GEMS : GEometrical Models of Natural Language Semantics

Distributional models and semantic spaces represent a core topic in contemporary computational linguistics for their impact on advanced tasks and on other knowledge fields (such as social science and the humanities).

The goal of the GEMS workshop is to further stimulate research on semantic spaces  and distributional methods for NLP, by adopting an interdisciplinary

approach to allow a proper exchange of ideas, results and resources among often independent  communities. In particular, the workshop will provide a common ground for a fruitful discussion among experts of distributional approaches, collocational corpus analysis and machine learning; researchers interested in the use of statistical models in NLP applications (e.g.
question
answering, summarization and textual entailment) and in other fields of science; and experts in formal computational semantics.

The workshop aims at gathering contemporary contributions to large scale problems in meaning representation, acquisition and use, based on distributional and vector space models. The workshop aims also to shed new light on the use of such  techniques  on complex linguistic tasks, such as linguistic knowledge acquisition, semantic role labeling, textual entailment recognition, question answering, document understanding/summarization and ontology learning.

The workshop is endorsed by SIGLEX, the ACL Special Interest Group on the Lexicon.

TOPICS OF INTEREST

We invite submissions on any topic of current interest related to the
application of semantic spaces to NLP and related disciplines, such as:

  - Document-based, Collocational and Syntagmatic spaces
  - Eigenvector methods (e.g. Singular Value and Tucker Decomposition)
  - Higher order tensors and Quantum Logic extensions
  - Feature engineering in machine learning models
  - Computational complexity and evaluation issues
  - Graph-based models over semantic spaces
  - Logic and inference in semantic spaces
  - Psychological and cognitive theories of semantic space models
  - Applications in the humanities and social sciences

We also especially encourage submissions on the empirical  evaluation of the

above computational models within the  following NLP tasks:

  - Word sense disambiguation and discrimination
  - Selectional preference induction
  - Acquisition of lexicons and linguistic patterns
  - Conceptual clustering
  - Kernels methods for NLP (e.g. relation extraction and textual entailment)
  - Quantitative extensions of Formal Concept Analysis
  - Modeling of linguistic theories and ontological knowledge:

IMPORTANT DATES

Submission deadline: December 19, 2008
Notification of acceptance: January 30, 2009
Camera-ready papers due: February 13, 2009
Workshop: either March 30 or 31, 2009 (to be announced)