5.1 Policy
As part of the end-to-end information management framework, all GTOS datasets will have a designated data custodian
5.2 Discussion
Datasets are not generated in isolation. They are developed within organisations and institutions to meet corporate objectives. A custodian is the body responsible for the development, maintenance and quality of a dataset, and for the arrangements for access to it. The most important aspect of a custodian is that they should have the scientific and technical knowledge and expertise to be in the best position to assess and ensure data quality, and indicate the appropriate uses and limitations of the data.
The dataset custodian is responsible for:
building the dataset (primary or secondary data entry);
maintaining the dataset (up-date and additions);
ensuring quality (data validation, error correction);
ensuring integrity (back-up and security precautions);
providing (and controlling) access;
providing directory level metadata to a central or shared metadatabase;
maintaining full and accurate additional data needed to correctly use the data;
providing advice on appropriate uses and interpretation of the dataset;
providing for the archiving of the dataset.
"Controlling access" may include limiting access to only authorised users, and/or imposing usage charges.
Users in turn have the responsibilities to:
use the data for appropriate applications;
observe any conditions on use agreed with the custodian;
report observed errors and problems to the custodian;
not re-distribute the dataset without the authority to do so
acknowledge the original source(s)
not maintain modified copies or alternative "versions" of the dataset
The custodianship concept is designed to ensure the availability of the highest quality data, while reducing redundancy of data collection and maintenance - such as the maintenance of multiple copies or various "versions" of the same data. It thus serves the dual purpose of promoting quality and cost-efficiency.
The most valuable aspect of designated custodianship is clarity of roles - especially with regard to data input and update. Some form of active "management" of the custodianship functions must be put in place, including the identification and designation of custodians, and monitoring of their conforming to the agreed principles.
It is not unusual to find access controls where the custodial organisation (or groups or individuals within it) have reservations about making the data widely available. However the basic principles and purpose of GTOS make it imperative that custodians do not impose unnecessary conditions to inhibit open exchange i.e. they abide by the GTOS Policy on Data Access and Release (see 6.1).
A custodian may choose to delegate some of the specified functions - such as archiving or day-to-day operational management of the data - but the custodian must maintain the responsibility for these activities by setting standards and monitoring compliance.
Custodian roles within GTOS will be undertaken by at least some of the proposed Data Centres (though it should be noted that carrying out analysis is not a basic function of a custodian). As discussed above (see 3.2), such Centres will have varying levels of capability and facilities and so delegation may be necessary. For instance, a Centre may not operate a shared metadatabase or provide full archiving functions but would delegate responsibility to one which does have the capability in question. This is in line with the possibility of designating different categories of DC.
5.3 Actions
Develop detailed minimum requirements for a GTOS dataset custodian in the form of a proforma custodianship agreement
Ensure the defined core datasets have identified custodians
Establish procedures for assigning, managing and reviewing custodians.
