LINGUIST List 26.1498
Wed Mar 18 2015
Calls: Computational Linguistics; Semantics/ Computational Linguistics (Jrnl)
Editor for this issue: Andrew Lamont <alamontlinguistlist.org>
Date: 18-Mar-2015
From: Aurelie Herbelot <aurelie.herbelot
cantab.net>
Subject: Computational Linguistics; Semantics/ Computational Linguistics (Jrnl)
E-mail this message to a friend Full Title: Computational Linguistics
Linguistic Field(s): Computational Linguistics; Semantics
Call Deadline: 10-Apr-2015
Special Issue of Computational Linguistics: Formal Distributional Semantics [Deadline extended]
Final Call for Papers
Extended submission deadline: April 10th 2015
http://www.cl.cam.ac.uk/~ah433/CL-special-issue-formal-distributional-semantics.html The semantics of natural language consists of complex phenomena encompassing functional aspects such as quantification (e.g. ''the cat'' vs. ''a cat'') and conceptual aspects related to word meaning (e.g. ''cat'' vs. ''animal''; ''visit Boston'' vs. ''visit a friend''). No existing theory of meaning accounts for both, and existing approaches are typically biased towards one or the other. For instance, formal semantics focuses on functional aspects, providing a systematic treatment of compositionality through a clear syntax-semantics interface -- at the expense of lexical semantics. Distributional, or vector-space, semantics (Turney & Pantel, 2010), on the other hand, excels at lexical semantics phenomena ranging from word similarity to categorization, and it has recently made progress towards the treatment of composition (Baroni 2013); however, functional aspects remain mostly unaccounted for.
Because of the complementary strengths of the two approaches, the computational linguistics community has started investigating proposals for an overarching architecture, combining formal and distributional semantics (e.g. Coecke et al., 2011; Erk, 2013; Lewis and Steedman 2013; Baroni et al., 2014). This effort holds the promise of significantly advancing the state of the art. However, given the fundamentally different nature of formal and distributional semantics, the enterprise also poses great challenges. The aim of this special issue is to explore the boundaries of a formal distributional semantics, by proposing relevant computational accounts of meaning and applying the corresponding frameworks to specific linguistic phenomena.
Topics
For this special issue, we solicit article submissions describing original research on any aspect of formal distributional semantics integrating a computational perspective. Possible topics include, but are not limited to:
- Theoretical questions: What is meaning in formal distributional semantics, and how well do computational models simulate the relevant theories? How can distributional representations be related to the traditional components of a semantics for natural languages, especially reference and truth? Is similarity (the chief notion in distributional semantics) at odds with inference (one of the test beds of formal semantics), or can it support it?
- Framework issues: Should a framework be developed that encompasses both formal and distributional semantics in a single formalism (Baroni et al., 2014), or should the two approaches be kept separate and linked via systematic interactions (Lewis and Steedman 2013; Garrette et al., 2014)? How do different frameworks fare in standard computational semantics benchmarks (RTE, STS, etc.)?
- Linguistic phenomena: Can formal distributional semantics account for known phenomena? Can it shed new light on old puzzles? Can it handle newly observed phenomena?
Submission Instructions:
Articles submitted to this special issue must adhere to the Computational Linguistics Style Guidelines. The submission guidelines can be found on the CL web site (
http://cljournal.org/submissions.html).
Guest Editors
Gemma Boleda
Universitat Pompeu Fabra, Barcelona, Spain
gemma.boleda
upf.edu
Aurelie Herbelot
University of Cambridge, UK
aurelie.herbelot
cantab.net
Page Updated: 18-Mar-2015