Hi,
On 02/11/2017 10:18 PM, Ben Goertzel wrote:
I'm not familiar enough with the pattern matcher yet to suggest a Scheme
API. I do believe I know however the next steps to clean it up. So I'll look
into that first, create a github issue, then look into the API design.
Thanks, that will be helpful!
https://github.com/opencog/opencog/issues/2615
it is far from being complete but it's a start, and probably enough to
keep someone busy for a while. Happy to provide more details if needed.
Maybe we can try, first of all, a wiki page for summarizing the
near-term examples for which we'd like to see pattern mining work...
E.g. on my radar for this year we have (in no particular order)
-- Mining of inference histories, for inference control
-- Mining of dialogue histories, for learning dialogue patterns (or
more generally, verbal/nonverbal interaction patterns)
-- Mining of sets of genomic datasets or medical patient records, to
find surprisingly common combinations of features
-- Mining of surprising combinations of visual features in the output
of a relatively "disentangled" deep NN (such as the
pyramid-of-InfoGANs that Ralf, Selameab, Tesfa, Yenat and I are
working on)
-- Mining of surprising combinations of semantic relationships, in the
R2L output of a large number of simple sentences read into Atomspace
-- Mining of surprising combinations of syntactic relationships, in an
Atomspace containing a set of syntactic relationships corresponding to
each word in the dictionary of a given language (to be done
iteratively within the language learning algorithm Linas is
implementing)
-- Mining of surprising (link-parser link combination,
Lojban-Atomese-output combination) pairs, in a corpus of (link parses,
Lojban-Atomese outputs) obtained from a parallel (English, Lojban)
corpus
Thanks. I've added that to pattern miner wiki page
http://wiki.opencog.org/w/Pattern_miner#Use_cases
Personally, I do would like to see an API that facilitates interaction with
the URE. I don't really have concrete examples, just an idea.
Hmmm...
Well one could say there is a rule
FindSignificantPatterns
whose input is
-- a template T restricting what kinds of patterns to look for
-- a GroundedSchemaNode containing the significance measure one wants to use
and whose output is, say, a SetLink containing the most significant
patterns found
This is a pretty computationally expensive inference rule though ;)
True, but so is any evidence-based/direct-computation rule producing
ImpliciationLinks and such. The thing is the pattern miners has a set of
constraints, that could perhaps be seen as preconditions by the BC, to
avoid generating meaningless candidates, such as those with too little
support, etc. Just an idea...
Nil
But maybe you're thinking of something else...
-- Ben
--
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 opencog+unsubscr...@googlegroups.com.
To post to this group, send email to opencog@googlegroups.com.
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/e98b76bc-44f7-a397-bd20-1ffc4595e531%40gmail.com.
For more options, visit https://groups.google.com/d/optout.