tailieunhanh - Báo cáo khoa học: "Evaluating Distributional Models of Semantics for Syntactically Invariant Inference"

A major focus of current work in distributional models of semantics is to construct phrase representations compositionally from word representations. However, the syntactic contexts which are modelled are usually severely limited, a fact which is reflected in the lexical-level WSD-like evaluation methods used. In this paper, we broaden the scope of these models to build sentence-level representations, and argue that phrase representations are best evaluated in terms of the inference decisions that they support, invariant to the particular syntactic constructions used to guide composition. . | Evaluating Distributional Models of Semantics for Syntactically Invariant Inference Jackie CK Cheung and Gerald Penn Department of Computer Science University of Toronto Toronto ON M5S 3G4 Canada j cheung gpenn @ Abstract A major focus of current work in distributional models of semantics is to construct phrase representations compositionally from word representations. However the syntactic contexts which are modelled are usually severely limited a fact which is reflected in the lexical-level WSD-like evaluation methods used. In this paper we broaden the scope of these models to build sentence-level representations and argue that phrase representations are best evaluated in terms of the inference decisions that they support invariant to the particular syntactic constructions used to guide composition. We propose two evaluation methods in relation classification and QA which reflect these goals and apply several recent compositional distributional models to the tasks. We find that the models outperform a simple lemma overlap baseline slightly demonstrating that distributional approaches can already be useful for tasks requiring deeper inference. 1 Introduction A number of unsupervised semantic models Mitchell and Lapata 2008 for example have recently been proposed which are inspired at least in part by the distributional hypothesis Harris 1954 that a word s meaning can be characterized by the contexts in which it appears. Such models represent word meaning as one or more high-dimensional vectors which capture the lexical and syntactic contexts of the word s occurrences in a training corpus. Much of the recent work in this area has following Mitchell and Lapata 2008 focused on the notion of compositionality as the litmus test of a truly semantic model. Compositionality is a natural way to construct representations of linguistic units larger than a word and it has a long history in Montagovian semantics for dealing with argument structure and assembling