Skip to content

2. Data quality and completeness

Jump to bottom
Alberto Cottica edited this page Jul 1, 2017 · 1 revision

Improvements

We are happy to acknowledge an improvement in the quality of the data with respect to previous years. Specifically, unlike in the past, all organisations are now identified by their unique PIC codes. This removes the need for time-consuming and messy disambiguation based on the free text organisationName field (Consiglio Nazionale delle Ricerche, CNRand C.N.R.are the same organisation with three different spellings). Big win!

Remaining issues

What are projects about?

Come si fa a capire la disciplina a cui afferiscono i progetti? Il codice SIC qualche volta è parte del campo chiamato topic, ma più spesso no.

TODO Tommaso?

Adding the non-funded proposals to the dataset

Data on research funding proposals could yield valuable insights on what makes a proposal "excellent" or "successful". In order to do that, however, data scientists need information on the non-funded proposals as a control group, against which to compare the successfully funded ones. At the time of writing, only the latter are included in the CORDIS dataset. We asked the European Commission to release data on the former (they are, after all, carefully archived in the Participant Portal's DB), but they refused, on IP protection grounds.

We respectfully disagree on several grounds.

  1. There is no evidence that publishing data of unsuccessful proposals in the CORDIS format has any IP implications. Details on the proposals is not published at all in CORDIS, not even for the funded proposals. The only hint is a free-form field called objectives; if this is the concern, a compromise could be to omit it.

  2. Full information on participation to Horizon 2020 has a potential for "market making", in the sense of empowering companies, universities and research organisations to better choose their prospective partners. Suppose the University of X invited company Y to be a part of a project's consortium. Writing H2020 applications is hard work: Y would like to know what X's track record is. What's their acceptance percentage? Are they the kind of organisations that submits 20 projects to get one funded, or are they focusing their effort on a small number of projects per call? Knowing this would allow everyone to choose partners with the desired risk profile, and that way increase transparency and efficiency.

  3. We believe the right to confidentiality of participant organisations conflicts with their right to vet prospective partners, and with that of citizens to transparency. We intend to take up our request formally, with the Commission's Common service for Horizon 2020 information and data.

Clone this wiki locally