Co-Dispersion: A Windowless Approach to Lexical Association

Proceedings of the 12th Conference of the European Chapter of the ACL, pages 861–869, Athens, Greece, 30 March – 3 April 2009. c⃝2009 Association for Computational Linguistics Co-dispersion: A Windowless Approach to Lexical Association Justin Washtell University of Leeds Leeds, UK washtell@comp.leeds.ac.uk Abstract We introduce an alternative approach to ex- tracting word pair associations from corpora, based purely on surface distances in the text. We contrast it with the prevailing window- based co-occurrence model and show it to be more statistically robust and to disclose a broader selection of significant associative re- lationships - owing largely to the property of scale-independence. In the process we provide insights into the limiting characteristics of window-based methods which complement the sometimes conflicting application-oriented lit- erature in this area. 1 Introduction The principle of using statistical measures of co- occurrence from corpora as a proxy for word association - by comparing observed frequencies of co-occurrence with expected frequencies - is relatively young. One of the most well known computational studies is that of Church & Hanks (1989). The method by which co-occurrences are counted, now as then, is based on a device which dates back at least to Weaver (1949): the context window. While variations on the specific notion of context have been explored (separation of content and function words, asymmetrical and non-contiguous contexts, the sentence or the document as context) and increasingly sophisti- cated association measures have been proposed (see Evert, 2007, for a thorough review) the basic principle – that of counting token frequencies within a context region – remains ubiquitous. Herein we discuss some of the intrinsic limi- tations of this approach, as are being felt in re- cent research, and present a principled solution which does not rely on co-occurrence windows at all, but instead on measurements of the surface distance between words. 2 The impact of window size The issue of how to determine appropriate win- dow size (and shape) has often been glossed over in the literature, with such parameters being de- termined arbitrarily, or empirically on a per- application basis, and often receiving little more than a cursory mention under the description of method. For reasons that we will discuss how- ever, the issue has been receiving increasing at- tention. Some have attempted to address it intrin- sically (Sahlgren 2006; Schulte im Walde & Melinger, 2008; Hung et al, 2001); others no less earnestly in the interests of specific applications (Lamjiri, 2003; Edmonds, 1997; Wang 2005; Choueka & Lusignan, 1985) (note that this di- vide is sometimes subtle). The 2008 Workshop on Distributional Lexi- cal Semantics, held in conjunction with the European Summer School on Logic, Language and Learning (ESSLLI) – hereafter the ESSLLI Workshop - saw this issue (along with other “problem” parameters in distributional lexical semantics) as one of its central themes, and wit- nessed many different takes upon it. Interest- ingly, there was little consensus, with some stud- ies appearing on the surface to starkly contradict one-another. It is now generally recognized that window …

Content truncated to 500 words