Guy Emerson
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Guy Emerson has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
Alexis Palmer
1 shared publication- Ann Copestake
University of Cambridge
1 shared publication Liling Tan
1 shared publication- Michaela Regneri
Saarland University
1 shared publication - Olga Zamaraeva
University of Washington
1 shared publication Stefania Preda
1 shared publication
Latest Publications
- Open Access
Using dependency parsing for few-shot learning in distributional semantics
In this work, we explore the novel idea of employing dependency parsing information in the context of few-shot learning, the task of learning the meaning of a rare word based on a limited amount of context sentences. Firstly, we use dependency-based word embedding models as background spaces for few-shot learning. Secondly, we introduce two few-shot learning methods which enhance the additive baseline model by using dependencies.
Open AccessUsing dependency parsing for few-shot learning in distributional semantics
In this work, we explore the novel idea of employing dependency parsing information in the context of few-shot learning, the task of learning the meaning of a rare word based on a limited amount of context sentences. Firstly, we use dependency-based word embedding models as background spaces for few-shot learning. Secondly, we introduce two few-shot learning methods which enhance the additive baseline model by using dependencies.
- Open Access
- 2 February 2021
- Published by Cambridge University Press (CUP) in Cambridge Open Engage
Autoencoding Pixies: Amortised Variational Inference with Graph Convolutions for Functional Distributional Semantics
The aim of distributional semantics is to learn the meanings of words from a corpus of text. The aim of formal semantics is to develop mathematical models of meaning. Functional Distributional Semantics provides a framework for distributional semantics which is interpretable in formal semantic terms, by representing the meaning of a word as a truth-conditional function (a binary classifier). However, the model introduces a large number of latent variables, which means that inference is computationally expensive, and training a model is therefore slow to converge. In this work, I introduce the Pixie Autoencoder, which augments the generative model of Functional Distributional Semantics with a graph-convolutional neural network to perform amortised variational inference. This allows the model to be trained more effectively, achieving better results on semantic similarity in context, and outperforming BERT, a large pre-trained language model.
- 2 February 2021
- Published by Cambridge University Press (CUP) in Cambridge Open Engage
Open AccessAutoencoding Pixies: Amortised Variational Inference with Graph Convolutions for Functional Distributional Semantics
The aim of distributional semantics is to learn the meanings of words from a corpus of text. The aim of formal semantics is to develop mathematical models of meaning. Functional Distributional Semantics provides a framework for distributional semantics which is interpretable in formal semantic terms, by representing the meaning of a word as a truth-conditional function (a binary classifier). However, the model introduces a large number of latent variables, which means that inference is computationally expensive, and training a model is therefore slow to converge. In this work, I introduce the Pixie Autoencoder, which augments the generative model of Functional Distributional Semantics with a graph-convolutional neural network to perform amortised variational inference. This allows the model to be trained more effectively, achieving better results on semantic similarity in context, and outperforming BERT, a large pre-trained language model.
- 15 October 2020
- Published by Not Crossref prefixes - transferred records in Proceedings of the International Conference on Head-Driven Phrase Structure Grammar
- p. 157–177
Multiple question fronting without relational constraints: An analysis of Russian as a basis for cross-linguistic modeling
We present an analysis of multiple question fronting in a restricted variant of the HPSG formalism (DELPH-IN) where unification is the only natively defined operation. Analysing multiple fronting in this formalism is challenging, because it requires carefully handling list appends, something that HPSG analyses of question fronting heavily rely on. Our analysis uses the append list type to address this challenge. We focus the testing of our analysis on Russian, although we also integrate it into the Grammar Matrix customization system where it serves as a basis for cross-linguistic modeling. In this context, we discuss the relationship of our analysis to lexical threading and conclude that, while lexical threading has its advantages, modeling multiple extraction cross-linguistically is easier without the lexical threading assumption.
- 15 October 2020
- Published by Not Crossref prefixes - transferred records in Proceedings of the International Conference on Head-Driven Phrase Structure Grammar
- p. 157–177
Multiple question fronting without relational constraints: An analysis of Russian as a basis for cross-linguistic modeling
We present an analysis of multiple question fronting in a restricted variant of the HPSG formalism (DELPH-IN) where unification is the only natively defined operation. Analysing multiple fronting in this formalism is challenging, because it requires carefully handling list appends, something that HPSG analyses of question fronting heavily rely on. Our analysis uses the append list type to address this challenge. We focus the testing of our analysis on Russian, although we also integrate it into the Grammar Matrix customization system where it serves as a basis for cross-linguistic modeling. In this context, we discuss the relationship of our analysis to lexical threading and conclude that, while lexical threading has its advantages, modeling multiple extraction cross-linguistically is easier without the lexical threading assumption.
- 16 October 2015
- Published by Not Crossref prefixes - transferred records in Proceedings of the International Conference on Head-Driven Phrase Structure Grammar
- p. 75–95
Lacking integrity: HPSG as a morphosyntactic theory
Standard accounts of HPSG assume a distinction between morphology and syntax. However, despite decades of research, no cross-linguistically valid definition of 'word' has emerged (Haspelmath, 2010), suggesting that no sharp distinction is justified. Under such a view, the basic units are morphemes, rather than words, but it has been argued this raises problems when analysing phenomena such as zero inflection, syncretism, stem alternations, and extended exponence. We argue that with existing HPSG machinery, a morpheme-based approach can in fact deal with such issues. To illustrate this, we consider Slovene nominal declension and Georgian verb agreement, which have both been used to argue against constructive morpheme-based approaches. We overcome these concerns through use of a type hierarchy, and give a morpheme-based analysis which is simpler than the alternatives. Furthermore, we can recast notions from Word-and-Paradigm morphology, such as 'rule of referral' and 'stem space', in our framework. We conclude that using HPSG as a unified morphosyntactic theory is not only feasible, but also yields fruitful insights.
- 16 October 2015
- Published by Not Crossref prefixes - transferred records in Proceedings of the International Conference on Head-Driven Phrase Structure Grammar
- p. 75–95
Lacking integrity: HPSG as a morphosyntactic theory
Standard accounts of HPSG assume a distinction between morphology and syntax. However, despite decades of research, no cross-linguistically valid definition of 'word' has emerged (Haspelmath, 2010), suggesting that no sharp distinction is justified. Under such a view, the basic units are morphemes, rather than words, but it has been argued this raises problems when analysing phenomena such as zero inflection, syncretism, stem alternations, and extended exponence. We argue that with existing HPSG machinery, a morpheme-based approach can in fact deal with such issues. To illustrate this, we consider Slovene nominal declension and Georgian verb agreement, which have both been used to argue against constructive morpheme-based approaches. We overcome these concerns through use of a type hierarchy, and give a morpheme-based analysis which is simpler than the alternatives. Furthermore, we can recast notions from Word-and-Paradigm morphology, such as 'rule of referral' and 'stem space', in our framework. We conclude that using HPSG as a unified morphosyntactic theory is not only feasible, but also yields fruitful insights.
- 1 January 2014
- Published by Association for Computational Linguistics (ACL)
- p. 30-38
SentiMerge: Combining Sentiment Lexicons in a Bayesian Framework
- 1 January 2014
- Published by Association for Computational Linguistics (ACL)
- p. 30-38
SentiMerge: Combining Sentiment Lexicons in a Bayesian Framework