On Wed, Nov 2, 2016 at 1:49 AM, André Senna <[email protected]> wrote:
> Man Hin, do you think SuReal will work with this simplification? > > I am not sure... I'll find out what is minimally required for sureal to generate sentences > 2016-10-31 18:57 GMT-02:00 Linas Vepstas <[email protected]>: > >> >> >> On Fri, Oct 28, 2016 at 9:15 PM, Ben Goertzel <[email protected]> wrote: >>> >>> >>> Be aware however that parsing Simple English Wikipedia currently >>> results in a lot of Atoms, i.e. way more than you're gonna fit in RAM >>> one one machine unless you have a supercomputer... >>> >> >> Really? That's wrong. I am guessing that is because every single >> sentence was stored, and that is a mistake: instead, we should extract >> plausible meanings for words, and keep only those, and discard the >> individual words. >> >> I discovered -- what maybe 5 or more years ago -- that the LG disjuncts >> correlate very well with word-meanings. Thus, I would recommend the >> following "beginner" database: parse SEW, assign to each word its >> "meaning" , i.e. its disjunct, and then keep only the network of these word >> "meanings". Discard all the individual word instances. >> >> I bet the result of that would fit in 8GB, and would be an adequate rough >> cut as input to PLN. I doubt that it would be any worse, in terms of data >> quality, than, say, WordNet or FrameNet. It would probably be higher >> quality, is my guess, based on previously fumbling with this stuff. >> >> --linas >> >> >> > > > -- > Andre Senna > IGENESIS - http://www.igenesis.com.br > (31) 9120 3292 > -- You received this message because you are subscribed to the Google Groups "opencog" 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/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAMfi0BLFeqRQsB66F6oOC_7ecmDxOV%3D0-K-Pbc3-%2B7pMFWJdwA%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
