The Microsoft Academic Graph (MAG) is a graph containing scientific publication data and (citation) relationships between the graph entities (publications, authors, institutions, journals, conferences, and fields of study), see https://www.microsoft.com/en-us/research/project/microsoft-academic-graph/ , and currently comprises about 150 million publications. It can be accessed using the Microsoft Cognitive Services Academic Knowledge API (AK API, https://www.microsoft.com/cognitive-services/en-us/academic-knowledge-api ).
This software repository provides a bin/academic_search script that can be used to access the AK API and the MAG data starting from an EPrints repository.
It has been employed for a study to assess coverage of ZORA (Zurich Open Repository and Archive data) publications in MAG. The findings of this study have been submitted to the Journal of the Association for Information Science and Technology (JASIST) and are available on arXiv: https://arxiv.org/abs/1703.05539
When using this code or parts of it for scientific purposes, please cite this study as follows:
Sven E. Hug, Martin P. Brändle, The coverage of Microsoft Academic: Analyzing the publication output of a university, arXiv:1703.05539 [cs.DL] (2017).
While this code is specific to checking coverage of an EPrints repository against MAG and is submitted here as supplemental material to the article, parts of it can be used to write own code to access the AK API and use citation or other metadata.
To access the AK API, an API key (free or subscribed access) is required. See https://www.microsoft.com/cognitive-services/en-us/pricing
The API key, the accessed entities and other settings can be configured in cfg/cfg.d/z_ms_knowledge_api.pl
mapping_example/mapping_example.csv provides an example for mappings of institutes to the research fields according to the field of science and technology (FOS) classification in the Frascati manual.