Support initial and final skips in skipgram training? [vote if desired]

[proycon]

Currently, Colibri Core only extracts skipgrams in which the skip is not at an initial or final position, but in the middle. For example, patterns like x {*} and {*} x are never extracted, only x {*} y. This is to keep complexity down, to find patterns left and right of x you are better of using the neigbour relations using an indexed patternmodel.

Vote for this feature here if you do want support for initial and final skips.