Really interesting. I'm approaching a similar problem from the similar side. I start with a parsing algorithm, and extend it to support left side conjunctions and right side disjunctions. This extension makes the rules equivalent to normalized sequents known from sequent calculus. The result is very interesting: any logical formula can be converted to these normalized sequents suitable for (enhanced) parsing. All rules then become:
A1 /\ A2 /\ A3 /\ ... -> B1 \/ B2 \/ B3 \/ ... The parsing process then becomes equivalent to the logical abduction process. We start from the `Goal` atom and chain rules backwards to verify if the input string matches the rules. I like to think that this system resembles a kind of URE on steroids. pet, 2. tra 2021. u 04:27 Linas Vepstas <[email protected]> napisao je: > Not everyone is aware that Link Grammar's underlying theory is far more > broad than simply describing natural language. It would seem that the > theory generalizes to one of describing graphs, in general (and knowledge > graphs, in particular). > > I just updated the wiki page https://wiki.opencog.org/w/Link_Grammar to > give a brief sketch of how this theory generalizes, and how it influences > the AtomSpace design. The generalization is reviewed in > https://wiki.opencog.org/w/Connectors_and_Sections > > If you've heard these ideas before, then this page "doesn't say anything > new"; it just reminds you of the names of the Atoms that implement the > various concepts, such as connectors and disjuncts (connector sequences). > It also points to the half-dozen PDF's that justify and articulate this > generalization. > > If you haven't heard of these ideas before... well, enjoy! Just be aware > that the ideas are deceptively simple. You might read this stuff and nod to > yourself "yes, of course, obvious" until you get to something like the > https://wiki.opencog.org/w/CrossSection and wonder "what the heck?" The > apparent simplicity of these concepts makes them sometimes hard to > understand. > > -- Linas > > -- > 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 view this discussion on the web visit > https://groups.google.com/d/msgid/opencog/CAHrUA378PEv7%3Dm4A2yte-1icmL32JYrd-EndgE%2BuHGZvzAgc%3Dg%40mail.gmail.com > <https://groups.google.com/d/msgid/opencog/CAHrUA378PEv7%3Dm4A2yte-1icmL32JYrd-EndgE%2BuHGZvzAgc%3Dg%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > -- 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 view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAB5%3Dj6W5VnC-2A_k%3D3Vc2rV29%3DKuKLhddQzBXgpXsxc%3Dm%2BbuWw%40mail.gmail.com.
