Cognitive Aspects of the Lexicon (CogALex-v)

https://sites.google.com/site/cogalex2016/home

Workshop co-lated with coling (the 26th International Conference on

Computational Linguistics, Osaka, Japan), December 12, 2016

Invited speaker :  Chris Biemann (Technische Universität, Darmstadt)

We are pleased to announce the 5th Workshop on ‘Cognitive Aspects of the Lexicon’ (Cogalex-V), taking place just before coling (Osaka, Japan), december 12, 2016.

1        Context and background

The way we look at the lexicon (creation and use) has changed dramatically over the past 30 years. While in the past being considered as an appendix to grammar, the lexicon has now moved to centre stage. Indeed, there is hardly any task in NLP which can be conducted without it. Also, rather than considering it as a static entity (database view), dictionaries are now viewed as dynamic networks, akin to the human brain, whose nodes and links (connection strengths) may change over time.

Linguists work on products, while psychologists and computer scientists deal with processes. They decompose the task into a set of subtasks, i.e. modules between  which information flows. There are inputs, outputs and processes in between. A typical task in language processing is to go from meanings to sound or vice versa, the two extremes of language production and language understanding. Since this mapping is hardly ever direct, various intermediate steps or layers (syntax, morphology) are necessary.

Most of the work done by psycholinguists has dealt with the information flow from meaning (or concepts) to sound or the other way around. What has not been addressed though is the creation of a map of the mental lexicon, that is a  represention of the way how words are organized or connected.

In this respect WordNet and Roget's Thesaurus are probably closest to what one can expect these days. This being said, to find a word in a resource one has to reduce the search space (entire lexicon) and this is done via the knowledge one has at the onset of search. While the information stored in the lexicon is a product, its access is clearly a (cognitive, i.e. knowledge-based) process.

1.1      Goal

The goal of Cogalex is to provide a forum for researchers in NLP, psychologists, computational lexicographers and users of lexical resources to share their knowledge and needs concerning the construction, organization and use of a lexicon by people (lexical access) and machines (NLP, IR, data-mining).

Like in the past (2004, 2008, 2010, 2012 and 2014), we will invite researchers to address various unsolved problems, by putting this time stronger emphasis though on distributional semantics (DS). Indeed, we would like to see work showing the relevance of DS as a cognitive model of the lexicon. The interest in distributional approaches has grown considerably over the last few year, both in computational linguistics and cognitive sciences. A further boost has been provided by the recent hype around deep learning and neural embeddings. While all these approaches seem to have great potential, their added value to address cognitive and semantic aspects of the lexicon still needs to be shown.

This workshop is about possible enhancements of lexical resources and electronic dictionaries, as well as on any aspect relevant to the achieve a better understanding of the mental lexicon and semantic memory.We solicit contributions including but not limited to the topics listed here below, topics, which can be considered from any of the following points of view:

• (computational, corpus) linguistics,
• neuro- or psycholinguistics (tip of the tongue problem, associations),
• network related sciences (sociology, economy, biology),
• mathematics (vector-based approaches, graph theory, small-world problem), etc.
  
  

We also plan to organize a “friendly competition” for corpus-based models of lexical networks and navigation, i.e. lexical access (see below).

1.2      Possible Topics

1.2.1   Analysis of the conceptual input of a dictionary user

• What does a language producer start out with and how does this input relate to the target form? (meaning, collocation, topically related, etc.)
• What is in the authors' minds when they are generating a message and looking for a word?
• What does it take to bridge the gap between this input and the desired output (target word)?
• Lexical representation (holistic, decomposed)
• Meaning representation (concept based, primitives)
• Distributional semantics (count models, neural embeddings, etc. )
• Neurocomputational theories of content representation.
• Discovering structures in the lexicon: formal and semantic point of view (clustering, topical structure)
• Evolution, i.e. dynamic aspects of the lexicon (changes of weights)
• Neural models of the mental lexicon (distribution of information concerning words, organization of words)
• Manual, automatic or collaborative building of dictionaries and indexes (crowd-sourcing, serious games, etc.)
• Impact and use of social networks (Facebook, Twitter) for building dictionaries, for organizing and indexing the  data (clustering of words), and for allowing to track navigational strategies, etc.
• (Semi-) automatic induction of the link type (e.g. synonym, hypernym, meronym, association, collocation, ...)
• Use of corpora and patterns (data-mining) for getting access to words, their uses, combinations and associations
• Search based on sound, meaning or associations
• Search (simple query vs. multiple words)
• Search-space determination based on user's knowledge, meta-knowledge and cognitive state (information available at the onset, knowledge concerning the relationship between the input and the target word, ...)
• Context-dependent search (modification of users’ goals during search)
• Navigation (frequent navigational patterns or search strategies used by people)
• Interface problems, data-visualization
• Creative ways of getting access to and using word associations (reading between the lines, subliminal communication).

1.2.2   The meaning of words

1.2.3    Structure of the lexicon

1.2.4     Methods for crafting dictionaries or indexes

1.2.5     Dictionary access (navigation and search strategies), interface issues,

2        Description of the shared tasks associated with the workshop.

We plan to organize a “friendly competition” of corpus-based models of lexical access and semantic/associative relations between words.  This competition will be based on an existing, publicly available data set.  We provide an official separation of the data set into training, development and test data as well as a detailed specification of the task and evaluation metrics (implemented as easy-to-use scripts), so that the results obtained by different participants can be compared directly.

The precise design of the task has not been finalized yet, but it will be based on one or more of the following data sets:

• free association norms from the Edinburgh Associative Thesaurus (EAT)
• free association norms from the University of South Florida (USF)
• prime-target pairs from the Semantic Priming Project (SPP)
• semantically related word pairs from EVALution 1.0 (https://github.com/esantus/EVALution)

3        INVITED SPEAKER

Chris Biemann, leader of the LT research group in Darmstadt, and well known for his work on graph-based-approaches for NLP, has kindly accepted to give the invited presentation.

4        Deadlines.

• September 25:   Submission deadline
• October 16:       Author notification
• October 30:       Camera ready due by Authors
• November 6:     Proceedings due by Workshop Organisers to Workshop & Publication Chairs.
• December 12 :   Workshop

5        Submission

The submissions should be written in English and be anonymized for review. They must comply with the style-sheets provided by Coling: http://coling2016.anlp.jp/#instructions

• Long papers may consist of 8 pages of content, plus 2 pages for references;
• Short paper may consist of up to 4 pages of content, plus 2 pages for references
• The respective final versions may be up to 9 pages for long papers and 5 pages for short ones. In both cases the number of pages for references is limited to 3 pages

Papers should be in PDF format and have to be submitted electronically via the START submission system (https://www.softconf.com/coling2016/ CogALex-V/). You probably have to register first, and then choose: submission, i.e. (https://www.softconf.com/coling2016/CogALex-V/user/scmd.cgi?scmd=submitPaperCustom&pageid=0).

6       Organizers.

• Michael Zock (LIF, CNRS, Aix-Marseille University, Marseille, France)
• Alessandro Lenci (Computational Linguistics Laboratory, University of Pisa,, Italy)
• Stefan Evert (FAU, Erlangen-Nürnberg, Germany)

7       Contact persons

For general questions, please get in touch with Michael Zock (michael.zock@lif.univ-mrs.fr), for questions concerning the shared task, send an e-mail to Stefan Evert (stefan.evert@fau.de)

8       Program committee

Bieman Chris (Technische Universität, Darmstadt, Germany)

Babych, Bogdan (University of Leeds, UK)

Brysbaert, Marc (Experimental Psychology, Ghent University, Belgium)

Cristea Dan ('Al. I. Cuza' University, Iasi, Romania)

deDeyne Simon (University of Adelaide, Australia)

de Melo Gerard (IIIS, Tsinghua University, Beijing, China)

Evert, Stefan (University of Erlangen, Germany)

Ferret Olivier (CEA LIST, France)

Fontenelle Thierry (CDT, Luxemburg)

Gala Nuria (University of Aix-Marseille, France)

Geeraerts Dirk (University of Leuven, Belgium)

Granger Sylviane (Université Catholique de Louvain, Belgium)

Grefenstette Gregory (Inria, Paris, France)

Hirst Graeme (University of Toronto, Canada)

Hovy Ed (CMU, Pittsburgh, USA)

Hsieh, Shu-Kai (National Taiwan University, Taipei, Taiwan)

Joyce Terry (Tama University, Kanagawa-ken, Japan)

Lafourcade, Matthieu (LIRMM, université de Montepellier, France

Lapalme Guy (RALI, University of Montreal, Canada

Lebani Gianluca (University of Pisa, Italy)

Lenci Alessandro (University of Pisa, Italy)

L'Homme Marie Claude (University of Montreal, Canada)

Mititelu Verginica (RACAI, Bucharest, Romania)

Navigli, Roberto  (Sapienza, Rome, Italy)

Paradis Carita (Centre for Languages and Literature Lund University, Sweden)

Pihlevar, Taher  (university of Cambridge, UK)

Pirrelli, Vito (ILC, Pisa, Italy)

Polguère Alain (ATILF-CNRS, Nancy, France)

Purver, Matthew (King's College, London, UK)

Ramisch Carlos (AMU, Marseille, France)

Rayson Paul (UCREL, university of Lancaster, UK

Rosso, Paol (NLEL, Universitat Politècnica de València, Spain)

Sahlgren, Magnus (Gavagai Inc. & SICS, Sweden)

Schulte im Walde Sabine (University of Stuttgart, Germany)

Schwab Didier (LIG, Grenoble, France)

Sharoff Serge (University of Leeds, UK)

Stella Massimo (Institute for Complex Systems Simulation, university of Southhampton, UK)

Tokunaga Takenobu (TITECH, Tokyo, Japan)

Tufis Dan (RACAI, Bucharest, Romania)

Zarcone, Alessandra (Saarland University, Germany)

Zock Michael (LIF-CNRS, Marseille, France)