Topic models, an unsupervised technique for inferring translation domains improve machine translation quality. However, previous work uses only the source language and completely ignores the target language, which can disambiguate domains. We propose new polylingual tree-based topic models to extract domain knowledge that considers both source and target languages and derive three different inference schemes. We evaluate our model on a Chinese to English translation task and obtain up to 1.2 BLEU improvement over strong baselines.

In this paper, we describe and evaluate a bigram part-of-speech (POS) tagger that uses latent annotations and then investigate using additional genre-matched unlabeled data for self-training the tagger. The use of latent annotations substantially improves the performance of a baseline HMM bigram tagger, outperforming a trigram HMM tagger with sophisticated smoothing. The performance of the latent tagger is further enhanced by self-training with a large set of unlabeled data, even in situations where standard bigram or trigram taggers do not benefit from selftraining when trained on greater amounts of labeled training data. Our best model obtains a state-of-the-art Chinese tagging accuracy of 94.78% when evaluated on a representative test set of the Penn Chinese Treebank 6.0.

Predicting how Congressional legislators will vote is important for understanding their past and future behavior. However, previous work on roll-call prediction has been limited to single session settings, thus did not consider generalization across sessions. In this paper, we show that metadata is crucial for modeling voting outcomes in new contexts, as changes between sessions lead to changes in the underlying data generation process. We show how augmenting bill text with the sponsors' ideologies in a neural network model can achieve an average of a 4% boost in accuracy over the previous stateof-the-art. 1 A session is a 2-year period of legislative business.

Automatic summarization methods have been studied on a variety of domains, including news and scientific articles. Yet, legislation has not previously been considered for this task, despite US Congress and state governments releasing tens of thousands of bills every year.In this paper, we introduce BillSum, the first dataset for summarization of US Congressional and California state bills (https://github.com/ FiscalNote/BillSum). We explain the properties of the dataset that make it more challenging to process than other domains. Then, we benchmark extractive methods that consider neural sentence representations and traditional contextual features. Finally, we demonstrate that models built on Congressional bills can be used to summarize California bills, thus, showing that methods developed on this dataset can transfer to states without human-written summaries.

Many applications in NLP, such as questionanswering and summarization, either require or would greatly benefit from the knowledge of when an event occurred. Creating an effective algorithm for identifying the activity time of an event in news is difficult in part because of the sparsity of explicit temporal expressions. This paper describes a domain-independent machine-learning based approach to assign activity times to events in news. We demonstrate that by applying topic models to text, we are able to cluster sentences that describe the same event, and utilize the temporal information within these event clusters to infer activity times for all sentences. Experimental evidence suggests that this is a promising approach, given evaluations performed on three distinct news article sets against the baseline of assigning the publication date. Our approach achieves 90%, 88.7%, and 68.7% accuracy, respectively, outperforming the baseline twice.