Explore the vast range of titles available.

This thesis investigates how the sub-structure of words can be accounted for in probabilistic models of language. Such models play an important role in natural language processing tasks such as translation or speech recognition, but often rely on the simplistic assumption that words are opaque symbols. This assumption does not fit morphologically complex language well, where words can have rich internal structure and sub-word elements are shared across distinct word forms. Our approach is to encode basic notions of morphology into the assumptions of three different types of language models, with the intention that leveraging shared sub-word structure can improve model performance and help overcome data sparsity that arises from morphological processes. In the context of n-gram language modelling, we formulate a new Bayesian model that relies on the decomposition of compound words to attain better smoothing, and we develop a new distributed language model that learns vector representations of morphemes and leverages them to link together morphologically related words. In both cases, we show that accounting for word sub-structure improves the models' intrinsic performance and provides benefits when applied to other tasks, including machine translation. We then shift the focus beyond the modelling of word sequences and consider models that automatically learn what the sub-word elements of a given language are, given an unannotated list of words. We formulate a novel model that can learn discontiguous morphemes in addition to the more conventional contiguous morphemes that most previous models are limited to. This approach is demonstrated on Semitic languages, and we find that modelling discontiguous sub-word structures leads to improvements in the task of segmenting words into their contiguous morphemes.

BackgroundTime of diagnosis (TOD) of benign esophageal perforation is regarded as an important risk factor for clinical outcome, although convincing evidence is lacking. The aim of this study is to assess whether time between onset of perforation and diagnosis is associated with clinical outcome in patients with iatrogenic esophageal perforation (IEP) and Boerhaave’s syndrome (BS).MethodsWe searched MEDLINE, Embase and Cochrane library through June 2018 to identify studies. Authors were invited to share individual patient data and a meta-analysis was performed (PROSPERO: CRD42018093473). Patients were subdivided in early (≤ 24 h) and late (> 24 h) TOD and compared with mixed effects multivariable analysis while adjusting age, gender, location of perforation, initial treatment and center. Primary outcome was overall mortality. Secondary outcomes were length of hospital stay, re-interventions and ICU admission.ResultsOur meta-analysis included IPD of 25 studies including 576 patients with IEP and 384 with BS. In IEP, early TOD was not associated with overall mortality (8% vs. 13%, OR 2.1, 95% CI 0.8–5.1), but was associated with a 23% decrease in ICU admissions (46% vs. 69%, OR 3.0, 95% CI 1.2–7.2), a 22% decrease in re-interventions (23% vs. 45%, OR 2.8, 95% CI 1.2–6.7) and a 36% decrease in length of hospital stay (14 vs. 22 days, p < 0.001), compared with late TOD. In BS, no associations between TOD and outcomes were found. When combining IEP and BS, early TOD was associated with a 6% decrease in overall mortality (10% vs. 16%, OR 2.1, 95% CI 1.1–3.9), a 19% decrease in re-interventions (26% vs. 45%, OR 1.9, 95% CI 1.1–3.2) and a 35% decrease in mean length of hospital stay (16 vs. 22 days, p = 0.001), compared with late TOD.ConclusionsThis individual patient data meta-analysis confirms the general opinion that an early (≤ 24 h) compared to a late diagnosis (> 24 h) in benign esophageal perforations, particularly in IEP, is associated with improved clinical outcome.