issues.