Download random sample of documents

[lukavdplas]

Users often don't have the rights to download an entire corpus. Also, for a preliminary analysis or educational purposes, working with the whole data may not be desired.

However, the default sorting of corpora is not random - I think this is based on indexing order. In some cases, this may be considered random in that the factors that determine indexing order have no relation with anything you would be analysing, but it usually introduces a bias of some kind.

I think it would be nice if users could download a random sample of documents.

This requires some randomisation, which can be achieved in one of the following ways:

• Add a "random" sort option in the search results
• Add an option to the download interface where users can choose between "most relevant" or "random selection". Of course, this would only apply if the download size exceeds the number of results.

In addition, you could add the option to limit the download size.