SIGNLL (pronounce as signal) is the Special Interest Group on Natural Language Learning of the Association for Computational Linguistics (ACL).

This page contains the following information:

Natural Language Learning before SIGNLL

Language Learning is the primary focus of a entire subdiscipline of Psychology, viz. Psycholinguistics, and has always been an arena for experimentation in the Artificial Intelligence Machine Learning community. More recently still, relevant work is being pursued in the emerging connectionist movement, and Natural Language Learning (NLL) is in addition proving to be a natural outgrowth of current work in statistical, information-theoretic and corpus-based methods in Computational Linguistics.

In the years before the foundation of SIGNLL, we have seen NLL sessions, workshops and symposia under auspices as varied as COLING (The Unfinished Language, 90), DARPA (90/91), AAAI (MLNLO/CNLP, 91/93), IJCAI (NLL, 91) and ECML (Machine Learning and Text Analysis, 93), although the history of NLL goes back at least a couple of decades.

Events have been held under names as diverse as Natural Language Learning (NLL: reflecting the relationship with the AI subfield of Natural Language), Machine Learning of Natural Language and Ontology (MLNLO: reflecting the intersection with AI subfield of Machine Learning and giving special prominence to the problems of Semantics and Representation), Extraction of Hierarchical Structure (SHOE: emphasizing the development of general hierarchical clustering techniques), Connectionist Natural Language Processing (CNLP), etc.

At the latest of these, the NLL community proposed that we should seek to identify ourselves formally with the ACL as a SIG, and noted that while we are just a source of esoteric applications for ML (primarily concerned with the development of new ML algorithms and their commercial exploitation as Automated Knowledge Acquisition) our aim is by contrast the answering of fundamental scientific questions about the nature of the human language acquisition process and the development of practical NLL techniques for solving current problems across the full range of Computational Linguistics, whilst admitting the widest possible range of computational approaches. Not only does this range encompass machine learning, connectionist, genetic, statistical and information-theoretic methods, but we are prepared also to borrow techniques from further afield (e.g. cryptography, compression and higher criticism).

The aims and purposes of SIGNLL

Aims

SIGNLL aims to promote research in:

  • automated acquisition of syntax, morphology and phonology
  • automated acquisition of semantic / ontological structure
  • automated acquisition of inter-linguistic correspondences
  • learning to recognize or produce spoken and written forms
  • modelling human language acquisition theory and processes

Purview

SIGNLL will foster the application and development of new techniques for the automatic analysis of language, including but not limited to:

  • case-based, example-based and explanation-based learning
  • connectionist, statistical and information-theoretic models
  • inductive, deductive and analytic techniques
  • clustering and classification procedures

Approaches and Paradigms

SIGNLL emphasizes paradigms which can be exploited automatically:

  • corpus-based analysis including automated tagging and testing
  • learning in interactive environments with minimal supervision
  • automatic preprocessing feeding overtly supervised techniques
  • unsupervised and naturally / implicitly supervised techniques

Functions and Activities

SIGNLL aims to perform and encourage the following functions and activities:

  • promotion and development of the field and avenues for publication
  • provision and coordination of a library of language learning software and data
  • facilitation of communication between researchers in this field
  • provision of information about relevant research and resources
  • development of standard corpora and interface formats for NLL
  • coordination of the organization of workshops and symposia
  • liaison with other SIGs, funding organizations, etc.

SIGNLL Officers

President

Alexander Clark <alexc (at) cs.rhul.ac.uk> Royal Holloway University of London (United Kingdom)

Secretary

Xavier Carreras <carreras (at) lsi.upc.edu> Universitat Politècnica de Catalunya (Spain)

Steering Committee

Antal van den Bosch, Tilburg University (Netherlands)
Claire Cardie, Cornell University (USA)
Walter Daelemans, University of Antwerp (Belgium)
Lluís Màrquez Universitat Politècnica de Catalunya (Spain)
Hwee Tou Ng, National University of Singapore (Singapore)
Joakim Nivre Uppsala University (Sweden)
David Powers, Flinders University (Australia)
Dan Roth, University of Illinois at Urbana-Champaign (USA)

International Advisory Committee

Antal van den Bosch, Tilburg University (Netherlands)
Thorsten Brants, Google Inc. (USA)
Claire Cardie, Cornell University (USA)
James Cussens, University of York (UK)
Walter Daelemans, University of Antwerp (Belgium)
Ido Dagan, Bar Ilan University (Israel)
Diane Litman, University of Pittsburgh (USA)
Lluís Màrquez Universitat Politècnica de Catalunya (Spain)
Rada Mihalcea, University of North Texas (USA)
Yuji Matsumoto, Nara Institute of Science and Technology (Japan)
Raymond Mooney, University of Texas at Austin (USA)
John Nerbonne, University of Groningen (Netherlands)
Hwee Tou Ng, National University of Singapore (Singapore)
Grace Ngai, The Hong Kong Polytechnic University (Hong Kong)
Joakim Nivre Uppsala University (Sweden)
Miles Osborne, University of Edinburgh (UK)
David Powers, Flinders University (Australia)
Ellen Riloff, University of Utah (USA)
Dan Roth, University of Illinois at Urbana-Champaign (USA)

SIGDAT Liaison Representative

David Yarowsky, Johns Hopkins University (USA)

Information Officer

Erik Tjong Kim Sang, University of Groningen (The Netherlands)

SIGNLL Past Officers

The SIGNLL President and the SIGNLL Secretary are elected for a period of two years. The current office period runs from November 1, 2011 to October 31, 2013.

  • 2009-2011: President: Joakim Nivre. Secretary: Lluís Màrquez
  • 2007-2009: President: Lluís Màrquez. Secretary: Joakim Nivre
  • 2005-2007: President: Antal van den Bosch. Secretary:: Hwee Tou Ng
  • 2003-2005: President: Dan Roth. Secretary: Antal van den Bosch
  • 2001-2003: President: Walter Daelemans. Secretary: Dan Roth
  • 1999-2001: President: Walter Daelemans. Secretary: Claire Cardie
  • 1997-1999: President: Michael Brent. Secretary: Walter Daelemans
  • 1995-1997: President: David Powers. Secretary: Walter Daelemans

Constitution

The constitution governing ACL SIGNLL is available from the secretary on request. Members should be, and elected officers must be, members of the Association for Computational Linguistics.