This paper introduces a novel methodology for extracting semantic frames from text corpora. Building on recent advances in computational construction grammar, the method captures expert knowledge of how semantic frames can be expressed in the form of conventionalised form-meaning pairings, called constructions. By combining these constructions in a semantic parsing process, the frame-semantic structure of a sentence is retrieved through the intermediary of its morpho-syntactic structure. The main advantage of this approach is that state-of-the-art results are achieved, without the need for annotated training data. We demonstrate the method in a case study where causation frames are extracted from English newspaper articles, and compare it to a commonly used approach based on Conditional Random Fields (CRFs). The computational construction grammar approach yields a word-level F1 score of 78.5%, outperforming the CRF approach by 4.5 percentage points.

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Going from natural language directions to fully specified executable plans for household robots involves a challenging variety of reasoning steps. In this paper, a processing pipeline to tackle these steps for natural language directions is proposed and implemented. It uses the ontological Socio-physical Model of Activities (SOMA) as a common interface between its components. The pipeline includes a natural language parser and a module for natural language grounding. Several reasoning steps formulate simulation plans, in which robot actions are guided by data gathered using human computation. As a last step, the pipeline simulates the given natural language direction inside a virtual environment. The major advantage of employing an overarching ontological framework is that its asserted facts can be stored alongside the semantics of directions, contextual knowledge, and annotated activity models in one central knowledge base. This allows for a unified and efficient knowledge retrieval across all pipeline components, providing flexibility and reasoning capabilities as symbolic knowledge is combined with annotated sub-symbolic models.

Purpose Deep neural networks need to be able to indicate error likelihood via reliable estimates of their predictive uncertainty when used in high-risk scenarios, such as medical decision support. This work contributes a systematic overview of state-of-the-art approaches for decomposing predictive uncertainty into aleatoric and epistemic components, and a comprehensive comparison for Bayesian neural networks (BNNs) between mutual information decomposition and the explicit modelling of both uncertainty types via an additional loss-attenuating neuron. Methods Experiments are performed in the context of liver segmentation in CT scans. The quality of the uncertainty decomposition in the resulting uncertainty maps is qualitatively evaluated, and quantitative behaviour of decomposed uncertainties is systematically compared for different experiment settings with varying training set sizes, label noise, and distribution shifts. Results Our results show the mutual information decomposition to robustly yield meaningful aleatoric and epistemic uncertainty estimates, while the activation of the loss-attenuating neuron appears noisier with non-trivial convergence properties. We found that the addition of a heteroscedastic neuron does not significantly improve segmentation performance or calibration, while slightly improving the quality of uncertainty estimates. Conclusions Mutual information decomposition is simple to implement, has mathematically pleasing properties, and yields meaningful uncertainty estimates that behave as expected under controlled changes to our data set. The additional extension of BNNs with loss-attenuating neurons provides no improvement in terms of segmentation performance or calibration in our setting, but marginal benefits regarding the quality of decomposed uncertainties.