HIC: What counts as "risky" access to datasets?

[szwells]

It has been suggested that queries over project datasets which return either too much or too little data can be considered risky.

• Too much data returned could indicate a user is trying to circumvent access restrictions in order to "steal" the data set -- This represents a commercial risk given the value inherent in large medical datasets
• Too little data returned could risk the identification of individuals. This is especially true where the intersection of individual queries isolates very small numbers of individuals
• Trying to link across projects shouldn't be done by researchers and should be flagged up if this happens

We are trying to identify the conditions under which particular patterns of behaviour in accessing the datasets could be recognised and flagged to the HIC data manager.

[abell-x]

Trying to link across projects shouldn't be done by researchers and should be flagged up if this happens

[szwells]

Thanks Alison, added to the list in the body of the ticket

[jamoore]

Closing all specific HIC tasks.