profile
viewpoint
Becky Sharp BeckySharp Tucson, AZ

BeckySharp/OdinsonWebapp 8

Odinson development webapp

enoriega/cs665 1

Repository for question answering for CSC 665

BeckySharp/data 0

Databases & Data Science Playground

BeckySharp/deep_qa 0

Using deep learning to answer Aristo's science questions

BeckySharp/eidos 0

Protests Reading system

BeckySharp/eidosstub 0

Stub from the Eidos system for use to extract things like properties...!

pull request commentlum-ai/odinson

Promotion + matches from State are StateMatches

per discussion extractNoState should still promote but query not

BeckySharp

comment created time in 5 days

PR opened lum-ai/odinson

Reviewers
Promotion + matches from State are StateMatches

This is an attempt at a simple "for now" solution to argument promotion into the state that is compatible with the current version of the State (i.e., trying to handle merge conflicts).

Code from #182 included that makes the matches coming out of the state all be StateMatches and that fixes some bugs in the passing of the promotion params.

Undecided are:

  • do we expect promotion when we call extractNoState ? meaning, do we expect to be given the args in the results too? or only with the mentions?
  • do we expect promotion when we call query

Currently, we don't promote in either situation, but that can be discussed for sure.

Likely this is only semi-compatible with the SQL state, but we can address that in many ways (esp if it's moving to a plugin)

+455 -324

0 comment

14 changed files

pr created time in 6 days

push eventlum-ai/odinson

bsharpataz

commit sha a0d1faf5c5770d2d15d84161d2ec60fc0c0548e0

remove redundant check

view details

push time in 6 days

create barnchlum-ai/odinson

branch : promotion_dev

created branch time in 6 days

delete branch WorldModelers/Ontologies

delete branch : rename_duplicate_nodes

delete time in 12 days

push eventWorldModelers/Ontologies

Yee Seng Chan

commit sha cf46e80199a899b237975db1b00364ab21e03eb7

rename nodes with the same name

view details

Becky Sharp

commit sha c4990b5e544ff60e4fa53dc630250c4467ddd540

Merge pull request #93 from WorldModelers/rename_duplicate_nodes rename nodes with the same name

view details

push time in 12 days

PR merged WorldModelers/Ontologies

rename nodes with the same name

Changes according to an email I sent on 10/07 and Robyn's reply on 10/08. Thanks!

+16 -16

0 comment

2 changed files

chanys

pr closed time in 12 days

PullRequestReviewEvent

issue commentclulab/eidos

MockGlove

I think i did a millenia ago :) but "anything I can do* you can do better*, you can do anything better than me" (some words altered for the sake of the message) https://www.youtube.com/watch?v=WO23WBji_Z0

:D I LOVE this team -- seriously -- if anyone ever has the chance to work with you turkeys, they should snatch it up!!!

On Wed, Oct 7, 2020 at 1:51 PM Keith Alcock notifications@github.com wrote:

External Email

This is Becky's idea, but she shouldn't always have to make her own.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/clulab/eidos/issues/911#issuecomment-705186432, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCK63URY3VPBKFGCFFLSJTID3ANCNFSM4SH3RQCQ .

kwalcock

comment created time in 12 days

create barnchclulab/eidos

branch : debug_srl

created branch time in 13 days

PR closed clulab/tomcat-speech

Reviewers
added comment

not done yet...

+13 -2

0 comment

2 changed files

BeckySharp

pr closed time in 14 days

pull request commentclulab/eidos

SRL grounding

I think it will improve once we don't need to reannotate too. any chance you know the hot methods?

On Mon, Oct 5, 2020 at 11:52 AM Keith Alcock notifications@github.com wrote:

External Email

Runtime on four files without the SRL grounding is 11:40 minutes. With the grounding it is 31:00 minutes. This seems to be a very expensive grounding. It will improve some with parallel dynet.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/clulab/eidos/pull/909#issuecomment-703822264, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCIO3CT4NF6KI34IYV3SJIIXVANCNFSM4SFBRMUA .

kwalcock

comment created time in 14 days

issue openedlum-ai/odinson

Add and class level comments

describing the design choices

e.g.,

  • that Matches are internal, and that they're really there for debug
  • that StateMatches don't need to know what they used to be, if someone want to debug they can execute query instead of extractMentions
  • other things

created time in 15 days

issue openedlum-ai/odinson

Add info about trigger to SQL state

To facilitate remembering that a Match was originally an EventMatch (i.e., has a trigger, etc)

created time in 15 days

issue commentclulab/eidos

SRLCompositionalGrounder crashes

also i’m not looking at the code, but i thought i switched to cluprocessor, right? or am i crazy?

On Thursday, October 1, 2020, Mihai Surdeanu notifications@github.com wrote:

External Email

Oh, I think I know what's going on. Hang on.

On Thu, Oct 1, 2020 at 17:08 Keith Alcock notifications@github.com wrote:

It may be that I am being deceived, but it doesn't seem to have worked. I see things like words = String[30] and dependencies.outgoingEdges = Tuple2[29] and preferredSize = None. The processor in question is a FastNLPProcessorWithSemanticRoles and it appears that the DirectedGraphs are initialized along a different route than what was changed. They come from FastNLPProcessor.parseWithStanford which calls CoreNLPUtils.toDirectedGraph(basicDeps, in) which makes a DirectedGraph without the preferredSize. In IntelliJ you can right click on the constructor for DirectedGraph and select "Find Usages". "New instance creation" is probably the right category to double check. I'm not sure which all of them are necessary to convert, but at least one is missing the new argument.

— You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub https://github.com/clulab/eidos/issues/907#issuecomment-702459177, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ AAI75TUVUSDX3Y5M374QXSLSIUKWBANCNFSM4R7ZQXWA> .

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/clulab/eidos/issues/907#issuecomment-702470650, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCPOUUEDTYZP3B52THDSIUP6HANCNFSM4R7ZQXWA .

kwalcock

comment created time in 19 days

issue commentclulab/eidos

SRLCompositionalGrounder crashes

though fwiw you may want to add a helper method to return Array.empty when someone tries to get the outgoing for a SRL beyond the array...? they could all point to the same Array.empty

On Thu, Oct 1, 2020 at 9:49 AM Becky Sharp bsharpataz@gmail.com wrote:

OK, i can check when i get back if this isn't solved yet -- and no, Mihai, these are the syntax deps, not the SRLs, so no changes are needed there

On Thu, Oct 1, 2020 at 8:33 AM Keith Alcock notifications@github.com wrote:

External Email

BTW, thanks @BeckySharp https://github.com/BeckySharp for so cleanly adding things like IndividualGrounding to the messy OntologyGrounder.scala code. The JSONLD deserialization is ready and the present issue was only found while I was giving the serialization a good workout.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/clulab/eidos/issues/907#issuecomment-702217544, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCOHFZZWLFNVBAWXUQTSISOLJANCNFSM4R7ZQXWA .

kwalcock

comment created time in 20 days

issue commentclulab/eidos

SRLCompositionalGrounder crashes

OK, i can check when i get back if this isn't solved yet -- and no, Mihai, these are the syntax deps, not the SRLs, so no changes are needed there

On Thu, Oct 1, 2020 at 8:33 AM Keith Alcock notifications@github.com wrote:

External Email

BTW, thanks @BeckySharp https://github.com/BeckySharp for so cleanly adding things like IndividualGrounding to the messy OntologyGrounder.scala code. The JSONLD deserialization is ready and the present issue was only found while I was giving the serialization a good workout.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/clulab/eidos/issues/907#issuecomment-702217544, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCOHFZZWLFNVBAWXUQTSISOLJANCNFSM4R7ZQXWA .

kwalcock

comment created time in 20 days

issue commentclulab/eidos

SRLCompositionalGrounder crashes

the dependencies array isn't shorter like the srl one, right @MihaiSurdeanu ?

kwalcock

comment created time in 20 days

PR opened lum-ai/odinson

State v2

opening for "sharing"/facilitation

+915 -553

0 comment

22 changed files

pr created time in 22 days

pull request commentlum-ai/odinson

Travis CI changes

my accounts are mostly second hand, through people trying to use tools and having java 11 and then they downgrade to 8 and it works. @kwalcock @MihaiSurdeanu @marcovzla do any of you remember this?

myedibleenso

comment created time in 23 days

issue closedWorldModelers/Ontologies

Add edits from the semi-automated process/tool

from the edits branch, probably wait until the edits from the experiment are in (or at least the first pass), i.e., issue #38

closed time in 25 days

BeckySharp

issue closedWorldModelers/Ontologies

roads and bridges

https://github.com/WorldModelers/Ontologies/blob/45d9aa80ffcff538dd45f1ff5489ce08072de733/CompositionalOntology_v2.1_metadata.yml#L914

it seems to me this category is too broad. do we need to separate railways from roads? and shouldn't we add bridges under this concept?

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

roads and bridges

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

group of people

https://github.com/WorldModelers/Ontologies/blob/25690a258d02fdf1f35ce9140f7cd54145e2b30c/CompositionalOntology_v2.1_metadata.yml#L1559

so, is a group of people a "community" or a "person_or_group"? maybe "community" is_a "person_or_group"?

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

group of people

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

bad examples

https://github.com/WorldModelers/Ontologies/blob/25690a258d02fdf1f35ce9140f7cd54145e2b30c/CompositionalOntology_v2.1_metadata.yml#L1608

"lean" and "crop" must be "lean season" and "crop season", i suppose? (is there such a thing as "crop season"??)

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

bad examples

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

suggest merging

https://github.com/WorldModelers/Ontologies/blob/25690a258d02fdf1f35ce9140f7cd54145e2b30c/CompositionalOntology_v2.1_metadata.yml#L1526

is there a conceptual distinction between "safety" and "security"? i think these should be merged into a single property. the words themselves are (nearly) synonymous.

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

suggest merging

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

the art and practice of gardening

https://github.com/WorldModelers/Ontologies/blob/25690a258d02fdf1f35ce9140f7cd54145e2b30c/CompositionalOntology_v2.1_metadata.yml#L23

How is "horticulture" like "beets" and "carrots"??? I suggest removing "horticulture" as an example, and renaming the type to "vegetables".

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

the art and practice of gardening

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

refactor

https://github.com/WorldModelers/Ontologies/blob/45d9aa80ffcff538dd45f1ff5489ce08072de733/CompositionalOntology_v2.1.yml#L90

{COVID, malaria, flu} IS_A disease

case_volume and vector are not diseases, so they belong in the same brach as the other 3.

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

refactor

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

too specific

https://github.com/WorldModelers/Ontologies/blob/45d9aa80ffcff538dd45f1ff5489ce08072de733/CompositionalOntology_v2.1.yml#L124

can someone explain why we need to make the distinction between post-harvest infrastructure and other infrastructure? i'm not sure that distinction is easy to make (perhaps there are some pieces of infrastructure only used in that stage, but, if so, the distinction is inherent in the concept, we don't need to make it explicit in the ontology). presumably a warehouse can be used just as well in post-harvest and in other stages of production.

btw, "packaging" is not really an example of infrastructure, but a specific process done in the post-harvest stage. there may be specific facilities for packing/packaging, but i wouldn't know what they would be called.

as a general comment, for these obscure concepts we need domain specialists to give us a better idea of what vocabulary might be mapped to them. i would also like to see examples of real usage. i find the current examples seriously insufficient (more often than not just a spelling out of the concept name).

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

too specific

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

bad example

https://github.com/WorldModelers/Ontologies/blob/45d9aa80ffcff538dd45f1ff5489ce08072de733/CompositionalOntology_v2.1_metadata.yml#L853

perhaps we need a process type for "rehabilitation", but "school rehabilitation" is not an example of an educational facility.

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

bad example

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

food riots?

https://github.com/WorldModelers/Ontologies/blob/45d9aa80ffcff538dd45f1ff5489ce08072de733/CompositionalOntology_v2.1_metadata.yml#L686

what do "food riots" have to do with "malnutrition"?!?

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

food riots?

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

alcohol & drugs (& rock 'n roll?)

https://github.com/WorldModelers/Ontologies/blob/45d9aa80ffcff538dd45f1ff5489ce08072de733/CompositionalOntology_v2.1_metadata.yml#L718

maybe "alcoholism", but clearly not just "alcohol", no?

"drugs" by itself also shouldn't map to "substance abuse".

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

alcohol & drugs (& rock 'n roll?)

closed with #83

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

upsurge

https://github.com/WorldModelers/Ontologies/blob/45d9aa80ffcff538dd45f1ff5489ce08072de733/CompositionalOntology_v2.1_metadata.yml#L111

an "upsurge" is an "increase", not a large group of insects.

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

upsurge

closed with #83

lgalescu

comment created time in 25 days

delete branch WorldModelers/Ontologies

delete branch : more_edits

delete time in 25 days

push eventWorldModelers/Ontologies

zupon

commit sha 025ad29f4b3404cb78e59a57e7dec320d867f130

addressed github issues for ontology

view details

zupon

commit sha fa24346ba490f340d89bc2b43c1b357ef5590625

broke up compound concepts in compositional ontology

view details

zupon

commit sha e5f379c46bee87555daccd0968e69df9fa3a05d4

edits based on github thread

view details

zupon

commit sha b5044072d1d8bf817e514439d725bc70a608fb98

add welfare concept node

view details

bsharpataz

commit sha dc4cbe955011d81aa3ec1c8465f2578dcc73311c

addressed issue #84

view details

bsharpataz

commit sha 68154cba1b522e4fd530c74662b43c598dbcbdc3

merging farm and post-harvest infrastructure

view details

bsharpataz

commit sha 807d123902d01c0ac9ae8e7c506fded0e65cabfc

combine transportation infrastructure, split road and rail

view details

bsharpataz

commit sha 387dda2250e470b7fe6ec030b97160f6f8e72107

food riots

view details

bsharpataz

commit sha b9107bdf3cf33ce41e9ee990ce35eb7e68009cf4

drug and alcohol abuse

view details

bsharpataz

commit sha 49c3f0f3297b115cc01e658f7fc388f85c7b83de

locust upsurge

view details

Becky Sharp

commit sha 081931695b062a0d5cc43994afd78a0eca01055a

Merge branch 'master' into more_edits

view details

Becky Sharp

commit sha d916699506c99802e6f8153b8ee6dc5870c6840b

Merge pull request #83 from WorldModelers/more_edits More edits

view details

push time in 25 days

PR merged WorldModelers/Ontologies

Reviewers
More edits

Addresses issues raised about the compositional ontology AND breaks up compositional concepts into their respective branches.

Some remaining questions:

  1. food_preference and food_intake nodes: There's already a food CONCEPT, so I removed "food" from the names/examples of these nodes. Is that alright, or are these something more specific where we do want to keep "food" involved?
  2. living_condition: condition has been added as a PROPERTY, but after removing it from this node I'm not sure what's left. Is a life CONCEPT alright?
  3. economic_development and economic_crisis: Not sure how to separate out development and crisis here.
  4. physical_insecurity: There's an insecurity PROPERTY, but if we get rid of that here I'm not sure what we're left with. Maybe that same potential life CONCEPT?

Closes #74, #77, #78, #79, #81, #84, #86, #87, #88, #89, #90, #91

+196 -174

7 comments

2 changed files

zupon

pr closed time in 25 days

issue closedWorldModelers/Ontologies

spelling error

https://github.com/WorldModelers/Ontologies/blob/25690a258d02fdf1f35ce9140f7cd54145e2b30c/CompositionalOntology_v2.1_metadata.yml#L1365

should be "emigrate"

closed time in 25 days

lgalescu

pull request commentWorldModelers/Ontologies

More edits

Thanks @zupon and @lgalescu for the edits and comments! As soon as the tests pass I'll merge it and we can roll with this next version!

zupon

comment created time in 25 days

push eventWorldModelers/Ontologies

Ben Gyori

commit sha aa29cc9c05c8a4f220dfdc38bfd83ecb2e06a89f

Fix infrastructure typo

view details

Becky Sharp

commit sha 534fa925fc7f25ad6a118d3a7a8673502a931b27

Merge pull request #80 from WorldModelers/fix_infrastructure Fix infrastructure typo

view details

Becky Sharp

commit sha 45d9aa80ffcff538dd45f1ff5489ce08072de733

Merge pull request #82 from WorldModelers/edits some edits, not all

view details

Becky Sharp

commit sha 081931695b062a0d5cc43994afd78a0eca01055a

Merge branch 'master' into more_edits

view details

push time in 25 days

PullRequestReviewEvent

push eventWorldModelers/Ontologies

bsharpataz

commit sha dc4cbe955011d81aa3ec1c8465f2578dcc73311c

addressed issue #84

view details

bsharpataz

commit sha 68154cba1b522e4fd530c74662b43c598dbcbdc3

merging farm and post-harvest infrastructure

view details

bsharpataz

commit sha 807d123902d01c0ac9ae8e7c506fded0e65cabfc

combine transportation infrastructure, split road and rail

view details

bsharpataz

commit sha 387dda2250e470b7fe6ec030b97160f6f8e72107

food riots

view details

bsharpataz

commit sha b9107bdf3cf33ce41e9ee990ce35eb7e68009cf4

drug and alcohol abuse

view details

bsharpataz

commit sha 49c3f0f3297b115cc01e658f7fc388f85c7b83de

locust upsurge

view details

push time in 25 days

issue commentWorldModelers/Ontologies

too specific

found a compromise, merged farm_infrastructure and post_harvest, as I htink we can make the other distinctions

lgalescu

comment created time in 25 days

issue commentWorldModelers/Ontologies

too specific

I agree, this is a node that Alli added as a domain expert, but i also highly doubt we can resolve the distinction.

lgalescu

comment created time in 25 days

issue closedWorldModelers/Ontologies

polarity in ont concepts

https://github.com/WorldModelers/Ontologies/blob/45d9aa80ffcff538dd45f1ff5489ce08072de733/CompositionalOntology_v2.1_metadata.yml#L48

I struggle to understand what a polarity of -1 means when attached to a concept in the ontology. can someone add to the README something about this?

i do understand the concept in extractions (as an aspect of the KR, not of the ontology). for example, i can say "the prices are unstable" and represent that as either using property/stability with polarity -1 or using property/instability with polarity 1 (and for "stable" i would use the same concepts but reversing polarities). in this case the polarity reflects a relation between a particular concept and its realization in an utterance. but in the ontology the latter is missing, which is why i struggle to understand what the polarity sign means there.

unless there's a good reason to keep the polarity in there, i would advocate for its removal.

closed time in 25 days

lgalescu

issue commentWorldModelers/Ontologies

polarity in ont concepts

ok, closing for now, but when you want to ask again we can reopen :D

lgalescu

comment created time in 25 days

create barnchclulab/tomcat-speech

branch : for-becky-edits

created branch time in a month

issue commentWorldModelers/Ontologies

polarity in ont concepts

:)

On Thu, Sep 24, 2020 at 10:32 PM lgalescu notifications@github.com wrote:

External Email

ok, thanks for clearing things up. i know i asked before, but this polarity thing just doesn't seem to stick. i may ask again a couple months hence.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WorldModelers/Ontologies/issues/85#issuecomment-698729012, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCM4KRBUHSRGRXGJ24LSHQTPPANCNFSM4RYZO4BA .

lgalescu

comment created time in a month

issue commentWorldModelers/Ontologies

polarity in ont concepts

so it's just a signal for how these concepts get affected by verbs such as "worsen"?

yes

then why does a "solitary locust" get polarity -1? can you worsen a solitary locust? and does that mean it's increased? a single locust can only increase in size!

meh. we're going for pragmatic > purity, I'd argue that worsened locusts = more locusts...

[[ha! i finally found a hook to get something off my chest, not about the WM ontology, but about... i don't know, biologists' use of language? wtfh is "solitarious"?? i almost flagged that as some kind of typo, and then i googled it! turns out that hoppers can be "solitarious", the darn critters! i really think biologists should all be forced to use just Latin -- over the last couple of years I've become convinced that English is simply too much for them.]]

haha, I agree! :)

lgalescu

comment created time in a month

issue commentWorldModelers/Ontologies

polarity in ont concepts

and we're definitely using it, and I am pretty sure that @bgyori is working on incorporating it in INDRA (if he hasn't already).

lgalescu

comment created time in a month

issue commentWorldModelers/Ontologies

polarity in ont concepts

I can add something to the readme (and I'll leave this issue open to ensure I do) but, the basic idea is that the polarity encodes some sense of the inherent "goodness" of the concepts. The only use for it (and AFAIK eidos is the only reader using it) is to resolve extractions where the final representation depends on an interaction between this inherent "polarity" and the inherent sentiment/polarity of a verb. This is the example I use to explain it:

To account for this:

  • X worsens drought ==> X increases/promotes dought
  • X worsens wages ==> X decreases/inhibits wages

Essentially, the final polarity of the “node” to be represented in the UI/used in modeling can only be resolved through the combination of the polarity of the concept (here in the ontology) and the polarity of the edge (e.g., "worsen" = NEG and "improved" = POS). Note: as Lucian observed, other discrete values for polarity are possible; they will be added as needed

lgalescu

comment created time in a month

pull request commentlum-ai/odinson

Save state

yes, that's good. I think this branch is going to be the "working branch" for the state

On Thu, Sep 24, 2020 at 8:42 PM Keith Alcock notifications@github.com wrote:

External Email

I can rename if this is the place to do it. How about stateV2?

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/lum-ai/odinson/pull/205#issuecomment-698702842, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCMG4SMSKQOTORLCIVLSHQGS7ANCNFSM4RT3LILA .

BeckySharp

comment created time in a month

delete branch lum-ai/odinson

delete branch : save_state

delete time in a month

push eventlum-ai/odinson

Becky Sharp

commit sha efb53be23e3836a9af7d528d2aed5c74f39dcf21

Save state (#205) * wip * wip * made with state and noState versions of extract * add getAllResults * helper method * update tests and backend for iterator * Remove state factory, ExtractorEngine manages ExtractorEngine needs to be closed now, methods in State trait for save, clear, close, and in the Object for load. * moving state things around * state-specific blocks in conf * removed resultitem and statefactory * mentionFactory makes the mentionIterators * Removed ResultItem, States all use Mentions ExtractorEngine adds Mentions to the State * changes to tests * Persist SQL state (#208) Co-authored-by: Keith Alcock <github@keithalcock.com>

view details

push time in a month

PR merged lum-ai/odinson

Reviewers
Save state

As with previous PRs and the discussion, there the engine adds the mentions to its state, the State lasts the life of the Engine, and optionally can be persisted or dumped.

  • Not all methods are in place (e.g., some SQLState methods and the jsonlines serialization of Mentions)
  • I'd like someone to check/help write tests for the Ordering adaptation for Mentions vs ResultsItems in MemoryState (for the TreeSet)
  • I removed the StateFactory (didn't seem necessary)
  • I put the ResultsIterator in the MentionFactory, since the Iterator creation is where the Mentions get built, so the MentionFactory seemed like a good place to do it
  • I passed the IndexSearcher to SQLState. This is because the LazyIdGetter needs access to the IndexSearcher, and that's needed to make Mentions, and the SQLState essentially re-makes mentions when asked to "getAll/Some"

Please review and we can discuss!! :)

+905 -565

1 comment

22 changed files

BeckySharp

pr closed time in a month

pull request commentlum-ai/odinson

Save state

for the sake of a single state branch, merging

BeckySharp

comment created time in a month

delete branch lum-ai/odinson

delete branch : kwalcock-iterateOnSql

delete time in a month

push eventlum-ai/odinson

Keith Alcock

commit sha 5fb847ac3c18f25461fbd07c1e5e680dacca68ab

Persist SQL state (#208)

view details

push time in a month

PR merged lum-ai/odinson

Persist SQL state
+49 -22

2 comments

4 changed files

kwalcock

pr closed time in a month

delete branch lum-ai/odinson

delete branch : kwalcock-flatEventPromoter

delete time in a month

push eventlum-ai/odinson

Keith Alcock

commit sha d35ca62e6212db840c82aa0675a9e024cab5a2c1

Switch to FlatEventPromoter (#207) * Switch to FlatEventPromoter * Filter by promote * Fix comment

view details

push time in a month

PR merged lum-ai/odinson

Switch to FlatEventPromoter
+57 -14

4 comments

2 changed files

kwalcock

pr closed time in a month

issue commentclulab/ConceptAlignment

Cron job

good call.

On Wed, Sep 23, 2020 at 7:36 PM Keith Alcock notifications@github.com wrote:

External Email

The cron job is initiated by a human and an alarm clock. I think it is too early to automatic this. The SuperMaaS service is only available on a development machine, for instance.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/clulab/ConceptAlignment/issues/20#issuecomment-698075517, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCLI2V7E5DOJF2NKYD3SHKWARANCNFSM4RNSB3EQ .

BeckySharp

comment created time in a month

pull request commentWorldModelers/Ontologies

More edits

One more question about economic_crisis. If I split it into economy and crisis CONCEPTs, will that be a problem since both of the relevant ones are CONCEPTs? If we only take the top 1, we'd lose the other one. What if we just rename that node depression_recession and keep everything else as is? Is that cheating?

no, not a problem, at least for UAZ, others can chime in if so

zupon

comment created time in a month

pull request commentlum-ai/odinson

Persist SQL state

cool, thanks!

On Thu, Sep 24, 2020 at 8:42 AM Keith Alcock notifications@github.com wrote:

External Email

They seem to have been broken already and were going to be fixed around the same time as getAllMentions() is implemented, which is soon.

Iterator[Mention] may need to be a CloseableIterator[Mention].

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/lum-ai/odinson/pull/208#issuecomment-698426234, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCLGQP3VRCOOPFCLIC3SHNSGNANCNFSM4RYNFVIQ .

kwalcock

comment created time in a month

PullRequestReviewEvent

pull request commentlum-ai/odinson

Switch to FlatEventPromoter

I think I should merge this now, towards getting that "one true branch" for the state changes, any objections?

kwalcock

comment created time in a month

PullRequestReviewEvent

Pull request review commentlum-ai/odinson

Switch to FlatEventPromoter

 class EventPromoter {     )   } +  protected def promoteOdinsonMatch(odinsonMatch: OdinsonMatch, odinsonScoreDoc: OdinsonScoreDoc): Array[LabeledNamedOdinResults]++  protected def promoteOdinResults(odinResults: OdinResults): Array[LabeledNamedOdinResults] = {+    odinResults.scoreDocs.flatMap { odinsonScoreDoc =>+      odinsonScoreDoc.matches.flatMap { odinsonMatch =>+        promoteOdinsonMatch(odinsonMatch, odinsonScoreDoc)+      }+    }+  }+}++class FlatEventPromoter extends EventPromoter {++  protected def promoteOdinsonMatch(odinsonMatch: OdinsonMatch, odinsonScoreDoc: OdinsonScoreDoc): Array[LabeledNamedOdinResults] = {+    val result = odinsonMatch match {+      case eventMatch: EventMatch =>+        // Turn this back on for paranoia mode.+        // {+        //  val namesAndArgumentMetadatas = eventMatch.argumentMetadata.groupBy(_.name)+        //  namesAndArgumentMetadatas.foreach { case (name, argumentMetadatas) =>+        //    // Those with the same name should all be the same.+        //    argumentMetadatas.distinct.length == 1+        //  }+        // }+        val namesToPromote: Set[String] = eventMatch.argumentMetadata+            .map(_.name)

correct me if I'm wrong, but don't we need to check the .promote attribute here?

kwalcock

comment created time in a month

PullRequestReviewEvent

pull request commentWorldModelers/Ontologies

More edits

thanks @zupon !!

zupon

comment created time in a month

PullRequestReviewEvent

pull request commentWorldModelers/Ontologies

More edits

Some remaining questions:

  1. food_preference and food_intake nodes: There's already a food CONCEPT, so I removed "food" from the names/examples of these nodes. Is that alright, or are these something more specific where we do want to keep "food" involved?

nope, having just intake and preference should be fine, and I guess those should be processes (or preference a property)

  1. living_condition: condition has been added as a PROPERTY, but after removing it from this node I'm not sure what's left. Is a life CONCEPT alright?

fine with me :)

  1. economic_development and economic_crisis: Not sure how to separate out development and crisis here.

I think we can separate development and crisis and leave something like economy

  1. physical_insecurity: There's an insecurity PROPERTY, but if we get rid of that here I'm not sure what we're left with. Maybe that same potential life CONCEPT?

maybe? seems reasonable, maybe others have a better idea?

zupon

comment created time in a month

delete branch WorldModelers/Ontologies

delete branch : edits

delete time in a month

push eventWorldModelers/Ontologies

bsharpataz

commit sha f0cd203889ae1626cd9718fa71ea46bd83757c74

some edits, not all

view details

Becky Sharp

commit sha 45d9aa80ffcff538dd45f1ff5489ce08072de733

Merge pull request #82 from WorldModelers/edits some edits, not all

view details

push time in a month

PR merged WorldModelers/Ontologies

Reviewers
some edits, not all

Added some patterns, changed some examples. More edits coming, but I wanted these changes to serve as a basis for those edits, which will address the issues posted

FYI @zupon -- please use this branch as the basis (unless if gets merged, then use master)

+94 -36

2 comments

2 changed files

BeckySharp

pr closed time in a month

pull request commentWorldModelers/Ontologies

some edits, not all

@chanys I'm going to merge this bc it's primarily a change to the metadata, but we can roll it back if you don't want it, thanks!

BeckySharp

comment created time in a month

push eventclulab/eidos

bsharpataz

commit sha 6ebd4b3a674294d7ca24073197c614e119c66965

comments

view details

push time in a month

pull request commentWorldModelers/Ontologies

some edits, not all

An abbreviation for “desert locust” I think

On Tue, Sep 22, 2020 at 11:58 AM Mihai Surdeanu notifications@github.com wrote:

External Email

@MihaiSurdeanu approved this pull request.

Looks good!

In CompositionalOntology_v2.1_metadata.yml https://github.com/WorldModelers/Ontologies/pull/82#discussion_r492964289 :

@@ -113,7 +113,9 @@ name: hopper_band polarity: -1 - OntologyNode:

  •          patterns:
    
  •          pattern:
    
  •          - locust
    
  •          - \bDL\b
    

@BeckySharp https://github.com/BeckySharp : what is this?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/WorldModelers/Ontologies/pull/82#pullrequestreview-493757639, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJCPCNSMG4TKNLKKM43RKTSHDXUZANCNFSM4RWBWKRA .

BeckySharp

comment created time in a month

PR opened WorldModelers/Ontologies

Reviewers
some edits, not all

Added some patterns, changed some examples. More edits coming, but I wanted these changes to serve as a basis for those edits, which will address the issues posted

FYI @zupon -- please use this branch as the basis (unless if gets merged, then use master)

+94 -36

0 comment

2 changed files

pr created time in a month

create barnchWorldModelers/Ontologies

branch : edits

created branch time in a month

delete branch WorldModelers/Ontologies

delete branch : comp_v2_add_locusts_other_health

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : comp_v2

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : alli

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : locust_covid

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : polarity

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : new_nodes

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : wm_compositional-to-snapshot

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : fix_polarity_position

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : intervention_flattened

delete time in a month

delete branch WorldModelers/Ontologies

delete branch : bbn

delete time in a month

more