Pablo Duboue's Talks

For some talks, I am making here available the slides. Some slides have been corrected for typos, but for the most part are exactly as presented. If you would like to see some other slides available, contact me and I will try to upload them here.

Slides from published papers

2016. On the Robustness of Standalone Referring Expression Generation Algorithms Using RDF Data

pdf

Slides for [Duboue_al.2016]

As presented at the INLG 2016 workshop WebNLG 2016, Edinburgh, UK.

2014. Filtering Personal Queries From Mixed-use Query Logs

pdf

Slides for [BressaneNeto_al.2014]

As presented in Canadian AI 2014, Montreal, QC.

2013. On the Feasibility of Automatically Describing n-dimensional Objects

pdf

Slides for [Duboue.13a]

As presented (poster) in EWNLG 2013, Sofia, BG.

2012. On The Feasibility of Open Domain Referring Expression Generation Using Large Scale Folksonomies

pdf

Slides for [Pacheco_al.12]

As presented in NAACL-HLT 2012, Montreal, QC.

2012. Extractive email thread summarization: Can we do better than He Said She Said?

pdf

Slides for [Duboue.12]

As presented (poster) in INLG 2012, Starved Rock, IL.

2006. IBM in TREC 2006 Enterprise Track

pdf

Slides for [Chu-Carroll_al.07]

As presented in TREC 2006, Gaithersburg, MD.

2004. Indirect Supervised Learning of Content Selection Logic

pdf

Slides for [Duboue.04]

As presented in INLG 2004, New Forest, UK.

2003. Statistical Acquisition of Content Selection Rules for Natural Language Generation

pdf

Slides for [Duboue_McKeown.03a]

As presented in EMNLP 2003, Sapporo, Japan.

2003. ProGenIE: Biographical descriptions for Intelligence Analysis

pdf ps

Slides for [Duboue_al.03]

As presented in ISI 2003, Tucson, AZ.

2002. Content Planner Construction via Evolutionary Algorithms and a Corpus-based Fitness Function

pdf

Slides for [Duboue_McKeown.02]

As presented in INLG 2002, Ramapo Mountains, NY.

2001. Empirically Estimating Order Constraints for Content Planning in Generation

pdf see also the slides (pdf) from the Columbia NLP group statistical generation day

Slides for [Duboue_McKeown.01]

As presented in ACL 2001, Toulouse, France.

Other talks

2016. Are community-built ontologies robust enough to build maintainable NLG systems?

pdf

As presented at the Interactions LLC lab, New Jersey, USA.

A retrospective of our work in referring expressions from RDF data with the FaMAF NLP group.

2015. Aprendizaje sobre Grandes Volúmenes de Datos y el Sistema Watson Jeopardy

pdf

As presented at the Facultad de Ingeniería Eléctrica - Universidad Tecnológica de Panamá, Panamá.

A retrospective of my 2014 class and my work in IBM Watson.

2015. Haskell en la Industria

pdf

As presented at the FAMAF-UNC, Córdoba, Argentina.

A small presentation in Spanish about the use of Haskell in the automated math tutoring domain, with an emphasis on its industrial features.

2015. NLG for Statistical Reports

html

As presented at the MTL Data Meetup.An intro to NLG and classic work on the topic at UdeM plus my Thoughtland work.

2014. Apache UIMA y el Sistema Watson Jeopardy!

pdf

As presented at Universidad Tecnológica Nacional, Reginal Córdoba. A recap in Spanish of the Big Data Montreal talk.

2014. Apache UIMA and the Watson Jeopardy! System

pdf

As presented at Big Data Montreal meeting 21st. In this talk, I discussed some unique features of Apache UIMA that helped the DeepQA System beat the Jeopardy! champions in 2011. Particularly, I discussed how UIMA's two running modes (batch and Asynchronous Scale-out) were key to expedite R&D efforts while allowing for answering queries in less than 5 seconds for the show.

2013. Poisson Magique project et introductión à le cadre de logiciel "Lamson"

pdf

As presented at Montreal Python meeting 42th. A quick talk on a new project with a short tutorial on the underlining framework employed. This talk was delivered half in French and English (with slides in opposite languages).

2013. Machine learning for hackers and why it matters for Free Software

pdf

As presented at Observe, Hack, Make hackercamp. A mix between the Radialpoint talk and the DebConf12 talk. This has been my best attended talk so far.

2013. 24 pull requests: what/how/why (lighting talk)

pdf

As presented at Observe, Hack, Make hackercamp. Quick overview of my participation in 2012's 24 pull requests challenge.

2013. Machine Learning in Javascript (lighting talk)

html

As presented at Observe, Hack, Make hackercamp. Quick overview of the URL Classy project.

2013. Thout: a javascript distributed content system (lighting talk)

pdf

As presented at Observe, Hack, Make hackercamp. Quick overview of the Thout project.

2013. Practical Introduction to Machine Learning

pdf

As presented at RadialPoint's community tech talk series. A short recap of my past machine learning experience. The talk was well attended and led to interesting discussion of the use of ML within RadialPoint.

2013. Hunter-Gatherer, a 1-Click Summarizer

pdf

As presented at Montreal Python meeting 35th. A short talk on a particular scientific use of python. The video for the talk has automatic sub-titles, don't miss them.

2012. Machine learning threats and opportunities for Debian and Free Software

pdf. As presented at DebConf12, Managua, Nicaragua. This talk spanned a very interesting debate at the conference (that you can see in the video).

2011. Generating dynamic prose in PHP

html5

As presented at the DevCamp Montreal event by Make Web Not War. This is the first talk about the experimental NLG framework in PHP I have been developing. The emphasis was in Open Government and Multilingual generation. Talks were only 15' long and overall the talk puzzled the audience big time. If you can't convince them, confuse them! Next time I'll go for a more abstract talk.

2011. Natural Language Toolkit

pdf

As presented at the Montreal Python meeting 18th. I have been wanting to look into NLTK for a while and the Montreal Python people gave me a great excuse to look into it in more detail. The talk includes a short tutorial on French and English language identification using n-grams.

2010. Apache UIMA and Academic Research

pdf

As presented at the Grupo de PLN FAMAF, Córdoba, Argentina.

This is a very opinionated talk regarding whether or not to use UIMA as the central back-bone in an academic research group. The overall theme in the talk is that the UIMA overhead in extra work for team members might not make it worth it. Interestingly, they decided to adopt UIMA for a spin-off sub-project. My views on this matter will probably change as UIMA simplifies itself and further mature.

2010. A Battle Cry for a System-level JVM in Debian

pdf

As presented at DebConf10, New York, NY.

This talk sparked very interesting discussion with other members of Debian-Java. Interestingly, I was for the most part convinced of not pursuing this line of work.

2010. What if.... we could automatically test our textual instructions?

pdf

As presented at DebConf10, New York, NY

This talk is the quick-off of my work on NLP-for-FLOSS work. It went well and I got a few people interested plus a pointer to very relevant data.

2010. How do Free Software ideals relate to Cloud Services?

pdf

Joint talk with Richard Darst.

As presented at BarCampNYC5, New York, NY.

These slides contain more questions than answers. We enjoyed preparing the talk very much and learned quite a bit in the process. We have an almost empty room so I guess it was the wrong venue. Worth repeating.

2010. Unstructured Information Processing with Apache UIMA

pdf

As presented at the NYC Search and Discovery Meetup, New York, NY.

These slides are a little dated and the main example contains a typo. See the FaMAF talk for an updated version.

2009. Natural Language Generation Tutorial

pdf

As presented at BarCampNYC4, New York, NY.

I really like this tutorial, it was a huge success.

2006. Procesamiento de Lenguaje Humano y Posibles Repercusiones a Niveles Laborales y Gubernamentales

pdf

As presented at Jornadas de Ciencia y Sociedad a los 50 años de FaMAF, Córdoba, Argentina.

My alma mater had its 50 anniversary and I was honored with a talk during a symposium organized as part of the festivities. This talk ( inSpanish) explains my thesis work with brand new examples in Spanish and hints about some possible societal changes due to new NLP technologies.

2004. Indirect Supervised Learning Of Content Selection Rules

pdf

As presented at the Cambridge Natural Language and Information Processing Research Group, Cambridge, UK.

These slides are for the most part my job talk and comparable to similar talks I gave at a number of other groups during 2004.

2003. L'Apprenant de Schemas pour la Structuration Rhétorique du Génération Automatique du Textes

pdf

As presented at LIMSI, Orsay, France.

These slides are in French, wrapping up the EMNLP 2003 slides.