Re: Ann: SWI-Prolog 6.5.1

Dear Samer,

On Thu, 15 Aug 2013 09:50:33 +0000
"Abdallah, Samer" <[email protected]> wrote:

> Hi Nicos,
> Thanks for this information - I didn't realise Prism had been ported to YAP.
> I'm trying it out now!  And it will be a great help that SWI and YAP are
> already so close to each other. 
> 
	Yap also has Problog.

> Given that Prism is already in YAP and YAP has good CHR implementation, 
> I'll also have a look at combining CHR with Prism along the lines of Sneyers, 
> Meer and Vennekens' CHRiSM system. The problem I'm working on involves
> parsing a probabilistic context sensitive grammar, and the multi-headedness
> of CHR suggests a way to handle the context sensitivity.
> 
> Samer.
> 
		Do you plan to do parameter estimation (learning the probabilities from
		examples), or you just want to parse sentences from your grammar ? 

		Another alternative is Stochastic Logic Programs. 
		There exists an implementation for parameter estimation [1] but is 
		no way as complex as the systems mentioned previously in this
		thread. The nice thing about SLP is that the probability/label goes on
		the clause rather than on switches. So it feels nearer to DCGs.

		If you only need parsing and something bespoke, it is probably quite straightforward
		to right your own probabilistic CH rules [2].  

Hope it helps.

Nicos Angelopoulos


[1] http://bioinformatics.nki.nl/~n.angelopoulos/sware/slps/pe/
[2] http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.85.1986&rep=rep1&type=pdf