bgoertzel

Welcome to the Haskell-verse, my friend ;-D ... On Tue, Feb 24, 2015 at 12:46 AM, Ben Goertzel [email protected] wrote: > > IMPLICATION_LINK :: Predicate -> Predicate -> TV >...

OK, so it's clear Nil's proposal needs more explanation and more examples, yeah... On Tue, Feb 24, 2015 at 1:18 PM, Linas Vepštas [email protected] wrote: > Well, like I said:...

This very nice, elementary treatment of category theory (mostly examples, not proofs) http://www.amazon.com/Category-Theory-Sciences-David-Spivak/dp/0262028131 frequently highlights the relationship btw databases and category-theoretic constructs... I'd recommend the book to anyone who wants...

Hi Nil, I think I know how to map the current Atom hierarchy into Haskell, but I have limited time today so I can't write a long email or document...

Nil, This strategy is interesting: https://hackage.haskell.org/package/swish-0.9.1.2/docs/Swish-RDF-Datatype.html#t:RDFDatatype https://hackage.haskell.org/package/swish-0.9.1.2/docs/Swish-Datatype.html#t:Datatype On Fri, Feb 27, 2015 at 5:13 PM, ngeiswei [email protected] wrote: > Here's my first attempt using Haskell type construct 'data' > >...

But ultimately, that Swish strategy is pretty complicated and ugly After looking around a bit at what other people have done, I'm not sure your "data" approach is bad... https://hackage.haskell.org/package/snap-predicates-0.1.0/docs/Data-Predicate.html...

Nice ... Now what about variablenode ? ;) On Monday, March 9, 2015, ngeiswei [email protected] wrote: > I think I'm finally going somewhere. Looks like GADT > https://en.wikibooks.org/wiki/Haskell/GADT provides exactly...

Hmm, yes I'd be interested to see if type classes could circumvent the problem you note with VariableNodes... But I agree, best to set this aside for the moment and...

Nil, I like your suggestions... I also though agree w/ LInas's observation that your Imperative construct is basically a special case of a Sample construct... the need for which we...

To deal with n-dimensional histograms efficiently, couldn't you use cover trees to approximate n-dimensional probability distributions? This paper explains how to represent probability distributions using kd-trees, https://pdfs.semanticscholar.org/00b3/87db5659824ba80c2f83a66abaa0eb6c7500.pdf and this paper...