Ken, That's on the order of what I was expecting, although it's more sophisticated than I'd figured on.
* Feature constraints sound promising, but AFAIK not implemented in any Kanrens * Your encoding is pretty darn cool, but it maybe isn't the clearest way to encode the data. The other options I'm coming up with are: * hack up the unifier and add it there * implement a feature-structure-unify as a miniKanren relation. I think the final choice is the traditional way to do this in Prolog, yes? JBH On Mon, Dec 26, 2016 at 4:44 PM, Chung-chieh Shan <[email protected]> wrote: > But we want the alists > ((a . _.0) (b . y)) > and > ((b . _.1) (a . x)) > to unify, yielding the mgu > ((a . x) (b . y)) > or equivalently > ((b . y) (a . x)) > > And we want the alists > ((a . x)) > and > ((b . y)) > to unify, yielding the same mgu as above. > > So the only way I know to encode (even unnested) feature structures > using trees is to map each feature name to a globally fixed address > in a binary tree. For example if we decide globally that "a" maps > to the address (left left right) and "b" maps to the address > (left right right left) then the alist above > ((a . _.0) (b . y)) > can be encoded by the tree > (((_.2 . _.0) . (_.3 . (y . _.4))) . _.5) > > Maybe this is what Jason has in mind? It's related to expressing > memoization as a lazy data structure. > > On 2016-12-24T12:07:14-0500, Dan Friedman wrote: > > I agree with Jason, processing alists (even split alists)i standard fare. > > Nested alists might want some constraints Haven't given that observation > > much thought. > > > > On Sat, Dec 24, 2016 at 11:08 AM, Jason Hemann <[email protected]> > > wrote: > > > I think these can be encoded using trees, yes? (Wiki seems to also > suggest > > > so, if I'm reading that correctly). I know of no implementations that > > > implement feature constraints presently, but I know those have been > investigated > > > in other contexts > > > <http://www.sciencedirect.com/science/article/pii/0743106694900442>. > > > > > > On Sat, Dec 24, 2016 at 8:06 AM, Amirouche Boubekki < > > > [email protected]> wrote: > > > > > >> Did anyone implement minikanren for feature structure > > >> <https://en.wikipedia.org/wiki/Feature_structure>? > > -- > Edit this signature at http://conway.bostoncoop.net/~ccshan/cgi-bin/sig > Information is not knowledge. Knowledge is not wisdom. Wisdom is not truth. > Truth is not beauty. Beauty is not love. Love is not music. Music is THE > BEST. > ― Frank Zappa > > -- > You received this message because you are subscribed to the Google Groups > "minikanren" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at https://groups.google.com/group/minikanren. > For more options, visit https://groups.google.com/d/optout. > -- JBH -- You received this message because you are subscribed to the Google Groups "minikanren" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/minikanren. For more options, visit https://groups.google.com/d/optout.
