Universal Dependencies (UD) is becoming a standard annotation scheme cross-linguistically, but it is argued that this scheme centering on content words is harder to parse than the conventional one centering on function words. To improve the parsability of UD, we propose a back-and-forth conversion algorithm, in which we preprocess the training treebank to increase parsability, and reconvert the parser outputs to follow the UD scheme as a postprocess. We show that this technique consistently improves LAS across languages even with a state-of-the-art parser, in particular on core dependency arcs such as nominal modifier. We also provide an in-depth analysis to understand why our method increases parsability.

We introduce the URIEL knowledge base for massively multilingual NLP and the lang2vec utility, which provides information-rich vector identifications of languages drawn from typological, geographical, and phylogenetic databases and normalized to have straightforward and consistent formats, naming, and semantics. The goal of URIEL and lang2vec is to enable multilingual NLP, especially on less-resourced languages and make possible types of experiments (especially but not exclusively related to NLP tasks) that are otherwise difficult or impossible due to the sparsity and incommensurability of the data sources. lang2vec vectors have been shown to reduce perplexity in multilingual language modeling, when compared to one-hot language identification vectors.

Noise Contrastive Estimation (NCE) is a learning procedure that is regularly used to train neural language models, since it avoids the computational bottleneck caused by the output softmax. In this paper, we attempt to explain some of the weaknesses of this objective function, and to draw directions for further developments. Experiments on a small task show the issues raised by an unigram noise distribution, and that a context dependent noise distribution, such as the bigram distribution, can solve these issues and provide stable and data-efficient learning.

Deep neural networks have achieved remarkable results across many language processing tasks, however they have been shown to be susceptible to overfitting and highly sensitive to noise, including adversarial attacks. In this work, we propose a linguistically-motivated approach for training robust models based on exposing the model to corrupted text examples at training time. We consider several flavours of linguistically plausible corruption, include lexical semantic and syntactic methods. Empirically, we evaluate our method with a convolutional neural model across a range of sentiment analysis datasets. Compared with a baseline and the dropout method, our method achieves better overall performance.

We explore whether social media can provide a window into community real estate -foreclosure rates and price changes- beyond that of traditional economic and demographic variables. We find language use in Twitter not only predicts real estate outcomes as well as traditional variables across counties, but that including Twitter language in traditional models leads to a significant improvement (e.g. from Pearson r = :50 to r = :59 for price changes). We overcome the challenge of the relative sparsity and noise in Twitter language variables by showing that training on the residual error of the traditional models leads to more accurate overall assessments. Finally, we discover that it is Twitter language related to business (e.g. ‘company’, ‘marketing’) and technology (e.g. ‘technology’, ‘internet’), among others, that yield predictive power over economics.

We present a new Lexical Simplification approach that exploits Neural Networks to learn substitutions from the Newsela corpus - a large set of professionally produced simplifications. We extract candidate substitutions by combining the Newsela corpus with a retrofitted context-aware word embeddings model and rank them using a new neural regression model that learns rankings from annotated data. This strategy leads to the highest Accuracy, Precision and F1 scores to date in standard datasets for the task.

This paper discusses some central caveats of summarisation, incurred in the use of the ROUGE metric for evaluation, with respect to optimal solutions. The task is NP-hard, of which we give the first proof. Still, as we show empirically for three central benchmark datasets for the task, greedy algorithms empirically seem to perform optimally according to the metric. Additionally, overall quality assurance is problematic: there is no natural upper bound on the quality of summarisation systems, and even humans are excluded from performing optimal summarisation.

We present an iterative annotation process for producing aligned, parallel corpora of abstractive and extractive summaries for narrative. Our approach uses a combination of trained annotators and crowd-sourcing, allowing us to elicit human-generated summaries and alignments quickly and at low cost. We use crowd-sourcing to annotate aligned phrases with the text-to-text generation techniques needed to transform each phrase into the other. We apply this process to a corpus of 476 personal narratives, which we make available on the Web.

Progress in text understanding has been driven by large datasets that test particular capabilities, like recent datasets for reading comprehension (Hermann et al., 2015). We focus here on the LAMBADA dataset (Paperno et al., 2016), a word prediction task requiring broader context than the immediate sentence. We view LAMBADA as a reading comprehension problem and apply comprehension models based on neural networks. Though these models are constrained to choose a word from the context, they improve the state of the art on LAMBADA from 7.3% to 49%. We analyze 100 instances, finding that neural network readers perform well in cases that involve selecting a name from the context based on dialogue or discourse cues but struggle when coreference resolution or external knowledge is needed.

Several corpora have been annotated with negation scope—the set of words whose meaning is negated by a cue like the word “not”—leading to the development of classifiers that detect negation scope with high accuracy. We show that for nearly all of these corpora, this high accuracy can be attributed to a single fact: they frequently annotate negation scope as a single span of text delimited by punctuation. For negation scopes not of this form, detection accuracy is low and under-sampling the easy training examples does not substantially improve accuracy. We demonstrate that this is partly an artifact of annotation guidelines, and we argue that future negation scope annotation efforts should focus on these more difficult cases.

Cross-lingual information extraction is the task of distilling facts from foreign language (e.g. Chinese text) into representations in another language that is preferred by the user (e.g. English tuples). Conventional pipeline solutions decompose the task as machine translation followed by information extraction (or vice versa). We propose a joint solution with a neural sequence model, and show that it outperforms the pipeline in a cross-lingual open information extraction setting by 1-4 BLEU and 0.5-0.8 F1.

We learn a mapping that negates adjectives by predicting an adjective’s antonym in an arbitrary word embedding model. We show that both linear models and neural networks improve on this task when they have access to a vector representing the semantic domain of the input word, e.g. a centroid of temperature words when predicting the antonym of ‘cold’. We introduce a continuous class-conditional bilinear neural network which is able to negate adjectives with high precision.

Instances (“Mozart”) are ontologically distinct from concepts or classes (“composer”). Natural language encompasses both, but instances have received comparatively little attention in distributional semantics. Our results show that instances and concepts differ in their distributional properties. We also establish that instantiation detection (“Mozart – composer”) is generally easier than hypernymy detection (“chemist – scientist”), and that results on the influence of input representation do not transfer from hyponymy to instantiation.

There has recently been a lot of work trying to use images of referents of words for improving vector space meaning representations derived from text. We investigate the opposite direction, as it were, trying to improve visual word predictors that identify objects in images, by exploiting distributional similarity information during training. We show that for certain words (such as entry-level nouns or hypernyms), we can indeed learn better referential word meanings by taking into account their semantic similarity to other words. For other words, there is no or even a detrimental effect, compared to a learning setup that presents even semantically related objects as negative instances.

We propose the semantic proto-role linking model, which jointly induces both predicate-specific semantic roles and predicate-general semantic proto-roles based on semantic proto-role property likelihood judgments. We use this model to empirically evaluate Dowty’s thematic proto-role linking theory.

There is a relationship between what we say and where we say it. Word embeddings are usually trained assuming that semantically-similar words occur within the same textual contexts. We investigate the extent to which semantically-similar words occur within the same geospatial contexts. We enrich a corpus of geolocated Twitter posts with physical data derived from Google Places and OpenStreetMap, and train word embeddings using the resulting geospatial contexts. Intrinsic evaluation of the resulting vectors shows that geographic context alone does provide useful information about semantic relatedness.

Emojis are ideograms which are naturally combined with plain text to visually complement or condense the meaning of a message. Despite being widely used in social media, their underlying semantics have received little attention from a Natural Language Processing standpoint. In this paper, we investigate the relation between words and emojis, studying the novel task of predicting which emojis are evoked by text-based tweet messages. We train several models based on Long Short-Term Memory networks (LSTMs) in this task. Our experimental results show that our neural model outperforms a baseline as well as humans solving the same task, suggesting that computational models are able to better capture the underlying semantics of emojis.

A traditional claim in linguistics is that all human languages are equally expressive—able to convey the same wide range of meanings. Morphologically rich languages, such as Czech, rely on overt inflectional and derivational morphology to convey many semantic distinctions. Languages with comparatively limited morphology, such as English, should be able to accomplish the same using a combination of syntactic and contextual cues. We capitalize on this idea by training a tagger for English that uses syntactic features obtained by automatic parsing to recover complex morphological tags projected from Czech. The high accuracy of the resulting model provides quantitative confirmation of the underlying linguistic hypothesis of equal expressivity, and bodes well for future improvements in downstream HLT tasks including machine translation.

Derivational morphology is a fundamental and complex characteristic of language. In this paper we propose a new task of predicting the derivational form of a given base-form lemma that is appropriate for a given context. We present an encoder-decoder style neural network to produce a derived form character-by-character, based on its corresponding character-level representation of the base form and the context. We demonstrate that our model is able to generate valid context-sensitive derivations from known base forms, but is less accurate under lexicon agnostic setting.

What is the information captured by neural network models of language? We address this question in the case of character-level recurrent neural language models. These models do not have explicit word representations; do they acquire implicit ones? We assess the lexical capacity of a network using the lexical decision task common in psycholinguistics: the system is required to decide whether or not a string of characters forms a word. We explore how accuracy on this task is affected by the architecture of the network, focusing on cell type (LSTM vs. SRN), depth and width. We also compare these architectural properties to a simple count of the parameters of the network. The overall number of parameters in the network turns out to be the most important predictor of accuracy; in particular, there is little evidence that deeper networks are beneficial for this task.

In this paper we pursue the hypothesis that the distribution of article omission specifically is constrained by principles of Information Theory (Shannon 1948). In particular, Information Theory predicts a stronger preference for article omission before nouns which are relatively unpredictable in context of the preceding words. We investigated article omission in German newspaper headlines with a corpus and acceptability rating study. Both support our hypothesis: Articles are inserted more often before unpredictable nouns and subjects perceive article omission before predictable nouns as more well-formed than before unpredictable ones. This suggests that information theoretic principles constrain the distribution of article omission in headlines.

We present a computational analysis of the language of drug users when talking about their drug experiences. We introduce a new dataset of over 4,000 descriptions of experiences reported by users of four main drug types, and show that we can predict with an F1-score of up to 88% the drug behind a certain experience. We also perform an analysis of the dominant psycholinguistic processes and dominant emotions associated with each drug type, which sheds light on the characteristics of drug users.

Latent structured prediction theory proposes powerful methods such as Latent Structural SVM (LSSVM), which can potentially be very appealing for coreference resolution (CR). In contrast, only small work is available, mainly targeting the latent structured perceptron (LSP). In this paper, we carried out a practical study comparing for the first time online learning with LSSVM. We analyze the intricacies that may have made initial attempts to use LSSVM fail, i.e., a huge training time and much lower accuracy produced by Kruskal’s spanning tree algorithm. In this respect, we also propose a new effective feature selection approach for improving system efficiency. The results show that LSP, if correctly parameterized, produces the same performance as LSSVM, being much more efficient.

The task of implicit discourse relation classification has received increased attention in recent years, including two CoNNL shared tasks on the topic. Existing machine learning models for the task train on sections 2-21 of the PDTB and test on section 23, which includes a total of 761 implicit discourse relations. In this paper, we’d like to make a methodological point, arguing that the standard test set is too small to draw conclusions about whether the inclusion of certain features constitute a genuine improvement, or whether one got lucky with some properties of the test set, and argue for the adoption of cross validation for the discourse relation classification task by the community.

We study the topmost weight matrix of neural network language models. We show that this matrix constitutes a valid word embedding. When training language models, we recommend tying the input embedding and this output embedding. We analyze the resulting update rules and show that the tied embedding evolves in a more similar way to the output embedding than to the input embedding in the untied model. We also offer a new method of regularizing the output embedding. Our methods lead to a significant reduction in perplexity, as we are able to show on a variety of neural network language models. Finally, we show that weight tying can reduce the size of neural translation models to less than half of their original size without harming their performance.

Multi-task learning (MTL) in deep neural networks for NLP has recently received increasing interest due to some compelling benefits, including its potential to efficiently regularize models and to reduce the need for labeled data. While it has brought significant improvements in a number of NLP tasks, mixed results have been reported, and little is known about the conditions under which MTL leads to gains in NLP. This paper sheds light on the specific task relations that can lead to gains from MTL models over single-task setups.

We present a novel, unsupervised, and distance measure agnostic method for search space reduction in spell correction using neural character embeddings. The embeddings are learned by skip-gram word2vec training on sequences generated from dictionary words in a phonetic information-retentive manner. We report a very high performance in terms of both success rates and reduction of search space on the Birkbeck spelling error corpus. To the best of our knowledge, this is the first application of word2vec to spell correction.

The popular skip-gram model induces word embeddings by exploiting the signal from word-context coocurrence. We offer a new interpretation of skip-gram based on exponential family PCA-a form of matrix factorization to generalize the skip-gram model to tensor factorization. In turn, this lets us train embeddings through richer higher-order coocurrences, e.g., triples that include positional information (to incorporate syntax) or morphological information (to share parameters across related words). We experiment on 40 languages and show our model improves upon skip-gram.

We present a dialogue generation model that directly captures the variability in possible responses to a given input, which reduces the ‘boring output’ issue of deterministic dialogue models. Experiments show that our model generates more diverse outputs than baseline models, and also generates more consistently acceptable output than sampling from a deterministic encoder-decoder model.

Children comprise a significant proportion of TV viewers and it is worthwhile to customize the experience for them. However, identifying who is a child in the audience can be a challenging task. We present initial studies of a novel method which combines utterances with user metadata. In particular, we develop an ensemble of different machine learning techniques on different subsets of data to improve child detection. Our initial results show an 9.2% absolute improvement over the baseline, leading to a state-of-the-art performance.

In this work, we tackle a problem of speech emotion classification. One of the issues in the area of affective computation is that the amount of annotated data is very limited. On the other hand, the number of ways that the same emotion can be expressed verbally is enormous due to variability between speakers. This is one of the factors that limits performance and generalization. We propose a simple method that extracts audio samples from movies using textual sentiment analysis. As a result, it is possible to automatically construct a larger dataset of audio samples with positive, negative emotional and neutral speech. We show that pretraining recurrent neural network on such a dataset yields better results on the challenging EmotiW corpus. This experiment shows a potential benefit of combining textual sentiment analysis with vocal information.

On-line dialogue policy learning is the key for building evolvable conversational agent in real world scenarios. Poor initial policy can easily lead to bad user experience and consequently fail to attract sufficient users for policy training. A novel framework, companion teaching, is proposed to include a human teacher in the dialogue policy training loop to address the cold start problem. Here, dialogue policy is trained using not only user’s reward, but also teacher’s example action as well as estimated immediate reward at turn level. Simulation experiments showed that, with small number of human teaching dialogues, the proposed approach can effectively improve user experience at the beginning and smoothly lead to good performance with more user interaction data.

This paper presents a hybrid dialog state tracker enhanced by trainable Spoken Language Understanding (SLU) for slot-filling dialog systems. Our architecture is inspired by previously proposed neural-network-based belief-tracking systems. In addition, we extended some parts of our modular architecture with differentiable rules to allow end-to-end training. We hypothesize that these rules allow our tracker to generalize better than pure machine-learning based systems. For evaluation, we used the Dialog State Tracking Challenge (DSTC) 2 dataset - a popular belief tracking testbed with dialogs from restaurant information system. To our knowledge, our hybrid tracker sets a new state-of-the-art result in three out of four categories within the DSTC2.

The task of morphological analysis is to produce a complete list of lemma+tag analyses for a given word-form. We propose a discriminative string transduction approach which exploits plain inflection tables and raw text corpora, thus obviating the need for expert annotation. Experiments on four languages demonstrate that our system has much higher coverage than a hand-engineered FST analyzer, and is more accurate than a state-of-the-art morphological tagger.

The Dravidian languages are one of the most widely spoken language families in the world, yet there are very few annotated resources available to NLP researchers. To remedy this, we create DravMorph, a corpus annotated for morphological segmentation and part-of-speech. Additionally, we exploit novel features and higher-order models to set state-of-the-art results on these corpora on both tasks, beating techniques proposed in the literature by as much as 4 points in segmentation F1.

In this paper we present BabelDomains, a unified resource which provides lexical items with information about domains of knowledge. We propose an automatic method that uses knowledge from various lexical resources, exploiting both distributional and graph-based clues, to accurately propagate domain information. We evaluate our methodology intrinsically on two lexical resources (WordNet and BabelNet), achieving a precision over 80% in both cases. Finally, we show the potential of BabelDomains in a supervised learning setting, clustering training data by domain for hypernym discovery.

We present a new parallel corpus, JHU FLuency-Extended GUG corpus (JFLEG) for developing and evaluating grammatical error correction (GEC). Unlike other corpora, it represents a broad range of language proficiency levels and uses holistic fluency edits to not only correct grammatical errors but also make the original text more native sounding. We describe the types of corrections made and benchmark four leading GEC systems on this corpus, identifying specific areas in which they do well and how they can improve. JFLEG fulfills the need for a new gold standard to properly assess the current state of GEC.

We present Arab-Acquis, a large publicly available dataset for evaluating machine translation between 22 European languages and Arabic. Arab-Acquis consists of over 12,000 sentences from the JRC-Acquis (Acquis Communautaire) corpus translated twice by professional translators, once from English and once from French, and totaling over 600,000 words. The corpus follows previous data splits in the literature for tuning, development, and testing. We describe the corpus and how it was created. We also present the first benchmarking results on translating to and from Arabic for 22 European languages.

The Parallel Meaning Bank is a corpus of translations annotated with shared, formal meaning representations comprising over 11 million words divided over four languages (English, German, Italian, and Dutch). Our approach is based on cross-lingual projection: automatically produced (and manually corrected) semantic annotations for English sentences are mapped onto their word-aligned translations, assuming that the translations are meaning-preserving. The semantic annotation consists of five main steps: (i) segmentation of the text in sentences and lexical items; (ii) syntactic parsing with Combinatory Categorial Grammar; (iii) universal semantic tagging; (iv) symbolization; and (v) compositional semantic analysis based on Discourse Representation Theory. These steps are performed using statistical models trained in a semi-supervised manner. The employed annotation models are all language-neutral. Our first results are promising.

We address the challenge of cross-lingual POS tagger evaluation in absence of manually annotated test data. We put forth and evaluate two dictionary-based metrics. On the tasks of accuracy prediction and system ranking, we reveal that these metrics are reliable enough to approximate test set-based evaluation, and at the same time lean enough to support assessment for truly low-resource languages.

In this paper, we present a Wikipedia-based approach to develop resources for the legal domain. We establish a mapping between a legal domain ontology, LKIF (Hoekstra et al. 2007), and a Wikipedia-based ontology, YAGO (Suchanek et al. 2007), and through that we populate LKIF. Moreover, we use the mentions of those entities in Wikipedia text to train a specific Named Entity Recognizer and Classifier. We find that this classifier works well in the Wikipedia, but, as could be expected, performance decreases in a corpus of judgments of the European Court of Human Rights. However, this tool will be used as a preprocess for human annotation. We resort to a technique called “curriculum learning” aimed to overcome problems of overfitting by learning increasingly more complex concepts. However, we find that in this particular setting, the method works best by learning from most specific to most general concepts, not the other way round.

This paper presents a new resource, called Content Types Dataset, to promote the analysis of texts as a composition of units with specific semantic and functional roles. By developing this dataset, we also introduce a new NLP task for the automatic classification of Content Types. The annotation scheme and the dataset are described together with two sets of classification experiments.

This paper presents work on using continuous representations for authorship attribution. In contrast to previous work, which uses discrete feature representations, our model learns continuous representations for n-gram features via a neural network jointly with the classification layer. Experimental results demonstrate that the proposed model outperforms the state-of-the-art on two datasets, while producing comparable results on the remaining two.

In this paper, we address the (to the best of our knowledge) new problem of extracting a structured description of real estate properties from their natural language descriptions in classifieds. We survey and present several models to (a) identify important entities of a property (e.g.,rooms) from classifieds and (b) structure them into a tree format, with the entities as nodes and edges representing a part-of relation. Experiments show that a graph-based system deriving the tree from an initially fully connected entity graph, outperforms a transition-based system starting from only the entity nodes, since it better reconstructs the tree.

State-of-the-art neural machine translation (NMT) systems are generally trained on specific domains by carefully selecting the training sets and applying proper domain adaptation techniques. In this paper we consider the real world scenario in which the target domain is not predefined, hence the system should be able to translate text from multiple domains. We compare the performance of a generic NMT system and phrase-based statistical machine translation (PBMT) system by training them on a generic parallel corpus composed of data from different domains. Our results on multi-domain English-French data show that, in these realistic conditions, PBMT outperforms its neural counterpart. This raises the question: is NMT ready for deployment as a generic/multi-purpose MT backbone in real-world settings?

Current evaluation metrics for timeline summarization either ignore the temporal aspect of the task or require strict date matching. We introduce variants of ROUGE that allow alignment of daily summaries via temporal distance or semantic similarity. We argue for the suitability of these variants in a theoretical analysis and demonstrate it in a battery of task-specific tests.

This paper tackles the reduction of redundant repeating generation that is often observed in RNN-based encoder-decoder models. Our basic idea is to jointly estimate the upper-bound frequency of each target vocabulary in the encoder and control the output words based on the estimation in the decoder. Our method shows significant improvement over a strong RNN-based encoder-decoder baseline and achieved its best results on an abstractive summarization benchmark.

Musical parody, i.e. the act of changing the lyrics of an existing and very well-known song, is a commonly used technique for creating catchy advertising tunes and for mocking people or events. Here we describe a system for automatically producing a musical parody, starting from a corpus of songs. The system can automatically identify characterizing words and concepts related to a novel text, which are taken from the daily news. These concepts are then used as seeds to appropriately replace part of the original lyrics of a song, using metrical, rhyming and lexical constraints. Finally, the parody can be sung with a singing speech synthesizer, with no intervention from the user.

This paper presents an efficient and optimal parsing algorithm for probabilistic context-free grammars (PCFGs). To achieve faster parsing, our proposal employs a pruning technique to reduce unnecessary edges in the search space. The key is to conduct repetitively Viterbi inside and outside parsing, while gradually expanding the search space to efficiently compute heuristic bounds used for pruning. Our experimental results using the English Penn Treebank corpus show that the proposed algorithm is faster than the standard CKY parsing algorithm. In addition, we also show how to extend this algorithm to extract k-best Viterbi parse trees.

Prepostitional phrase (PP) attachment is a well known challenge to parsing. In this paper, we combine the insights of different works, namely: (1) treating PP attachment as a classification task with an arbitrary number of attachment candidates; (2) using auxiliary distributions to augment the data beyond the hand-annotated training set; (3) using topological fields to get information about the distribution of PP attachment throughout clauses and (4) using state-of-the-art techniques such as word embeddings and neural networks. We show that jointly using these techniques leads to substantial improvements. We also conduct a qualitative analysis to gauge where the ceiling of the task is in a realistic setup.

This paper formalizes a sound extension of dynamic oracles to global training, in the frame of transition-based dependency parsers. By dispensing with the pre-computation of references, this extension widens the training strategies that can be entertained for such parsers; we show this by revisiting two standard training procedures, early-update and max-violation, to correct some of their search space sampling biases. Experimentally, on the SPMRL treebanks, this improvement increases the similarity between the train and test distributions and yields performance improvements up to 0.7 UAS, without any computation overhead.

In this paper, we propose efficient and less resource-intensive strategies for parsing of code-mixed data. These strategies are not constrained by in-domain annotations, rather they leverage pre-existing monolingual annotated resources for training. We show that these methods can produce significantly better results as compared to an informed baseline. Due to lack of an evaluation set for code-mixed structures, we also present a data set of 450 Hindi and English code-mixed tweets of Hindi multilingual speakers for evaluation.

We introduce a constituency parser based on a bi-LSTM encoder adapted from recent work (Cross and Huang, 2016b; Kiperwasser and Goldberg, 2016), which can incorporate a lower level character biLSTM (Ballesteros et al., 2015; Plank et al., 2016). We model two important interfaces of constituency parsing with auxiliary tasks supervised at the word level: (i) part-of-speech (POS) and morphological tagging, (ii) functional label prediction. On the SPMRL dataset, our parser obtains above state-of-the-art results on constituency parsing without requiring either predicted POS or morphological tags, and outputs labelled dependency trees.

People can refer to quantities in a visual scene by using either exact cardinals (e.g. one, two, three) or natural language quantifiers (e.g. few, most, all). In humans, these two processes underlie fairly different cognitive and neural mechanisms. Inspired by this evidence, the present study proposes two models for learning the objective meaning of cardinals and quantifiers from visual scenes containing multiple objects. We show that a model capitalizing on a ‘fuzzy’ measure of similarity is effective for learning quantifiers, whereas the learning of exact cardinals is better accomplished when information about number is provided.

While dependency parsers reach very high overall accuracy, some dependency relations are much harder than others. In particular, dependency parsers perform poorly in coordination construction (i.e., correctly attaching the conj relation). We extend a state-of-the-art dependency parser with conjunction-specific features, focusing on the similarity between the conjuncts head words. Training the extended parser yields an improvement in conj attachment as well as in overall dependency parsing accuracy on the Stanford dependency conversion of the Penn TreeBank.

We present a second-stage machine translation (MT) system based on a neural machine translation (NMT) approach to automatic post-editing (APE) that improves the translation quality provided by a first-stage MT system. Our APE system (APE_Sym) is an extended version of an attention based NMT model with bilingual symmetry employing bidirectional models, mt–pe and pe–mt. APE translations produced by our system show statistically significant improvements over the first-stage MT, phrase-based APE and the best reported score on the WMT 2016 APE dataset by a previous neural APE system. Re-ranking (APE_Rerank) of the n-best translations from the phrase-based APE and APE_Sym systems provides further substantial improvements over the symmetric neural APE model. Human evaluation confirms that the APE_Rerank generated PE translations improve on the previous best neural APE system at WMT 2016.

Meaningful conclusions about the relative performance of NLP systems are only possible if the gold standard employed in a given evaluation is both valid and reliable. In this paper, we explore the validity of human annotations currently employed in the evaluation of document-level quality estimation for machine translation (MT). We demonstrate the degree to which MT system rankings are dependent on weights employed in the construction of the gold standard, before proposing direct human assessment as a valid alternative. Experiments show direct assessment (DA) scores for documents to be highly reliable, achieving a correlation of above 0.9 in a self-replication experiment, in addition to a substantial estimated cost reduction through quality controlled crowd-sourcing. The original gold standard based on post-edits incurs a 10–20 times greater cost than DA.

We present a novel scheme to combine neural machine translation (NMT) with traditional statistical machine translation (SMT). Our approach borrows ideas from linearised lattice minimum Bayes-risk decoding for SMT. The NMT score is combined with the Bayes-risk of the translation according the SMT lattice. This makes our approach much more flexible than n-best list or lattice rescoring as the neural decoder is not restricted to the SMT search space. We show an efficient and simple way to integrate risk estimation into the NMT decoder which is suitable for word-level as well as subword-unit-level NMT. We test our method on English-German and Japanese-English and report significant gains over lattice rescoring on several data sets for both single and ensembled NMT. The MBR decoder produces entirely new hypotheses far beyond simply rescoring the SMT search space or fixing UNKs in the NMT output.

Translating into morphologically rich languages is difficult. Although the coverage of lemmas may be reasonable, many morphological variants cannot be learned from the training data. We present a statistical translation system that is able to produce these inflected word forms. Different from most previous work, we do not separate morphological prediction from lexical choice into two consecutive steps. Our approach is novel in that it is integrated in decoding and takes advantage of context information from both the source language and the target language sides.

Analysing translation quality in regards to specific linguistic phenomena has historically been difficult and time-consuming. Neural machine translation has the attractive property that it can produce scores for arbitrary translations, and we propose a novel method to assess how well NMT systems model specific linguistic phenomena such as agreement over long distances, the production of novel words, and the faithful translation of polarity. The core idea is that we measure whether a reference translation is more probable under a NMT model than a contrastive translation which introduces a specific type of error. We present LingEval97, a large-scale data set of 97000 contrastive translation pairs based on the WMT English->German translation task, with errors automatically created with simple rules. We report results for a number of systems, and find that recently introduced character-level NMT systems perform better at transliteration than models with byte-pair encoding (BPE) segmentation, but perform more poorly at morphosyntactic agreement, and translating discontiguous units of meaning.

Knowing which words have been attended to in previous time steps while generating a translation is a rich source of information for predicting what words will be attended to in the future. We improve upon the attention model of Bahdanau et al. (2014) by explicitly modeling the relationship between previous and subsequent attention levels for each word using one recurrent network per input word. This architecture easily captures informative features, such as fertility and regularities in relative distortion. In experiments, we show our parameterization of attention improves translation quality.

We put forward an approach that exploits the knowledge encoded in lexical resources in order to induce representations for words that were not encountered frequently during training. Our approach provides an advantage over the past work in that it enables vocabulary expansion not only for morphological variations, but also for infrequent domain specific terms. We performed evaluations in different settings, showing that the technique can provide consistent improvements on multiple benchmarks across domains.

This paper presents a large-scale evaluation study of dependency-based distributional semantic models. We evaluate dependency-filtered and dependency-structured DSMs in a number of standard semantic similarity tasks, systematically exploring their parameter space in order to give them a “fair shot” against window-based models. Our results show that properly tuned window-based DSMs still outperform the dependency-based models in most tasks. There appears to be little need for the language-dependent resources and computational cost associated with syntactic analysis.

One key property of word embeddings currently under study is their capacity to encode hypernymy. Previous works have used supervised models to recover hypernymy structures from embeddings. However, the overall results do not clearly show how well we can recover such structures. We conduct the first dataset-centric analysis that shows how only the Baroni dataset provides consistent results. We empirically show that a possible reason for its good performance is its alignment to dimensions specific of hypernymy: generality and similarity

We develop a novel cross-lingual word representation model which injects syntactic information through dependency-based contexts into a shared cross-lingual word vector space. The model, termed CL-DepEmb, is based on the following assumptions: (1) dependency relations are largely language-independent, at least for related languages and prominent dependency links such as direct objects, as evidenced by the Universal Dependencies project; (2) word translation equivalents take similar grammatical roles in a sentence and are therefore substitutable within their syntactic contexts. Experiments with several language pairs on word similarity and bilingual lexicon induction, two fundamental semantic tasks emphasising semantic similarity, suggest the usefulness of the proposed syntactically informed cross-lingual word vector spaces. Improvements are observed in both tasks over standard cross-lingual “offline mapping” baselines trained using the same setup and an equal level of bilingual supervision.

This paper proposes to use distributed representation of words (word embeddings) in cross-language textual similarity detection. The main contributions of this paper are the following: (a) we introduce new cross-language similarity detection methods based on distributed representation of words; (b) we combine the different methods proposed to verify their complementarity and finally obtain an overall F1 score of 89.15% for English-French similarity detection at chunk level (88.5% at sentence level) on a very challenging corpus.

We explore the ability of word embeddings to capture both semantic and morphological similarity, as affected by the different types of linguistic properties (surface form, lemma, morphological tag) used to compose the representation of each word. We train several models, where each uses a different subset of these properties to compose its representations. By evaluating the models on semantic and morphological measures, we reveal some useful insights on the relationship between semantics and morphology.

This paper explores a simple and efficient baseline for text classification. Our experiments show that our fast text classifier fastText is often on par with deep learning classifiers in terms of accuracy, and many orders of magnitude faster for training and evaluation. We can train fastText on more than one billion words in less than ten minutes using a standard multicore CPU, and classify half a million sentences among 312K classes in less than a minute.

It is often assumed that topic models benefit from the use of a manually curated stopword list. Constructing this list is time-consuming and often subject to user judgments about what kinds of words are important to the model and the application. Although stopword removal clearly affects which word types appear as most probable terms in topics, we argue that this improvement is superficial, and that topic inference benefits little from the practice of removing stopwords beyond very frequent terms. Removing corpus-specific stopwords after model inference is more transparent and produces similar results to removing those words prior to inference.

Measuring topic quality is essential for scoring the learned topics and their subsequent use in Information Retrieval and Text classification. To measure quality of Latent Dirichlet Allocation (LDA) based topics learned from text, we propose a novel approach based on grouping of topic words into buckets (TBuckets). A single large bucket signifies a single coherent theme, in turn indicating high topic coherence. TBuckets uses word embeddings of topic words and employs singular value decomposition (SVD) and Integer Linear Programming based optimization to create coherent word buckets. TBuckets outperforms the state-of-the-art techniques when evaluated using 3 publicly available datasets and on another one proposed in this paper.

The success of sentence classification often depends on understanding both the syntactic and semantic properties of word-phrases. Recent progress on this task has been based on exploiting the grammatical structure of sentences but often this structure is difficult to parse and noisy. In this paper, we propose a structure-independent ‘Gated Representation Alignment’ (GRA) model that blends a phrase-focused Convolutional Neural Network (CNN) approach with sequence-oriented Recurrent Neural Network (RNN). Our novel alignment mechanism allows the RNN to selectively include phrase information in a word-by-word sentence representation, and to do this without awareness of the syntactic structure. An empirical evaluation of GRA shows higher prediction accuracy (up to 4.6%) of fine-grained sentiment ratings, when compared to other structure-independent baselines. We also show comparable results to several structure-dependent methods. Finally, we analyzed the effect of our alignment mechanism and found that this is critical to the effectiveness of the CNN-RNN hybrid.

Recently, there has been a lot of activity in learning distributed representations of words in vector spaces. Although there are models capable of learning high-quality distributed representations of words, how to generate vector representations of the same quality for phrases or documents still remains a challenge. In this paper, we propose to model each document as a multivariate Gaussian distribution based on the distributed representations of its words. We then measure the similarity between two documents based on the similarity of their distributions. Experiments on eight standard text categorization datasets demonstrate the effectiveness of the proposed approach in comparison with state-of-the-art methods.

In many natural language processing (NLP) tasks, a document is commonly modeled as a bag of words using the term frequency-inverse document frequency (TF-IDF) vector. One major shortcoming of the frequency-based TF-IDF feature vector is that it ignores word orders that carry syntactic and semantic relationships among the words in a document. This paper proposes a novel distributed vector representation of a document, which will be labeled as DV-LSTM, and is derived from the result of adapting a long short-term memory recurrent neural network language model by the document. DV-LSTM is expected to capture some high-level sequential information in the document, which other current document representations fail to do. It was evaluated in document genre classification in the Brown Corpus and the BNC Baby Corpus. The results show that DV-LSTM significantly outperforms TF-IDF vector and paragraph vector (PV-DM) in most cases, and their combinations may further improve the classification performance.

We introduce a novel method to extract keywords from meeting speech in real-time. Our approach builds on the graph-of-words representation of text and leverages the k-core decomposition algorithm and properties of submodular functions. We outperform multiple baselines in a real-time scenario emulated from the AMI and ICSI meeting corpora. Evaluation is conducted against both extractive and abstractive gold standard using two standard performance metrics and a newer one based on word embeddings.

Task-oriented dialogue focuses on conversational agents that participate in dialogues with user goals on domain-specific topics. In contrast to chatbots, which simply seek to sustain open-ended meaningful discourse, existing task-oriented agents usually explicitly model user intent and belief states. This paper examines bypassing such an explicit representation by depending on a latent neural embedding of state and learning selective attention to dialogue history together with copying to incorporate relevant prior context. We complement recent work by showing the effectiveness of simple sequence-to-sequence neural architectures with a copy mechanism. Our model outperforms more complex memory-augmented models by 7% in per-response generation and is on par with the current state-of-the-art on DSTC2, a real-world task-oriented dialogue dataset.

We explore the problem of translating speech to text in low-resource scenarios where neither automatic speech recognition (ASR) nor machine translation (MT) are available, but we have training data in the form of audio paired with text translations. We present the first system for this problem applied to a realistic multi-speaker dataset, the CALLHOME Spanish-English speech translation corpus. Our approach uses unsupervised term discovery (UTD) to cluster repeated patterns in the audio, creating a pseudotext, which we pair with translations to create a parallel text and train a simple bag-of-words MT model. We identify the challenges faced by the system, finding that the difficulty of cross-speaker UTD results in low recall, but that our system is still able to correctly translate some content words in test data.

In this paper we present a comparative evaluation of various negotiation strategies within an online version of the game “Settlers of Catan”. The comparison is based on human subjects playing games against artificial game-playing agents (‘bots’) which implement different negotiation dialogue strategies, using a chat dialogue interface to negotiate trades. Our results suggest that a negotiation strategy that uses persuasion, as well as a strategy that is trained from data using Deep Reinforcement Learning, both lead to an improved win rate against humans, compared to previous rule-based and supervised learning baseline dialogue negotiators.

This paper introduces a new unsupervised approach for dialogue act induction. Given the sequence of dialogue utterances, the task is to assign them the labels representing their function in the dialogue. Utterances are represented as real-valued vectors encoding their meaning. We model the dialogue as Hidden Markov model with emission probabilities estimated by Gaussian mixtures. We use Gibbs sampling for posterior inference. We present the results on the standard Switchboard-DAMSL corpus. Our algorithm achieves promising results compared with strong supervised baselines and outperforms other unsupervised algorithms.

Grounded semantics is typically learnt from utterance-level meaning representations (e.g., successful database retrievals, denoted objects in images, moves in a game). We explore learning word and utterance meanings by continuous observation of the actions of an instruction follower (IF). While an instruction giver (IG) provided a verbal description of a configuration of objects, IF recreated it using a GUI. Aligning these GUI actions to sub-utterance chunks allows a simple maximum entropy model to associate them as chunk meaning better than just providing it with the utterance-final configuration. This shows that semantics useful for incremental (word-by-word) application, as required in natural dialogue, might also be better acquired from incremental settings.

Semantic analyses of the Perfect often defeat their own purpose: by restricting their attention to ‘real’ perfects (like the English one), they implicitly assume the Perfect has predefined meanings and usages. We turn the tables and focus on form, using data extracted from multilingual parallel corpora to automatically generate semantic maps (Haspelmath, 1997) of the sequence ‘Have/Be + past participle’ in five European languages (German, English, Spanish, French, Dutch). This technique, which we dub Translation Mining, has been applied before in the lexical domain (Wälchli and Cysouw, 2012) but we showcase its application at the level of the grammar.

We propose ECO: a new way to generate embeddings for phrases that is Efficient, Compositional, and Order-sensitive. Our method creates decompositional embeddings for words offline and combines them to create new embeddings for phrases in real time. Unlike other approaches, ECO can create embeddings for phrases not seen during training. We evaluate ECO on supervised and unsupervised tasks and demonstrate that creating phrase embeddings that are sensitive to word order can help downstream tasks.

Distributional word representations are widely used in NLP tasks. These representations are based on an assumption that words with a similar context tend to have a similar meaning. To improve the quality of the context-based embeddings, many researches have explored how to make full use of existing lexical resources. In this paper, we argue that while we incorporate the prior knowledge with context-based embeddings, words with different occurrences should be treated differently. Therefore, we propose to rely on the measurement of information content to control the degree of applying prior knowledge into context-based embeddings - different words would have different learning rates when adjusting their embeddings. In the result, we demonstrate that our embeddings get significant improvements on two different tasks: Word Similarity and Analogical Reasoning.

In this paper we present a cross-lingual extension of a neural tensor network model for knowledge base completion. We exploit multilingual synsets from BabelNet to translate English triples to other languages and then augment the reference knowledge base with cross-lingual triples. We project monolingual embeddings of different languages to a shared multilingual space and use them for network initialization (i.e., as initial concept embeddings). We then train the network with triples from the cross-lingually augmented knowledge base. Results on WordNet link prediction show that leveraging cross-lingual information yields significant gains over exploiting only monolingual triples.

One of the key problems in computational metaphor modelling is finding the optimal level of abstraction of semantic representations, such that these are able to capture and generalise metaphorical mechanisms. In this paper we present the first metaphor identification method that uses representations constructed from property norms. Such norms have been previously shown to provide a cognitively plausible representation of concepts in terms of semantic properties. Our results demonstrate that such property-based semantic representations provide a suitable model of cross-domain knowledge projection in metaphors, outperforming standard distributional models on a metaphor identification task.

Non-compositional phrases such as red herring and weakly compositional phrases such as spelling bee are an integral part of natural language (Sag, 2002). They are also the phrases that are difficult, or even impossible, for good compositional distributional models of semantics. Compositionality detection therefore provides a good testbed for compositional methods. We compare an integrated compositional distributional approach, using sparse high dimensional representations, with the ad-hoc compositional approach of applying simple composition operations to state-of-the-art neural embeddings.

Up to date, the majority of computational models still determines the semantic relatedness between words (or larger linguistic units) on the type level. In this paper, we compare and extend multi-sense embeddings, in order to model and utilise word senses on the token level. We focus on the challenging class of complex verbs, and evaluate the model variants on various semantic tasks: semantic classification; predicting compositionality; and detecting non-literal language usage. While there is no overall best model, all models significantly outperform a word2vec single-sense skip baseline, thus demonstrating the need to distinguish between word senses in a distributional semantic model.

We present a new approach to extraction of hypernyms based on projection learning and word embeddings. In contrast to classification-based approaches, projection-based methods require no candidate hyponym-hypernym pairs. While it is natural to use both positive and negative training examples in supervised relation extraction, the impact of positive examples on hypernym prediction was not studied so far. In this paper, we show that explicit negative examples used for regularization of the model significantly improve performance compared to the state-of-the-art approach of Fu et al. (2014) on three datasets from different languages.

Current models for stance classification often treat each target independently, but in many applications, there exist natural dependencies among targets, e.g., stance towards two or more politicians in an election or towards several brands of the same product. In this paper, we focus on the problem of multi-target stance detection. We present a new dataset that we built for this task. Furthermore, We experiment with several neural models on the dataset and show that they are more effective in jointly modeling the overall position towards two related targets compared to independent predictions and other models of joint learning, such as cascading classification. We make the new dataset publicly available, in order to facilitate further research in multi-target stance classification.

The polarity classification task aims at automatically identifying whether a subjective text is positive or negative. When the target domain is different from those where a model was trained, we refer to a cross-domain setting. That setting usually implies the use of a domain adaptation method. In this work, we study the single and cross-domain polarity classification tasks from the string kernels perspective. Contrary to classical domain adaptation methods, which employ texts from both domains to detect pivot features, we do not use the target domain for training. Our approach detects the lexical peculiarities that characterise the text polarity and maps them into a domain independent space by means of kernel discriminant analysis. Experimental results show state-of-the-art performance in single and cross-domain polarity classification.

Inferring the emotional content of words is important for text-based sentiment analysis, dialogue systems and psycholinguistics, but word ratings are expensive to collect at scale and across languages or domains. We develop a method that automatically extends word-level ratings to unrated words using signed clustering of vector space word representations along with affect ratings. We use our method to determine a word’s valence and arousal, which determine its position on the circumplex model of affect, the most popular dimensional model of emotion. Our method achieves superior out-of-sample word rating prediction on both affective dimensions across three different languages when compared to state-of-the-art word similarity based methods. Our method can assist building word ratings for new languages and improve downstream tasks such as sentiment analysis and emotion detection.

Neural network models have been used for target-dependent sentiment analysis. Previous work focus on learning a target specific representation for a given input sentence which is used for classification. However, they do not explicitly model the contribution of each word in a sentence with respect to targeted sentiment polarities. We investigate an attention model to this end. In particular, a vanilla LSTM model is used to induce an attention value of the whole sentence. The model is further extended to differentiate left and right contexts given a certain target following previous work. Results show that by using attention to model the contribution of each word with respect to the target, our model gives significantly improved results over two standard benchmarks. We report the best accuracy for this task.

We describe EmoBank, a corpus of 10k English sentences balancing multiple genres, which we annotated with dimensional emotion metadata in the Valence-Arousal-Dominance (VAD) representation format. EmoBank excels with a bi-perspectival and bi-representational design. On the one hand, we distinguish between writer’s and reader’s emotions, on the other hand, a subset of the corpus complements dimensional VAD annotations with categorical ones based on Basic Emotions. We find evidence for the supremacy of the reader’s perspective in terms of IAA and rating intensity, and achieve close-to-human performance when mapping between dimensional and categorical formats.

We introduce a tree-structured attention neural network for sentences and small phrases and apply it to the problem of sentiment classification. Our model expands the current recursive models by incorporating structural information around a node of a syntactic tree using both bottom-up and top-down information propagation. Also, the model utilizes structural attention to identify the most salient representations during the construction of the syntactic tree.

Users often use social media to share their interest in products. We propose to identify purchase stages from Twitter data following the AIDA model (Awareness, Interest, Desire, Action). In particular, we define the task of classifying the purchase stage of each tweet in a user’s tweet sequence. We introduce RCRNN, a Ranking Convolutional Recurrent Neural Network which computes tweet representations using convolution over word embeddings and models a tweet sequence with gated recurrent units. Also, we consider various methods to cope with the imbalanced label distribution in our data and show that a ranking layer outperforms class weights.

In this paper, we present an improved graph-based translation model which segments an input graph into node-induced subgraphs by taking source context into consideration. Translations are generated by combining subgraph translations left-to-right using beam search. Experiments on Chinese–English and German–English demonstrate that the context-aware segmentation significantly improves the baseline graph-based model.

We investigate the reranking of the output of several distributional approaches on the Bilingual Lexicon Induction task. We show that reranking an n-best list produced by any of those approaches leads to very substantial improvements. We further demonstrate that combining several n-best lists by reranking is an effective way of further boosting performance.

Phrase-based and hierarchical phrase-based (Hiero) translation models differ radically in the way reordering is modeled. Lexicalized reordering models play an important role in phrase-based MT and such models have been added to CKY-based decoders for Hiero. Watanabe et al. (2006) proposed a promising decoding algorithm for Hiero (LR-Hiero) that visits input spans in arbitrary order and produces the translation in left to right (LR) order which leads to far fewer language model calls and leads to a considerable speedup in decoding. We introduce a novel shift-reduce algorithm to LR-Hiero to decode with our lexicalized reordering model (LRM) and show that it improves translation quality for Czech-English, Chinese-English and German-English.

The task of unsupervised lexicon induction is to find translation pairs across monolingual corpora. We develop a novel method that creates seed lexicons by identifying cognates in the vocabularies of related languages on the basis of their frequency and lexical similarity. We apply bidirectional bootstrapping to a method which learns a linear mapping between context-based vector spaces. Experimental results on three language pairs show consistent improvement over prior work.

Many errors in phrase-based SMT can be attributed to problems on three linguistic levels: morphological complexity in the target language, structural differences and lexical choice. We explore combinations of linguistically motivated approaches to address these problems in English-to-German SMT and show that they are complementary to one another, but also that the popular verbal pre-ordering can cause problems on the morphological and lexical level. A discriminative classifier can overcome these problems, in particular when enriching standard lexical features with features geared towards verbal inflection.

We implement a fully probabilistic model to combine the hypotheses of a Spanish anaphora resolution system with those of a Spanish-English machine translation system. The probabilities over antecedents are converted into probabilities for the features of translated pronouns, and are integrated with phrase-based MT using an additional translation model for pronouns. The system improves the translation of several Spanish personal and possessive pronouns into English, by solving translation divergencies such as ‘ella’ vs. ‘she’/‘it’ or ‘su’ vs. ‘his’/‘her’/‘its’/‘their’. On a test set with 2,286 pronouns, a baseline system correctly translates 1,055 of them, while ours improves this by 41. Moreover, with oracle antecedents, possessives are translated with an accuracy of 83%.

In this paper we study the impact of using images to machine-translate user-generated e-commerce product listings. We study how a multi-modal Neural Machine Translation (NMT) model compares to two text-only approaches: a conventional state-of-the-art attentional NMT and a Statistical Machine Translation (SMT) model. User-generated product listings often do not constitute grammatical or well-formed sentences. More often than not, they consist of the juxtaposition of short phrases or keywords. We train our models end-to-end as well as use text-only and multi-modal NMT models for re-ranking n-best lists generated by an SMT model. We qualitatively evaluate our user-generated training data also analyse how adding synthetic data impacts the results. We evaluate our models quantitatively using BLEU and TER and find that (i) additional synthetic data has a general positive impact on text-only and multi-modal NMT models, and that (ii) using a multi-modal NMT model for re-ranking n-best lists improves TER significantly across different n-best list sizes.

Most existing models for multilingual natural language processing (NLP) treat language as a discrete category, and make predictions for either one language or the other. In contrast, we propose using continuous vector representations of language. We show that these can be learned efficiently with a character-based neural language model, and used to improve inference about language varieties not seen during training. In experiments with 1303 Bible translations into 990 different languages, we empirically explore the capacity of multilingual language models, and also show that the language vectors capture genetic relationships between languages.

We address for the first time unsupervised training for a translation task with hundreds of thousands of vocabulary words. We scale up the expectation-maximization (EM) algorithm to learn a large translation table without any parallel text or seed lexicon. First, we solve the memory bottleneck and enforce the sparsity with a simple thresholding scheme for the lexicon. Second, we initialize the lexicon training with word classes, which efficiently boosts the performance. Our methods produced promising results on two large-scale unsupervised translation tasks.

This paper presents a straightforward method to integrate co-reference information into phrase-based machine translation to address the problems of i) elided subjects and ii) morphological underspecification of pronouns when translating from pro-drop languages. We evaluate the method for the language pair Spanish-English and find that translation quality improves with the addition of co-reference information.

We propose a variant of Convolutional Neural Network (CNN) models, the Attention CNN (ACNN); for large-scale categorization of millions of Japanese items into thirty-five product categories. Compared to a state-of-the-art Gradient Boosted Tree (GBT) classifier, the proposed model reduces training time from three weeks to three days while maintaining more than 96% accuracy. Additionally, our proposed model characterizes products by imputing attentive focus on word tokens in a language agnostic way. The attention words have been observed to be semantically highly correlated with the predicted categories and give us a choice of automatic feature extraction for downstream processing.

We present a model to perform authorship attribution of tweets using Convolutional Neural Networks (CNNs) over character n-grams. We also present a strategy that improves model interpretability by estimating the importance of input text fragments in the predicted classification. The experimental evaluation shows that text CNNs perform competitively and are able to outperform previous methods.

Aspect extraction abstracts the common properties of objects from corpora discussing them, such as reviews of products. Recent work on aspect extraction is leveraging the hierarchical relationship between products and their categories. However, such effort focuses on the aspects of child categories but ignores those from parent categories. Hence, we propose an LDA-based generative topic model inducing the two-layer categorical information (CAT-LDA), to balance the aspects of both a parent category and its child categories. Our hypothesis is that child categories inherit aspects from parent categories, controlled by the hierarchy between them. Experimental results on 5 categories of Amazon.com products show that both common aspects of parent category and the individual aspects of sub-categories can be extracted to align well with the common sense. We further evaluate the manually extracted aspects of 16 products, resulting in an average hit rate of 79.10%.

The majority of approaches to author profiling and author identification focus mainly on lexical features, i.e., on the content of a text. We argue that syntactic and discourse features play a significantly more prominent role than they were given in the past. We show that they achieve state-of-the-art performance in author and gender identification on a literary corpus while keeping the feature set small: the used feature set is composed of only 188 features and still outperforms the winner of the PAN 2014 shared task on author verification in the literary genre.

Political text scaling aims to linearly order parties and politicians across political dimensions (e.g., left-to-right ideology) based on textual content (e.g., politician speeches or party manifestos). Existing models scale texts based on relative word usage and cannot be used for cross-lingual analyses. Additionally, there is little quantitative evidence that the output of these models correlates with common political dimensions like left-to-right orientation. Experimental results show that the semantically-informed scaling models better predict the party positions than the existing word-based models in two different political dimensions. Furthermore, the proposed models exhibit no drop in performance in the cross-lingual compared to monolingual setting.

Existing models based on artificial neural networks (ANNs) for sentence classification often do not incorporate the context in which sentences appear, and classify sentences individually. However, traditional sentence classification approaches have been shown to greatly benefit from jointly classifying subsequent sentences, such as with conditional random fields. In this work, we present an ANN architecture that combines the effectiveness of typical ANN models to classify sentences in isolation, with the strength of structured prediction. Our model outperforms the state-of-the-art results on two different datasets for sequential sentence classification in medical abstracts.

Topics generated by topic models are typically presented as a list of topic terms. Automatic topic labelling is the task of generating a succinct label that summarises the theme or subject of a topic, with the intention of reducing the cognitive load of end-users when interpreting these topics. Traditionally, topic label systems focus on a single label modality, e.g. textual labels. In this work we propose a multimodal approach to topic labelling using a simple feedforward neural network. Given a topic and a candidate image or textual label, our method automatically generates a rating for the label, relative to the topic. Experiments show that this multimodal approach outperforms single-modality topic labelling systems.

We present a robust approach for detecting intrinsic sentence importance in news, by training on two corpora of document-summary pairs. When used for single-document summarization, our approach, combined with the “beginning of document” heuristic, outperforms a state-of-the-art summarizer and the beginning-of-article baseline in both automatic and manual evaluations. These results represent an important advance because in the absence of cross-document repetition, single document summarizers for news have not been able to consistently outperform the strong beginning-of-article baseline.

We focus on named entity recognition (NER) for Chinese social media. With massive unlabeled text and quite limited labelled corpus, we propose a semi-supervised learning model based on B-LSTM neural network. To take advantage of traditional methods in NER such as CRF, we combine transition probability with deep learning in our model. To bridge the gap between label accuracy and F-score of NER, we construct a model which can be directly trained on F-score. When considering the instability of F-score driven method and meaningful information provided by label accuracy, we propose an integrated method to train on both F-score and label accuracy. Our integrated model yields 7.44% improvement over previous state-of-the-art result.

We propose a framework for discriminative IR atop linguistic features, trained to improve the recall of answer candidate passage retrieval, the initial step in text-based question answering. We formalize this as an instance of linear feature-based IR, demonstrating a 34%-43% improvement in recall for candidate triage for QA.

An important asset of using Deep Neural Networks (DNNs) for text applications is their ability to automatically engineering features. Unfortunately, DNNs usually require a lot of training data, especially for highly semantic tasks such as community Question Answering (cQA). In this paper, we tackle the problem of data scarcity by learning the target DNN together with two auxiliary tasks in a multitask learning setting. We exploit the strong semantic connection between selection of comments relevant to (i) new questions and (ii) forum questions. This enables a global representation for comments, new and previous questions. The experiments of our model on a SemEval challenge dataset for cQA show a 20% of relative improvement over standard DNNs.

Many important email-related tasks, such as email classification or search, highly rely on building quality document representations (e.g., bag-of-words or key phrases) to assist matching and understanding. Despite prior success on representing textual messages, creating quality user representations from emails was overlooked. In this paper, we propose to represent users using embeddings that are trained to reflect the email communication network. Our experiments on Enron dataset suggest that the resulting embeddings capture the semantic distance between users. To assess the quality of embeddings in a real-world application, we carry out auto-foldering task where the lexical representation of an email is enriched with user embedding features. Our results show that folder prediction accuracy is improved when embedding features are present across multiple settings.

In this paper, we present a method for temporal relation extraction from clinical narratives in French and in English. We experiment on two comparable corpora, the MERLOT corpus and the THYME corpus, and show that a common approach can be used for both languages.

We experiment with neural architectures for temporal relation extraction and establish a new state-of-the-art for several scenarios. We find that neural models with only tokens as input outperform state-of-the-art hand-engineered feature-based models, that convolutional neural networks outperform LSTM models, and that encoding relation arguments with XML tags outperforms a traditional position-based encoding.

We address fine-grained entity classification and propose a novel attention-based recurrent neural network (RNN) encoder-decoder that generates paths in the type hierarchy and can be trained end-to-end. We show that our model performs better on fine-grained entity classification than prior work that relies on flat or local classifiers that do not directly model hierarchical structure.

Many of the world’s languages contain an abundance of inflected forms for each lexeme. A critical task in processing such languages is predicting these inflected forms. We develop a novel statistical model for the problem, drawing on graphical modeling techniques and recent advances in deep learning. We derive a Metropolis-Hastings algorithm to jointly decode the model. Our Bayesian network draws inspiration from principal parts morphological analysis. We demonstrate improvements on 5 languages.