Actions

Ontolog Forum

Revision as of 05:01, 2 March 2017 by imported>KennethBaclawski
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Session Overview
Duration 1.5 hour
Date/Time Mar 01 2017 17:30 GMT
9:30am PST/12:30pm EST
5:30pm GMT/6:30pm CET
Convener KenBaclawski

Ontology Summit 2017 Overview

Video Teleconference: https://bluejeans.com/768423137

Meeting ID: 768423137

Chat room: http://bit.ly/2lRq4h5

When you use the Video Conference URL above, you will be given the choice of using the computer audio or using your own telephone. Some attendees had difficulties when using the computer audio choice. If this happens to you, please leave the meeting and reenter it using the telephone choice. You will be given a telephone number to call along with an access code.

Please use the chat room above, not the video conference chat room. The video conference chat room is only for communicating with the moderator.

Abstract

There are many connections between ontologies, AI, machine learning and reasoning. The Ontology Summit will explore, identify and articulate the relationships between these areas, with special attention to the three track themes.

This session will be an overview of the history and current state of the relationship between AI and ontologies. The purpose of the overview is to provide background and a framework for the subsequent summit discussions. This session will begin with short presentations by members of an invited panel. This will be followed by a general discussion period.

Agenda

Abstract In this talk I will first review the Ontology Learning Layer Cake, which was defined by Philipp Cimiano and myself in 2005. In recent years my research group in Natural Language Processing at the Insight Centre for Data Analytics (National University of Ireland, Galway) has made a number of contributions to some of the layers (Terms, Concepts, Concept Hierarchy), which I will present briefly. This includes also a few developments that were not part of the original Ontology Learning Layer Cake, but that have proven to be of importance, in particular Ontology Lexicalization and Ontology Translation.

  • Discussion

Attendees

Proceedings

[12:23] AlexShkotin: Hi All!

[12:29] AndreaWesterinen: Is the presentation available?

[12:36] ToddSchneider: Ken, yes I've started the audio recording.

[12:41] KenBaclawski: @[12:29] AndreaWesterinen: The slides have not yet been posted. I will post them after the meeting.

[12:46] ToddSchneider: Paul, how much 'training' is needed for the extraction tools?

[12:48] gary berg-cross: @Todd, or me might ask, what type of "training" is used?

[12:57] gary berg-cross: Paul, what techniques are used to create these taxonomies?

[12:58] ToddSchneider: No training data?

[13:00] Chuck Rehberg: Do you use hypernymic propositions to identify concept specializations/generalizations?

[13:02] AlexShkotin: @Paul, on which layer do you have definitions for concepts and relations?

[13:03] Robert Hoehndorf: How well does this method work in other languages?

[13:07] gary berg-cross: Jose, you can type questions below the blue line rather than in the "hand" box.

[13:07] gary berg-cross: My name jose Parente de Oliveira who asked this last question

Does this only work for English or can this be generalized to other languages?

[13:08] ToddSchneider: Paul, are the 'rules' you referred to domain specific? If not, would specializing them help?

[13:09] ToddSchneider: Jose, please put your question into the box/frame at the bottom of the chat room.

[13:10] MarkUnderwood: Twitter users - follow us @Ontologys

[13:10] MarkUnderwood: Correction @OntologySummit

[13:11] TerryLongstreth: Jose's questions: could you use any sort of initial taxonomy?

My name jose Parente de Oliveira who asked this last question

Does this only work for English or can this be generalized to other languages?

How do you choose the sense of the terms to represent the concept?

If it works for English, and ontologies can be translated to other languages, can you use a data-driven (machine learning) method to learn linguistic rules to apply in other languages?

Can your manual work be formally presented as examples to train neuro network for deep learning training?

[13:12] gary berg-cross: Also from Jose: How do you choose the sense of the terms to represent the concept?

If it works for English, and ontologies can be translated to other languages, can you use a data-driven (machine learning) method to learn linguistic rules to apply in other languages?

Can your manual work be formally presented as examples to train neuro network for deep learning training?

[13:18] Paul: Johanna Voelker

[13:19] ToddSchneider: Ken, will Paul's slides be posted to the meeting page?

[13:20] AlexShkotin: http://dws.informatik.uni-mannheim.de/en/people/alumni/drjohannavoelker/

[13:21] KenBaclawski: @[13:19] ToddSchneider: Yes, I will be doing it after the meeting.

[13:28] AlexShkotin: sorry for my system. and I put one q here: "@Paul, on which layer do you have definitions for concepts and relations?"

[13:29] Paul: sorry, I have to leave now - thanks for the discussion

[13:29] MikeBennett: I have to drop off now, at least from Audio.

[13:30] AlexShkotin: any

[13:30] Ognyan Kulev morphed into OgnyanKulev

[13:31] 0x1b: Thank you everyone - enjoyed the presentation and looking forward to many more

[13:31] AndreaWesterinen: GATE is the General Architecture for Text Engineering and is described at https://gate.ac.uk/

[13:32] Max Petrenko: no sound at my end -- apologies

[13:37] MarkUnderwood: David - do u have contacts there? Perhaps you can invite them?

[13:39] gary berg-cross: Wang, Hao. "Semantic Deep Learning." (2015).  ??

[13:39] AlexShkotin: just to give an answer to Ken and All. we are developing a language to write knowledge mathematically - YAFOLL. so we do not use any AI.

[13:39] Robert Hoehndorf: http://www.cs.uoregon.edu/Reports/ORAL-201509-Wang.pdf

[13:40] gary berg-cross: http://www.cs.uoregon.edu/Reports/ORAL-201509-Wang.pdf

[13:42] TerryLongstreth: xferred from Bluejean: Deep Learning of RDFS rules Bassem Makni, James Hendler Rensselaer Polytechnic Institute

[13:43] Donna Fritzsche: welcome Bruce!

[13:43] Robert Hoehndorf: http://www.aaai.org/ocs/index.php/AAAI/AAAI14/paper/viewFile/8531/8546

[13:43] Donna Fritzsche: Ken - you are breaking up?

[13:43] Robert Hoehndorf: https://ub-madoc.bib.uni-mannheim.de/41307/1/Ristoski_RDF2Vec.pdf

[13:44] gary berg-cross: Deep Learning of RDFS rules Bassem Makni, James Hendler Rensselaer Polytechnic Institute

Semantic Deep Learning Hao Wang

Ontology-Based Deep Restricted Boltzmann Machine Hao Wang(B), Dejing Dou, and Daniel Lowd

[13:46] KenBaclawski: I am still at the meeting, but I had a network problem.

[13:46] Robert Hoehndorf: https://arxiv.org/abs/1606.04422

[13:48] ToddSchneider: Ken, we see you in the BlueJeans participant window, but your audio isn't working.

[13:50] KenBaclawski: I should be back again.

[13:55] John Bottoms: Guest lecturer, SUNY. I teach AI for VR and working toward graphic simulations using ontologies. This is a grad level course in VR and the students have worked to develop a graphic-based cognitive computing system. The linguistic interface is in Prolog and allows dynamic editing of the ontology. I'm interested in talking with others interested in graphic ontologies.

[13:56] MarkUnderwood: John - Which SUNY?

[13:56] John Bottoms: SUNY Oswego

[13:57] MarkUnderwood: Great turnout, thanks for attending all. I must run to a NIST meeting

[13:59] BruceBray: I am at University of Utah (Biomedical Informatics and Cardiology), and we have been working with Topic modeling (LDA, etc) and starting to evaluate word embedding. We are also using ontologies to help with NLP feature extraction from clinical texts. It sounds like this summit will be very interesting.

Resources

Audio Recording

Previous Meetings

... further results