Metagraphs (the ram/cpu) paper- New version available!

[Linas Vepstas]

I've just published an expanded and revised version of the Metagraphs paper; the new version is here:

https://github.com/opencog/atomspace/blob/master/opencog/sheaf/docs/ram-cpu.pdf

It begins by describing how metagraphs are minor extensions to graphs, and examines the RAM and CPU savings that result. After that, it feels like a row of dominos falling over: each result leads to the next. Lots of ink spilled on the RAM & CPU implications of SQL, GraphQL, and all those other *QL languages, as compared to metagraph queries. And what s-expressions have to do with it. Type theory, of course. And why it's important to AGI and knowledge representation.

Several people had complimented me that it was one of the better, more informative papers, and one even noted he'd assigned it to his students for reading. With comments like that, it seemed that perhaps I should review it to make sure it didn't say anything silly.  And thus a new revision.

It is a major revision, more than doubling it's size. This may have ruined the charm of the original version. Or not: the original parts are intact.

I'm cc'ing a number of people on this email who are not subscribers to this mailing list. Discussions with them helped focus assorted vague statements into something concrete. (I've even thanked them in the credits! Really, it was useful!) Those discussions were under the label of "database for s-expressions". I now realize that, well, that is not quite it, either. You'd have to read the PDF to see why.

-- linas

p.s. take pot-shots at it.  I'll spin a version 2.0.1 for typos, and a version 2.1.0 for more serious damage.

--

Patrick: Are they laughing at us?

Sponge Bob: No, Patrick, they are laughing next to us.