Users can currently perform a search in a given corpus, e.g. to select documents that contain a certain term.

It should be possible to generate a new corpus based on such a search, for instance by clicking a dedicated button. Such a "subcorpus" would appear as a child node of the original corpus's node, and would itself, like any other corpus, have three nodes as children:

• a Docs node, containing those documents of the original corpus that match the search;
• a Terms node, either copied from the original corpus's terms list, or computed anew based on the new set of documents (whichever to choose could be left as an option for the user);
• a Graph node.