Hi, thanks,
I just implemented quoting the output patterns and the return AtomSpace:
1. There is a "if_quote_output_pattern" in config file, if it is set to
true, the patterns will be quoted.
2. The quote link type can be define with
"output_pattern_quoted_linktype" in the config file, the default is
MinedPatternLink.
3. The Frequency, InteractionInformation, Surpringness I and II value
are stored in the keyvalue pair of "PatternValues" of the MinedPatternLink.
Can be query by: MinedPatternLink->getValue(PatternValuesHandle).
4. Also after mining is finished, the AtomSpace contains all the
patterns can be get by the interface: getResultAtomSpace(). By AtomSpace
interface: get_handles_by_type (MINED_PATTERN_LINK ....).
For example:
(MinedPatternLink
(EvaluationLink
(PredicateNode "background")
(ListLink
(VariableNode "$var_1")
(ConceptNode "solo_singer")
)
)
(EvaluationLink
(PredicateNode "occupation")
(ListLink
(VariableNode "$var_1")
(ConceptNode "Singer-songwriter")
)
)
)
I had a quick look at the Pattern Miner clean up issue. I will try to fix
some of them by the way when I modify the related content. Of course you
are very welcome to fix things you think should be fixed. Just it would be
nice if you can notify me before you start to fix them. Because I am still
making a lot of changes, which always have a lot of conflicts with some
small changes you and Linas fixed, and because they usually occurred on a
lot of codes I had modified or just deleted, so I always cannot merge
them... Recently I just spent hours to fix all the conflicts. If they are
not very very important / core or emergent, maybe we can postpone all the
cleanups until I finish the current pattern mining work stage?
I picked some instances from the corpus you gave me early to experiment
with the new interestingness evaluation for them.
Thanks,
Shujing
On Mon, Jun 26, 2017 at 7:43 AM, Nil Geisweiller <[email protected]>
wrote:
> Shujing,
>
> On 06/25/2017 05:06 PM, Shujing Ke wrote:
>
>> My son seems better today, starts to be playful again and more willing
>> to eat and drink. But still have diarrhea.
>>
>
> Great!
>
>
>> I will implement exporting patterns first. And then the interestingness
>> evaluation inside 1-gram pattern when Nil gives the small test corpus.
>>
>
> BTW, there's gonna be a bit of time before I can hand you the corpus as
> I'm attempting to generate it from an actual inference control learning
> experiment, I'd say in a week or so.
>
> Meanwhile if you could take a look at
>
> https://github.com/opencog/opencog/issues/2787
>
> specifically the first item which is a question to you. Also, having the
> pattern miner unit test pass would be great. As I explain at the end of the
> issue, I may prefer to do these changes myself as they will help me to get
> familiar with the pattern miner code. Of course if you'd rather take care
> of them, for educational purpose or whatnot, you are free to do so as well.
>
> Nil
>
>
>> Thanks,
>> Shujing
>>
>> On Sun, Jun 25, 2017 at 9:33 AM, Ben Goertzel <[email protected]
>> <mailto:[email protected]>> wrote:
>>
>> Hi Shujing,
>>
>> > 4. About the interestingness evalution
>> >
>> > I didn't quite get the meaning of the rich(x) and z(y) and
>> married(x,y)
>> > example.
>> > I think it is also related to the pattern gram. For below 2
>> patterns: x,y,z
>> > are variables
>> > pattern A: rich(x) and z(y) and married(x,y)
>> > pattern B: rich(x) and cute(y) and married(x,y)
>> >
>> > If they are represented as 3 gram patterns, then it may be able to
>> just
>> > evaluate their interesingness by surpringness
>> > pattern A:
>> > InheritanceLink x rich
>> > InheritanceLink y z
>> > EvaluationLink married x y
>> >
>> > pattern B:
>> > InheritanceLink x rich
>> > InheritanceLink y cute
>> > EvaluationLink married x y
>> >
>> > If they are represented as 1 gram patterns, then I can implement an
>> > interestingness evalution based on the variables inside one root
>> link.
>> > pattern A:
>> > ImplicationLink
>> > AndLink
>> > InheritanceLink x rich
>> > InheritanceLink y z
>> > EvaluationLink married x y
>> >
>> > pattern B:
>> > ImplicationLink
>> > AndLink
>> > InheritanceLink x rich
>> > InheritanceLink y z
>> > EvaluationLink married x y
>>
>> Well it seems that the mathematical basis of the interestingness
>> evaluation is the same whether we view this as a 3-gram pattern or as
>> a 1-gram pattern
>>
>> So I guess the answer is that, yeah, the interestingness evaluation
>> has to be able to look at multiple variables within a "1-gram" and
>> also across different "grams" ...
>>
>> There may be surprisingness
>>
>> -- within a single "gram"
>>
>> -- among multiple "grams"
>>
>> -- involving, say, 2 variables in one gram, and then 2 variables in
>> another gram
>>
>> (where i'm using "gram" as a shorthand for "set of Atoms supervened
>> over by a single root link")
>>
>> 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 [email protected]
>> <mailto:opencog%[email protected]>.
>> To post to this group, send email to [email protected]
>> <mailto:[email protected]>.
>> Visit this group at https://groups.google.com/group/opencog
>> <https://groups.google.com/group/opencog>.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/opencog/CACYTDBe16V6TD6JC%
>> 3DQDVAd63fWpzbVe6sh3wtkqNz3bXpiYiuQ%40mail.gmail.com
>> <https://groups.google.com/d/msgid/opencog/CACYTDBe16V6TD6JC
>> %3DQDVAd63fWpzbVe6sh3wtkqNz3bXpiYiuQ%40mail.gmail.com>.
>> For more options, visit https://groups.google.com/d/optout
>> <https://groups.google.com/d/optout>.
>>
>>
>>
--
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/CALpD4-%2Bm2sVTUYAEJi%2Bx6r-kUUmrjU9_rXDmwa2cBiPB0qw0Kw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.
