You can do this by simple adding a text block: these are interpreted as rst, and thus you can add seealso. However, a very nice feature would be to have an optional automatic seealso, that is generated for the examples that contain the highest fraction of symbols in common (from an information retrieval, the right thing to do would be to compute a tf-idf encoding of the symbol counts, and then use a l2-distance threshold or nearest neighbor on these). G