The coordinates look as if you've not moved the camera at all, which is why they look like camera space. That method projects the 2D coordinate to a 3D point on the view plane, i.e. on z=1 in camera space. This point will then be transformed into world space. If the camera hasn't moved, both spaces are equal.
However, it's unlikely that you want to move in view plane.
Like i did in the other thread that is currently dealing with this issue, i'll link to a recent thread for desktop jPCT that offers a little example at the bottom of the thread:
http://www.jpct.net/forum2/index.php/topic,2413.0.htmlThe basic idea is to calculate the position in world space on a given plane (in world space, not in camera space) based on the result from reproject2D3DWS. You can do it that way or by doing a collision detection based on this vector. I'm mentioning that in the thread too.
Hope this helps.