**************************************************************** C A L L f o r P A P E R Journal of Natural Language Engineering Special Issue on: Distributional Lexical Semantics URL: http://art.uniroma2.it/jnle **************************************************************** In the last decades, vector space models (VSM) have received a growing attention in different fields of Artificial Intelligence, ranging from natural language processing (NLP) and cognitive science, to vision analysis and applications in the humanities. The basic idea of VSM is to represent entities as vectors in a geometric space, so that their similarity can be measured according to distance metrics in the space. VSM have demonstrated to successfully model and solve a variety of problems, such as metaphor detection and analysis, priming, discourse analysis, and information retrieval. In computational linguistics, the Distributional Hypothesis leverages the notion of VSM to model the semantics of words and other linguistic entities. The hypothesis was autonomously elaborated in different works, and has been since then applied through different settings. The hypothesis' core states that 'a word is defined by the company it keeps', i.e. by the set of linguistic contexts in which it appears. It follows that two target words appearing in similar contexts likely have similar or related meanings ('distributional similarity'). Different types of contexts (e.g. bag-of-words, syntactic relations, documents) tend to capture different semantic relations between target words (e.g. relatedness, similarity, topicality). Practical uses of the distributional hypothesis are today very popular, in various large scale linguistic learning tasks and applications, such as harvesting thesauri, word sense disambiguation, inference rules harvesting, selectional preference acquisition, conceptual clustering, modeling frame semantics information, question answering and synonym detection. Despite the growing popularity of distributional approaches, existing literature raises issues on many important aspects that have still to be addressed. Examples are: the need of comparative in depth analyses of the semantic properties captured by different types of distributional models; the application of new geometrical approaches as the use of quantum logic operators or tensor decomposition; the study of the interaction between distributional approaches and supervised machine learning, as the adoption of kernel methods based on distributional information; the application of distributional techniques in real world applications and in other fields. The special issue follows up most recent and similar efforts to summarize and harmonize researches on distributional techniques. We here refer to the 'Contextual Information in Semantic Space Models' workshop (2007); the ESSLLI workshop on 'Distributional Lexical Semantics' (2008); and the 'SigLex-SigSem GEMS workshop' (2009). All these workshops indicate the growing interest in the area in the last years. ------------------------------------------------------------------------------ Topics ====== The goal of the special issue is to offer a common journal venue where to gather and summarize the state of the art on distributional techniques applied to lexical semantics, as a cornerstone in computational linguistics research. As a side effect, the aim is also to propose a systematic and harmonized view of the works carried out independently by different researchers in the last years, which sometimes resulted in diverging and somehow inconsistent uses of terminology and axiomatizations. A further goal is to increase awareness in the computational linguistic community about cutting-edge studies on geometrical models, machine learning applications and experiences in different scientific fields. The special issue in particular focuses on the following areas of interest, building on topics proposed for the GEMS workshop (EACL 2009, Athens, http://art.uniroma2.it/gems): * Comparisons analysis of different distributional spaces (document-based, word-based, syntax based and others) and their parameters (dimension, corpus size, etc.) * 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 * Cognitive theories of semantic space models * Applications in the humanities and social sciences * Application of distributional approaches in : o Word sense disambiguation and discrimination o Selectional preference induction o Acquisition of lexicons and linguistic patterns o Conceptual clustering o Kernels methods for NLP (e.g. relation extraction and textual entailment) o Quantitative extensions of Formal Concept Analysis o Modeling of linguistic and ontological knowledge ---------------------------------------------------------------- Important Dates =============== Call for Paper : March 2009 Submissions deadline : 30 June 2009 First Evaluation Results : October 2009 Second Submission : January 2010 Final Acceptance : March 2010 Special Issue : May 2010 ---------------------------------------------------------------- Submission ========== Submission details will be soon announced through the Special Issue web page: http://art.uniroma2.it/jnle ---------------------------------------------------------------- Editorial Board =============== - Guest Editors Roberto Basili (University of Roma Tor Vergata, Italy) Marco Pennacchiotti (Yahoo! Inc., Santa Clara, USA) - Guest Editorial Board Marco Baroni (University of Trento, Italy) Michael W. Berry (University of Tenneesee) Johan Bos (University of Roma "La Sapienza", Italy) Paul Buitelaar (DERI, National University of Ireland, Galway) John A. Bullinaria (University of Birmingham, UK) Rodolfo Delmonte (University of Venice, Italy) Susan Dumais (Microsoft Research) Katrin Erk (University of Texas, US) Stefan Evert (University of Osnabruck, Germany) Gregory Grefenstette (Exalead S.A., France) Alfio Massimiliano Gliozzo (STLab - ISTC - CNR, Italy ) Mirella Lapata (University of Edinburgh, UK) Alessandro Lenci (University of Pisa, Italy) Jussi Karlgren (Swedish Institute of Computer Science, Sweden) Will Lowe (University of Nottingham, UK) Diana McCarthy (University of Sussex) Alessandro Moschitti (University of Trento, Italy) Saif Mohammad (University of Mryland, US) Sebastian Pado (Stanford University, US) Patrick Pantel (Yahoo! Inc., US) Ted Pedersen (University of Minnesota, Duluth, US) Massimo Poesio (University of Trento, Italy) Magnus Sahlgren (Swedish Institute of Computer Science, Sweden) Alessandro Moschitti (University of Trento, Italy) Saif Mohammad (University of Mryland, US) Sebastian Pado (Stanford University, US) Patrick Pantel (Yahoo! Inc., US) Ted Pedersen (University of Minnesota, Duluth, US) Massimo Poesio (University of Trento, Italy) Magnus Sahlgren (Swedish Institute of Computer Science, Sweden) Sabine Schulte im Walde (University of Stuttgart, Germany) Hinrich Schutze (Stuttgart University) Suzanne Stevenson (University of Toronto, Canada) Peter D. Turney (National Research Council, Canada) Carlo Strapparava (Fondazione Bruno Kessler, Italy) Dominic Widdows (Google Research, US) Yorick Wilks (University of Sheffield, UK) Fabio Massimo Zanzotto (University of Roma "Tor Vergata", Italy) ---------------------------------------------------------------- Contacts ======== Roberto Basili Marco Pennacchiotti Department of Computer Science Yahoo! Inc. University of Roma "Tor Vergata" Santa Clara, CA Italy USA Web page Web page