• Digital Humanities and Cultural Heritage
• Publications close
• Research data close
• Other research products close
• Conference object close
• FR close
• Edinburgh Research Explorer close
• Digital Humanities and Cultural Her... close

Machine translation (MT) models used in industries with constantly changing topics, such as translation or news agencies, need to adapt to new data to maintain their performance over time. Our aim is to teach a pre-trained MT model to translate previously unseen words accurately, based on very few examples. We propose (i) an experimental setup allowing us to simulate novel vocabulary appearing in human-submitted translations, and (ii) corresponding evaluation metrics to compare our approaches. We extend a data augmentation approach using a pre-trained language model to create training examples with similar contexts for novel words. We compare different fine-tuning and data augmentation approaches and show that adaptation on the scale of one to five examples is possible. Combining data augmentation with randomly selected training sentences leads to the highest BLEU score and accuracy improvements. Impressively, with only 1 to 5 examples, our model reports better accuracy scores than a reference system trained with on average 313 parallel examples. 14 pages includince 3 of appendices

We investigate a long-perceived shortcoming in the typical use of BLEU: its reliance on a single reference. Using modern neural paraphrasing techniques, we study whether automatically generating additional diverse references can provide better coverage of the space of valid translations and thereby improve its correlation with human judgments. Our experiments on the into-English language directions of the WMT19 metrics task (at both the system and sentence level) show that using paraphrased references does generally improve BLEU, and when it does, the more diverse the better. However, we also show that better results could be achieved if those paraphrases were to specifically target the parts of the space most relevant to the MT outputs being evaluated. Moreover, the gains remain slight even when human paraphrases are used, suggesting inherent limitations to BLEU's capacity to correctly exploit multiple references. Surprisingly, we also find that adequacy appears to be less important, as shown by the high results of a strong sampling approach, which even beats human paraphrases when used with sentence-level BLEU. Comment: Accepted in the Findings of EMNLP 2020

The University of Edinburgh participated in the WMT19 Shared Task on News Translation in six language directions: English-to-Gujarati, Gujarati-to-English, English-to-Chinese, Chinese-to-English, German-to-English, and English-to-Czech. For all translation directions, we created or used back-translations of monolingual data in the target language as additional synthetic training data. For English-Gujarati, we also explored semi-supervised MT with cross-lingual language model pre-training, and translation pivoting through Hindi. For translation to and from Chinese, we investigated character-based tokenisation vs. sub-word segmentation of Chinese text. For German-to-English, we studied the impact of vast amounts of back-translated training data on translation quality, gaining a few additional insights over Edunov et al. (2018). For English-to-Czech, we compared different pre-processing and tokenisation regimes. To appear in the Proceedings of WMT19: Shared Task Papers

Possibilistic networks are powerful graphical uncertainty representations based on possibility theory. This paper analyzes the computational complexity of querying min-based and product-based possibilistic networks. It particularly focuses on a very common kind of queries: computing maximum a posteriori explanation (MAP). The main result of the paper is to show that the decision problem of answering MAP queries in both min-based and product-based possibilistic networks is NP-complete. Such computational complexity results represent an advantage of possibilistic networks over probabilistic networks since MAP querying is NP^PP -complete in probabilistic Bayesian networks. We provide the proof based on reduction from the 3SAT decision problem to MAP querying possibilistic networks decision problem. As well as reductions that are useful for implementation of MAP queries using SAT solvers.

The accuracy of the predictions of the γ flux produced by a classical nova during the first hours after the outburst is limited by the uncertainties on several reaction rates, including the 18F(p,α)15O one. Better constraints on this reaction rate can be obtained by determining the spectroscopic properties of the compound nucleus 19Ne. This was achieved in a new inelastic scattering method using a 19Ne radioactive beam (produced by the GANIL-SPIRAL 1 facility) impinging onto a proton target. The experiment was performed at the VAMOS spectrometer. In this article the performances (excitation energy range covered and excitation energy resolution) and limitations of the new technique are discussed. Excitation energy resolution of σ = 33 keV and low background were obtained with this inverse kinematics method, which will allow extracting the spectroscopic properties of 19Ne. International audience

Document modeling is essential to a variety of natural language understanding tasks. We propose to use external information to improve document modeling for problems that can be framed as sentence extraction. We develop a framework composed of a hierarchical document encoder and an attention-based extractor with attention over external information. We evaluate our model on extractive document summarization (where the external information is image captions and the title of the document) and answer selection (where the external information is a question). We show that our model consistently outperforms strong baselines, in terms of both informativeness and fluency (for CNN document summarization) and achieves state-of-the-art results for answer selection on WikiQA and NewsQA.11Our TensorFlow code and datasets are publicly available at https://github.com/shashiongithub/Document-Models-with-Ext Information. 

We introduce a novel multi-source technique for incorporating source syntax into neural machine translation using linearized parses. This is achieved by employing separate encoders for the sequential and parsed versions of the same source sentence; the resulting representations are then combined using a hierarchical attention mechanism. The proposed model improves over both seq2seq and parsed baselines by over 1 BLEU on the WMT17 English-German task. Further analysis shows that our multi-source syntactic model is able to translate successfully without any parsed input, unlike standard parsed methods. In addition, performance does not deteriorate as much on long sentences as for the baselines. Comment: EMNLP 2018