From OntologPSMW

Jump to: navigation, search
[ ]
Session Synthesis Session 1
Duration 1.5 hour90 minute
5,400 second
0.0625 day
Date/Time Mar 29 2017 18:30 GMT
9:30am PDT/12:30pm EDT
5:30pm BST/6:30pm CEST
Convener MikeBennett

Ontology Summit 2017 Synthesis Session 1     (2)

Chat Room     (3)

Connection Details:     (5)


Abstract     (7)

The aim of this week's session is to synthesize the lessons learned so far on the three tracks that are under way. Each track has met once, and so we will have gained insights from a combination of invited speakers, chat log comments and blog page discussions.     (7A)

A second synthesis session will take place after the three tracks have met again, and that along with today's session outcome will form the basis of this year's Ontology Summit Communique.     (7B)

Agenda     (8)

Attendees for All Sessions so far     (9)

Proceedings     (10)

[12:31] MikeBennett: We are on GoToMeeting today - please set your name there on the way in.     (10A)

[12:31] MikeBennett: We will get started shortly, once everyone has had a chance to get themselves set up.     (10B)

[12:36] Donna Fritzsche: We sent them to Ken     (10D)

[12:37] Donna Fritzsche: I will be attending the FIBO conference next week - I look forward to meeting several of you in person!     (10E)

[12:37] Mark Underwood: Is the Gotomeeting switch a one-time?     (10F)

[12:37] Mark Underwood: Donna, wish I could go . . .work at Synchrony Financial now. Maybe you can debrief a subgroup here     (10G)

[12:37] NancyWiegand: I'm in. Nancy     (10H)

[12:38] Donna Fritzsche: yes Mark - I was thinking of doing something like that. Its exciting to see it move forward.     (10I)

[12:39] Mark Underwood: Donna +1     (10J)

[12:43] Mark Underwood: Donna If u see any colleagues (we were GE Capital) have them ping me if you get a sec     (10K)

[12:44] Donna Fritzsche: @mark - will do.     (10L)

[12:44] MikeBennett: @Mark the GoToMeeting is our back-up for when the Bluejeans facility is not available.     (10M)

[12:45] NancyWiegand: Have things started? I can hear only some faint audio.     (10N)

[12:48] Donna Fritzsche: yes Nancy - are you on the GTM?     (10O)

[12:48] Donna Fritzsche: I can hear fine     (10P)

[12:50] NancyWiegand: I was on something else! I re-dialed in, and it's ok now.     (10Q)

[12:50] Donna Fritzsche:     (10R)

[12:54] Mark Underwood: Mike B:Ack'd, thanks     (10S)

[13:03] Donna Fritzsche: Ram - what is the address of NIST? is there a recommended hotel?     (10T)

[13:11] Ram D. Sriram: @Donna: Gary indicated that the Ramada Inn is close by.     (10U)

[13:11] Donna Fritzsche: Nice summary Gary!     (10V)

[13:12] Donna Fritzsche: Thanks Ram     (10W)

[13:15] gary berg-cross: @Ram it is not a Ramada I was thinking of. There is a Best Western Plus Rockville between the Rockville Metro and the NCCOE. They may provide a shuttle each direction. They say they do to the Metro.     (10X)

[13:20] gary berg-cross: Next week Track A has its 2nd session with the following speakers/topics: Michael Yu (UCSD) - "Inferring the hierarchical structure and function of a cell from millions of biological measurements".     (10Y)

Francesco Corcoglioniti (Post-doc at Fondazione Bruno Kessler, Italy) "Frame-based Ontology Population from text with PIKES" and Evangelos Pafilis (Hellenic Center Marine Research [HCMR]) - EXTRACT: interactive extraction of environment metadata and term suggestion for metagenomic sample annotation.     (10Z)

[13:44] gary berg-cross: BTW, part of the OntoHub presentation mentioned using theory proving on the ontologies, including the idea of testing for inconsistencies with axioms in new ontologies. This is an important things for Ontology Learning.     (10AA)

[13:48] gary berg-cross: From OntoHub paper "When an ontology has been analysed, Theorems are shown in its Contents area. There, the user can either choose to prove all conjectures at once or only a specific one. Either way, the next step is to configure the proof attempts..."     (10AB)

[13:48] Alan Rector: apologies. I have to leave. Alan     (10AC)

[13:48] Ram D. Sriram: @Gary: Thanks for pointing that out     (10AD)

[13:49] Donna Fritzsche: overarching framework to capture the interrelationship between all 3     (10AE)

[13:49] MikeBennett: Chicken and egg: building ontologies v using ontologies to inform learning.     (10AF)

[13:50] MikeBennett: Reasoning - building the cross connections with the use / generation of ontologies. See above ref from Gary.     (10AG)

[13:51] Donna Fritzsche: gary points out - ontology design patterns     (10AH)

[13:51] MikeBennett: Consistent KR and semantic interoperability.     (10AI)

[13:52] gary berg-cross: Here is the OntoHub ref... Applied Ontology Y (2016) 1X 1 IOS Press Ontohub: A semantic repository for heterogeneous ontologies Mihai Codescu , Eugen Kuksa , Oliver Kutz , Till Mossakowski and Fabian Neuhaus     (10AJ)

[13:52] Donna Fritzsche: self awareness of local ontologies and techniques     (10AK)

[13:53] Donna Fritzsche: meta ontology to capture the interoperability capabilities     (10AL)

[13:53] Donna Fritzsche: good ref Gary     (10AM)

[13:53] MikeBennett: for the ones that buid ontologies (Track A- NELL, FRED etc.) - how do these form coherent theories of the world? do they create upper ontologies / families of thories?     (10AN)

[13:54] MikeBennett: Answer: there is something in place whereby they see human-curated ontologies as a "Gold standard" for them to compare with.     (10AO)

[13:54] MikeBennett: Some of the knowledge extraction already looks a lot like what we would create in a curated ontology. May result from the use of probability.     (10AP)

[13:55] MikeBennett: See NELL on this (sports ontology).     (10AQ)

[13:55] Donna Fritzsche: NELL Fragment on sports/players..     (10AR)

[13:56] Ram D. Sriram: There are several dimensions we need to look at here: 1) Cognition/Discourse or something like that (NLP, Retrieval, Interoperability, etc.); 2) Applications (Medicine, Law, Engineering); 3) Ontology Spectrum Usage (Simple taxonomies Vs Logical Theories) for each of the three tracks: 1) Learning to ontologies; 2) ML using Ontologies; 3) Inferencing. We also need to make sure that there are adequate evaluation strategies.     (10AS)

[13:57] Ram D. Sriram: OOPS! The first one is Learning Ontologies and not Learning to Ontologies (but you get it).     (10AT)

[13:59] Ram D. Sriram: @Gary: I think the Industrial Foundation Ontology initiative is trying to build a large ontology for the manufacturing domain.     (10AU)

[14:00] TorstenHahmann: I'm not sure how useful "gold standards" are. First, we often don't agree on them or have any good measure for what qualifies as gold standard. But a second aspect is that in domains where we have a gold standard, there isn't much of a point of learning an ontology.     (10AV)

[14:01] MikeBennett: Add: Adequate evaluation strategies.     (10AW)

[14:01] Ram D. Sriram: @gary: My last sentence talks about evaluation strategies     (10AX)

[14:02] Mark Underwood: I had to step out, but it would be useful for some of us to refine what we need our tools to do better -- e.g., representation for the outputs of TensorFlow, or canonical descriptions of training sets     (10AY)

[14:02] Donna Fritzsche: +1 Marks point     (10AZ)

[14:03] Donna Fritzsche: better tools are required     (10AAA)

[14:03] Donna Fritzsche: better training for Knowledge Engineers - how to leverage common ontology design patterns     (10AAB)

[14:03] Mark Underwood: Or thinking as NIST might, a reference architecture for how to link ML to an ontology     (10AAC)

[14:04] MikeBennett: If the "Gold Standard" does not include the ability to express reflexivity in it's relationaship-operator, I'd have a tough time agreeing it's a Standard at all.     (10AAD)

[14:05] MikeBennett: ^^ the above comment is from Jim someone     (10AAE)

[14:06] ChristiKapp: Would "gold standards" have a tendency to coalesce around marketable/main stream concepts though and not include specialized knowledge at the fringes? For example, vehicle taxonomies tend to leave out anything related to things such as motorcycles or snowmobiles. this seems skewed by sales volumes and marketability - not knowledge     (10AAF)

[14:08] Donna Fritzsche: Ecosystem of tools, learning techniques and knowledge engineers     (10AAG)

[14:10] Donna Fritzsche: I need to sign-off, thanks to all - Donna     (10AAH)

[14:12] TorstenHahmann: Thanks, very interesting discussion. Hope we can continue that     (10AAI)

Resources     (11)

Previous Meetings     (12)