Alexandru Coca, Mark Gaynor, Zhenxing Zhang, Jianpeng Cheng, Bo-Hsiang Tseng, Peter Boothroyd, Hector Martinez Alonso, Diarmuid O Seaghdha, Anders Johannsen. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Programmable task-oriented dialogue (TOD) agents enable language models to follow structured dialogue policies, but their effectiveness hinges on accurate dialogue state tracking (DST). We present PyTOD, an agent that generates executable code to track dialogue state and uses policy and execution feedback for efficient error correction. To achieve this, PyTOD employs a simple constrained decoding approach, using a language model instead of grammar rules to follow API schemata. This leads to state-of-the-art DST performance on the challenging SGD benchmark. Our experiments show that PyTOD surpasses strong baselines in both accuracy and cross-turn consistency, demonstrating the effectiveness of execution-aware state tracking.

Jianpeng Cheng, Devang Agrawal, Héctor Martínez Alonso, Shruti Bhargava, Joris Driesen, Federico Flego, Dain Kaplan, Dimitri Kartsaklis, Lin Li, Dhivya Piraviperumal, Jason D. Williams, Hong Yu, Diarmuid Ó Séaghdha, Anders Johannsen. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2020.

One of the core components of modern spoken dialogue systems is the belief tracker, which estimates the user's goal at every step of the dialogue.However, most current approaches have difficulty scaling to larger, more complex dialogue domains.This is due to their dependency on either: a) Spoken Language Understanding models that require large amounts of annotated training data; or b) hand-crafted lexicons for capturing some of the linguistic variation in users' language.We propose a novel Neural Belief Tracking (NBT) framework which overcomes these problems by building on recent advances in representation learning.NBT models reason over pre-trained word vectors, learning to compose them into distributed representations of user utterances and dialogue context.Our evaluation on two datasets shows that this approach surpasses past limitations, matching the performance of state-of-the-art models which rely on hand-crafted semantic lexicons and outperforming them when such lexicons are not provided.

Ivan Vulić, Nikola Mrkšić, Roi Reichart, Diarmuid Ó Séaghdha, Steve Young, Anna Korhonen. Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2017.

We present Attract-Repel, an algorithm for improving the semantic quality of word vectors by injecting constraints extracted from lexical resources. Attract-Repel facilitates the use of constraints from mono- and cross-lingual resources, yielding semantically specialized cross-lingual vector spaces. Our evaluation shows that the method can make use of existing cross-lingual lexicons to construct high-quality vector spaces for a plethora of different languages, facilitating semantic transfer from high- to lower-resource ones. The effectiveness of our approach is demonstrated with state-of-the-art results on semantic similarity datasets in six languages. We next show that Attract-Repel-specialized vectors boost performance in the downstream task of dialogue state tracking (DST) across multiple languages. Finally, we show that cross-lingual vector spaces produced by our algorithm facilitate the training of multilingual DST models, which brings further performance improvements.

Nikola Mrkšić, Diarmuid Ó Séaghdha, Blaise Thomson, Milica Gašić, Lina M. Rojas-Barahona, Pei-Hao Su, David Vandyke, Tsung-Hsien Wen, Steve Young. Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2016.

New scientific concepts, interpreted broadly, are continuously introduced in the literature, but relatively few concepts have a long‐term impact on society. The identification of such concepts is a challenging prediction task that would help multiple parties—including researchers and the general public—focus their attention within the vast scientific literature. In this paper we present a system that predicts the future impact of a scientific concept, represented as a technical term, based on the information available from recently published research articles. We analyze the usefulness of rich features derived from the full text of the articles through a variety of approaches, including rhetorical sentence analysis, information extraction, and time‐series analysis. The results from two large‐scale experiments with 3.8 million full‐text articles and 48 million metadata records support the conclusion that full‐text features are significantly more useful for prediction than metadata‐only features and that the most accurate predictions result from combining the metadata and full‐text features. Surprisingly, these results hold even when the metadata features are available for a much larger number of documents than are available for the full‐text features.

We describe a probabilistic framework for acquiring selectional preferences of linguistic predicates and for using the acquired representations to model the effects of context on word meaning. Our framework uses Bayesian latent-variable models inspired by, and extending, the well-known Latent Dirichlet Allocation (LDA) model of topical structure in documents; when applied to predicate–argument data, topic models automatically induce semantic classes of arguments and assign each predicate a distribution over those classes. We consider LDA and a number of extensions to the model and evaluate them on a variety of semantic prediction tasks, demonstrating that our approach attains state-of-the-art performance. More generally, we argue that probabilistic methods provide an effective and flexible methodology for distributional semantics.

Abstract This paper presents a classification-based approach to noun–noun compound interpretation within the statistical learning framework of kernel methods. In this framework, the primary modelling task is to define measures of similarity between data items, formalised as kernel functions. We consider the different sources of information that are useful for understanding compounds and proceed to define kernels that compute similarity between compounds in terms of these sources. In particular, these kernels implement intuitive notions of lexical and relational similarity and can be computed using distributional information extracted from text corpora. We report performance on classification experiments with three semantic relation inventories at different levels of granularity, demonstrating in each case that combining lexical and relational information sources is beneficial and gives better performance than either source taken alone. The data used in our experiments are taken from general English text, but our methods are also applicable to other domains and potentially to other languages where noun–noun compounding is frequent and productive.

Recommendation systems exist to help users discover content in a large body of items. An ideal recommendation system should mimic the actions of a trusted friend or expert, producing a personalised collection of recommendations that balance between the desired goals of accuracy, diversity, novelty and serendipity. We introduce the Auralist recommendation framework, a system that - in contrast to previous work - attempts to balance and improve all four factors simultaneously. Using a collection of novel algorithms inspired by principles of "serendipitous discovery", we demonstrate a method of successfully injecting serendipity, novelty and diversity into recommendations whilst limiting the impact on accuracy. We evaluate Auralist quantitatively over a broad set of metrics and, with a user study on music recommendation, show that Auralist's emphasis on serendipity indeed improves user satisfaction.

BACKGROUND: Applications of Natural Language Processing (NLP) technology to biomedical texts have generated significant interest in recent years. In this paper we identify and investigate the phenomenon of linguistic subdomain variation within the biomedical domain, i.e., the extent to which different subject areas of biomedicine are characterised by different linguistic behaviour. While variation at a coarser domain level such as between newswire and biomedical text is well-studied and known to affect the portability of NLP systems, we are the first to conduct an extensive investigation into more fine-grained levels of variation. RESULTS: Using the large OpenPMC text corpus, which spans the many subdomains of biomedicine, we investigate variation across a number of lexical, syntactic, semantic and discourse-related dimensions. These dimensions are chosen for their relevance to the performance of NLP systems. We use clustering techniques to analyse commonalities and distinctions among the subdomains. CONCLUSIONS: We find that while patterns of inter-subdomain variation differ somewhat from one feature set to another, robust clusters can be identified that correspond to intuitive distinctions such as that between clinical and laboratory subjects. In particular, subdomains relating to genetics and molecular biology, which are the most common sources of material for training and evaluating biomedical NLP tools, are not representative of all biomedical subdomains. We conclude that an awareness of subdomain variation is important when considering the practical use of language processing applications by biomedical researchers.