Deverbals in Portuguese? Why not?

Livy Real and I have been wondering about a short paper on deverbal nominalizations in Portuguese. The basic idea of the paper is that even a very preliminary resource on deverbal nouns and the verbs that they are related to would be useful to do knowledge representation in Portuguese. So we have a rough-and-ready translation of the the English original  NOMLEX up for improvements here.  The preliminary version of the paper, which was rejected from STIL2013 is here.

Common Sense, at last...

I always loved the name of this series of conferences CommonSense. They were somewhat inspired by John McCarthy and they have been going on since 1998.
So I was delighted earlier this year when they've invited me to their Program Committee, for the the 2013 installment that happened in Cyprus (here's the final program).

I had already asked to go for three other meetings, so I couldn't make it, but Charles Ortiz presented my paper for me and I just joined them by skype to answer questions. Very cool!

I've prepared two sets of slides for the talk, one using Powerpoint (Contexts for Quantification) and one using Latex, Contexts for Quantification. I am not satisfied with this work, yet.  So I need to keep working on the ideas...

But since Cleo asked for the  slides, here they are. Must upload them to Slideshare too.

Demonyms and other niceties...

Relating demonyms to the places they originate from is usual linguistic fare, but KR seems to play scant attention to it.

Here's a map of the world showing how the mapping goes: kind of... there are at least three mistakes in the code that generated it.
Here's the LanguageLog blog post giving due credit for the map above and explaining  the mistakes.

We need to cope with queries like "show me a French comedy", which is easy, but needs to be done.
Tina Boegel and I  never got around finishing it in the  PARC Transfer system.

 Actually Tina and I tried to do something a little more complicated, pertainyms, which include adjectives like Christian, buddhist, etc.

Happy to report that the Powerset version of the semantics has done it all, apparently.
I'm still investigating how does one add to the stuff they have in the best possible way...

Bridge and CommonSense

When I left PARC in May 2008 to try to do Cuil I realized that I didn't want to drop my research in Natural Language semantics. In particular I really wanted to carry on thinking about "proof-theoretic semantics for natural languages", but I was in a bit of a bind, given that I had no system to try experiments on (the Bridge system is proprietary to Xerox) and no people  to do this work with (friends were either in Powerset, a competitor of sorts,  or in PARC trying their own versions).
But I am nothing if not persistent (my dad would say 'teimosa') so I gave a talk about my plans to continue this kind of work at SRI (A Bridge not to far) and wrote a paper about it (Bridges from Language to Logic: Concepts, Contexts and Ontologies).

Later, together with Annie Zaenen, Cleo Condoravdi and Lauri Karttunen, I've been working (as an off campus member) of the Language and Natural Reasoning group. This is fun, we've organized  workshops and are editing a special volume of LiLT. Out of this work I have written a paper on quantification using the kind of logic discussed above, the paper Contexts for Quantification appeared in this year's CommonSense symposium.

Particles: Bah!...

I find it kind of difficult to keep two blogs, one at work, one personal...So I have not written much here. Which is just as well, as I have been behind with lots and lots of things to do and very few done. So this is not a proper post, but a 'hang-in-there' post, wait for me, I will be back. Soon. I hope.


 

Tania Rojas-Esponda gave a talk last week at Nuance on questions under discussion and particles. The abstract is below, together with Tania's  short bio, requested by me. This post is simply a page to hang her slides and paper from, before she puts them in her own website.

QUDs and S-trees: Formal tools for reading between the lines in dialogue

Abstract:

A central premise in pragmatics is that additional information can be obtained from utterances by factoring in contextual cues, such as the goals of interlocutors. I will talk about the notion of a Question under Discussion (QUD for short), a question that is seen as guiding the moves of participants in discourse. By leveraging work from the semantics of questions, we can give formal characterizations of the structure of conversation and of speaker cooperation. I will illustrate a variety of ways in which QUDs are useful. These include ambiguity resolution, inference, and new avenues for understanding non-truth-conditional expressions, such as discourse particles.

Bio:

Tania Rojas-Esponda's path started in mathematics (BA Princeton, Part III Univ. of Cambridge, MS Stanford). She later found a way to combine her formal skills with her long-standing passion for languages via the study of linguistics. Currently, Tania is a PhD candidate in the Stanford linguistics department. She has done research in semantics/pragmatics and in phonology. For her thesis, she is investigating formal approaches to understanding discourse particles.

(Cleo Condoravdi, Tania's co-advisor, joined us too)

OpenWN-PT going places...

Really glad to see that, despite its shortcomings, people are finding uses for our free, automatically created version of WordNet, OpenWN-PT described in the paper http://aclweb.org/anthology/C/C12/C12-3044.pdf.

(OpenWordNet-PT: An Open Brazilian Wordnet for Reasoning

Valeria de Paiva, Alexandre Rademaker, Gerard de Melo.)

The first printed version of the work appeared in the Global Wordnet Association meeting:
de Paiva, Valeria, and Alexandre Rademaker. 2012. “Revisiting a Brazilian WordNet”. In Proceedings of Global Wordnet Conference, Matsue, Japan. Global Wordnet Association.

The data on OpenWN-PT can be gotten from Github, http://github.com/arademaker/wordnet-br.

Meanwhile, besides appearing in Wordnets in the world  and  Open Multilingual Wordnet, the OpenWN-PT made it to FreeLing (as can be seen in http://devel.cpl.upc.edu/freeling/svn/trunk/COPYING, and as we were told by Marcos Garcia). Marcos and Pablo Gamallo also have a fairly big Portuguese dictionary incorporated to FreeLing, see their paper. 

Finally some other open resources for Portuguese (email in Nov, 30, 2012), thanks Claudia!
Onto.PT v.0.4.1: versão actual da ontologia lexical, estruturada
como uma wordnet (synsets + relações), e construída automaticamente a
partir de dicionários e tesauros para a língua portuguesa.
Para descarregar: http://ontopt.dei.uc.pt/index.php?sec=download
Para consultar: http://ontopt.dei.uc.pt/index.php?sec=consultar

- PAPEL 3.2: nova versão da rede lexical desenvolvida no âmbito da
Linguateca, com o apoio da Porto Editora.
Contém triplos semânticos na forma "palavra relação palavra" obtidos
automaticamente através de gramáticas, executadas sobre o conteúdo do
Dicionário PRO da Língua Portuguesa.
Para descarregar PAPEL e gramáticas: http://www.linguateca.pt/PAPEL/

- Redes lexicais extraídas a partir do Dicionário Aberto e do
Wikcionário.PT, seguindo o mesmo procedimento que no PAPEL.
Para descarregar: http://ontopt.dei.uc.pt/index.php?sec=recursos

NLCS in New Orleans, yeah...

Lately I have been bad at writing blog posts...
Or not writing them, as the case maybe.. Anyways I have recently got back from the Natural Language and Computer Science (NLCS) Workshop that I co-organized with Larry Moss (Indiana University), associated with Logic in Computer Science (LiCS) 2013.

The workshop turned out really well, despite some organizational hazards. As we said in the proposal and call for papers  for the workshop:

Formal tools coming from logic and category theory are important in both natural language semantics and in computational semantics. Moreover, work on these tools borrows heavily from all areas of theoretical computer science. In the other direction, applications having to do with natural language have inspired developments on the formal side. The workshop invites papers on both topics. Specific topics include, but are not limited to:

• logic for semantics of lexical items, sentences, discourse and dialog
• continuations in natural language semantics
• formal tools in textual inference, such as logics for natural language inference
• applications of category theory in semantics
• linear logic in semantics
• formal approaches to unifying data-driven and declarative approaches to semantics

Invited speakers:
Robin Cooper, University of Gothenburg, Sweden.

Ian Pratt-Hartmann, University of Manchester, UK.

Wlodek Zadrozny, UNC, Charlotte, North Carolina.

NLCS Workshop
120 Newcomb Hall
(Building 74 on the Campus Map)

Program

9:00 Valeria de Paiva, Nuance.com
Welcome [Slides]

9:10 Wlodek Zadrozny, University of North Carolina, Charlotte
After Watson [Slides]

10:10 Robin Cooper, University of Gothenburg, Sweden
Adapting Type Theory with Records for Natural Langauge Semantics [Slides]

11:20 Bruno Mery and Christian Retoré, Université de Bordeaux and IRIT, Toulouse
Advances in the Logical Representation of Lexical Semantics [Slides]

1:45 Christophe Fouqueré and Myriam Quatrini, Laboratoire d'Informatique de Paris-Nord and Institut de Mathématiques de Luminy A.N.R. LOCI
Inferences and Dialogues in Ludics [Slides]

2:30 Ian Pratt-Hartmann, University of Manchester
The Relational Syllogistic [Slides]

3:30 Alex Djalali, Stanford University
Extending a Natural Language Proof Theory: On Ordinary Comparatives [Slides]

4:40 Wren Thornton, Indiana University
Chiastic Lambda-Calculi [Slides]
5:30 Larry Moss, Indiana University
Current Work on Natural Logic [Slides]