Proceedings of the Workshop on Geometrical Models of Natural Language Semantics - GEMS '09

Abstract

The geometry of distributional models of lexical semantics represent a core topic in contemporary computational linguistics for its impact on several advanced Natural Language Processing tasks and some related knowledge fields (as social science and humanities). The goal of the EACL 2009 GEMS Workshop on "GEometrical Models of natural Language Semantics" was to stimulate research on semantic spaces and distributional methods in NLP, by adopting an interdisciplinary view. This aimed to enforce the proper exchange of ideas, results and resources among often independent communities. The workshop provided a common ground for a fruitful discussion among experts of distributional approaches, collocational corpus analysis and machine learning, researchers interested in the use of quantitative models in NLP applications (like question answering, summarization or textual entailment), experts in formal computational semantics and in other fields of science as well. The workshop successfully gathered a relevant number of high quality contributions to problems of meaning representation, acquisition and use, based on distributional and vector space models. We received 21 submissions, including short and long papers. Long papers were peer-reviewed by three members of the program committee, short papers by two. As an outcome of the review process, the program committee selected 11 papers for a full presentation, and 4 for short ones. All selected paper have been included in these proceedings. The paper are representative of the current state of the art in the subject, including: •cutting edge researches on geometric methods and machine learning, such as tensor factorization, kernel methods and Dirichlet process mixture models; •applications of semantic space models to NLP tasks, such as Textual Entailment Recognition, Ontology Learning, Induction of Selectional Preferences, Verb Classification and Machine Translation •novel uses of distributional methods for advanced linguistic studies, such as lexical variation and evolution as well as for educational purposes; •reference comparative studies among different types of semantic spaces. The papers included in this volume shed some light on the state of the art and the potential applications of semantic spaces in NLP and in related linguistic fields.