ORCID state of dblp

Since more than a year now, dblp has intensified its efforts to link dblp bibliographies to ORCIDs used by that author. ORCID information are now added regularly to the dblp data set. The primary sources for ORCIDs are: first, the annual ORCID open data dump and, second, metadata directly provided by publishers who have started to increasingly label author signatures with ORCID information. Neither of those data sources are free of errors and data hick-ups, so we are still manually cleaning the ORCID data prior to adding them to the corpus. But overall, ORCIDs have helped us to correct numerous cases of homonymous and synonymous bibliographies in dblp, so it is absolutely worth our time.

Since the past big ORCID update from last week (as of July 11, 2018), you can now find in dblp:

• 806,744 ORCID author signatures (i.e., author-publication pairs), that is 6.5% of all author signatures in the dataset
• 165,752 bibliographies that are linked to ORCIDs (either manually or implicit)
• 24,357 bibliographies with ORCIDs manually verified by the dblp team

If you look at computer science publications in dblp that have been published in the recent years, you will find that the fraction of author signatures with ORCID in dblp has climbed to now about 10.9% off all signatures in 2018. Of course, year 2018 is not done yet, and we are continuously working on improving the coverage further among all publications, regardless of the year of publication. The number of signatures and coverage for publication in recent years is:

• 2014: 892,240 signatures (8.2% coverage)
• 2015: 930,746 signatures (7.7% coverage)
• 2016: 984,559 signatures (6.7% coverage)
• 2017: 1,042,103 signatures (5.6% coverage)
• 2018: 439,350 signatures (10.9% coverage)

The oldest publication with an ORCID is from 1961. Because of the small number of publications in dblp from that time, the ORCID coverage for 1961 is a remarkable 0.12%