On February 15 2019, the ILL staff was invitated to a presentation of PUMA-BI, the new software developed by the ILL within our FILL2030 project.
PUMA combines the ILL databases – of proposals and publications – with Web of Science. As a result, it allows to
- match proposals with publications and
- provide a wide range of analyses and stats for both publications from ILL and from other sources.
Up to now, PUMA-BI has collected over 82500 publications related to ILL or neutron scattering. On the other hand, there was over 18000 accepted proposals in the ILL database.
The software ran over the +1.5 billion document comparison in only few hours thanks to a dedicated cluster of 256 Cores + 1TB Ram.
The characteristic used for comparison were:
- Authors and collaborators,
- Chemical formulae,
- Title and text content similarities,
- Image, DOI & Proposal Codes
As a result, the software suggests matches – ranked by a matching score – which take into account the results of different algorithms.
A tool for strategic decisions to be taken on the management level
The ILL instrument responsibles are invited to log into PUMA in order to check, validate or invalidate the matches suggested by PUMA-BI.
Once one will reach about 3000 validated matches, then the corpus will be sufficiently large to give a solid statistical sense to the figures PUMA-BI allows to compile.
Examples of the statistics extracted from the corpus are below:
Download this slideshow as a PDF!
During the presentation, several improvements were suggested by the audience, mostly composed by ILL scientists. Such feedback is always welcome and you may send your recommandations at any time to firstname.lastname@example.org. The PUMA-BI team will be happy to consider implementing them so that the software evolves towards more efficiency and precision!
The PUMA-BI team.
From left to right: Stuart Caunt, Jean-François Perrin, Sylvain Letreguilly, Virginie Teissier, and Fabien Pinet.