Paraphrasing methods recognize, generate, or extract phrases, sentences, or longer natural language expressions that convey almost the same information. Textual entailment methods, on the other hand, recognize, generate, or extract pairs of natural language expressions, such that a human who reads (and trusts) the first element of a pair would most likely infer that the other element is also true. Paraphrasing can be seen as bidirectional textual entailment and methods from the two areas are often similar. Both kinds of methods are useful, at least in principle, in a wide range of natural language processing applications, including question answering, summarization, text generation, and machine translation. We summarize key ideas from the two areas by considering in turn recognition, generation, and extraction methods, also pointing to prominent articles and resources.

Paraphrasing methods recognize, generate, or extract phrases, sentences, or longer natural language expressions that convey almost the same information. Textual entailment methods, on the other hand, recognize, generate, or extract pairs of natural language expressions, such that a human who reads (and trusts) the first element of a pair would most likely infer that the other element is also true. Paraphrasing can be seen as bidirectional textual entailment and methods from the two areas are often similar. Both kinds of methods are useful, at least in principle, in a wide range of natural language processing applications, including question answering, summarization, text generation, and machine translation. We summarize key ideas from the two areas by considering in turn recognition, generation, and extraction methods, also pointing to prominent articles and resources.

We present a novel framework for learning to interpret and generate language using only perceptual context as supervision. We demonstrate its capabilities by developing a system that learns to sportscast simulated robot soccer games in both English and Korean without any language-specific prior knowledge. Training employs only ambiguous supervision consisting of a stream of descriptive textual comments and a sequence of events extracted from the simulation trace. The system simultaneously establishes correspondences between individual comments and the events that they describe while building a translation model that supports both parsing and generation. We also present a novel algorithm for learning which events are worth describing. Human evaluations of the generated commentaries indicate they are of reasonable quality and in some cases even on par with those produced by humans for our limited domain.

We present a novel method for inducing synchronous context free grammars (SCFGs) from a corpus of parallel string pairs. SCFGs can model equivalence between strings in terms of substitutions, insertions and deletions, and the reordering of sub-strings. We develop a non-parametric Bayesian model and apply it to a machine translation task, using priors to replace the various heuristics commonly used in this field. Using a variational Bayes training procedure, we learn the latent structure of translation equivalence through the induction of synchronous grammar categories for phrasal translations, showing improvements in translation performance over previously proposed maximum likelihood models.

We address the problem of learning classifiers when observations have multiple views, some of which may not be observed for all examples. We assume the existence of view generating functions which may complete the missing views in an approximate way. This situation corresponds for example to learning text classifiers from multilingual collections where documents are not available in all languages. In that case, Machine Translation (MT) systems may be used to translate each document in the missing languages. We derive a generalization error bound for classifiers learned on examples with multiple artificially created views. Our result uncovers a trade-off between the size of the training set, the number of views, and the quality of the view generating functions. As a consequence, we identify situations where it is more interesting to use multiple views for learning instead of classical single view learning. An extension of this framework is a natural way to leverage unlabeled multi-view data in semi-supervised learning. Experimental results on a subset of the Reuters RCV1/RCV2 collections support our findings by showing that additional views obtained from MT may significantly improve the classification performance in the cases identified by our trade-off.

Texts and their translations are a rich linguistic resource that can be used to train and test statistics-based Machine Translation systems and many other applications. In this paper, we present a working system that can identify translations and other very similar documents among a large number of candidates, by representing the document contents with a vector of thesaurus terms from a multilingual thesaurus, and by then measuring the semantic similarity between the vectors. Tests on different text types have shown that the system can detect translations with over 96% precision in a large search space of 820 documents or more. The system was tuned to ignore language-specific similarities and to give similar documents in a second language the same similarity score as equivalent documents in the same language. The application can also be used to detect cross-lingual document plagiarism.

This paper describes an interdisciplinary approach which brings together the fields of corpus linguistics and translation studies. It presents ongoing work on the creation of a corpus resource in which translation shifts are explicitly annotated. Translation shifts denote departures from formal correspondence between source and target text, i.e. deviations that have occurred during the translation process. A resource in which such shifts are annotated in a systematic way will make it possible to study those phenomena that need to be addressed if machine translation output is to resemble human translation. The resource described in this paper contains English source texts (parliamentary proceedings) and their German translations. The shift annotation is based on predicate-argument structures and proceeds in two steps: first, predicates and their arguments are annotated monolingually in a straightforward manner. Then, the corresponding English and German predicates and arguments are aligned with each other. Whenever a shift - mainly grammatical or semantic -has occurred, the alignment is tagged accordingly.

Designers of statistical machine translation (SMT) systems have begun to employ tree-structured translation models. Systems involving tree-structured translation models tend to be complex. This article aims to reduce the conceptual complexity of such systems, in order to make them easier to design, implement, debug, use, study, understand, explain, modify, and improve. In service of this goal, the article extends the theory of semiring parsing to arrive at a novel abstract parsing algorithm with five functional parameters: a logic, a grammar, a semiring, a search strategy, and a termination condition. The article then shows that all the common algorithms that revolve around tree-structured translation models, including hierarchical alignment, inference for parameter estimation, translation, and structured evaluation, can be derived by generalizing two of these parameters -- the grammar and the logic. The article culminates with a recipe for using such generalized parsers to train, apply, and evaluate an SMT system that is driven by tree-structured translation models.

In this paper we describe a WSD experiment based on bilingual English-Spanish comparable corpora in which individual noun phrases have been identified and aligned with their respective counterparts in the other language. The evaluation of the experiment has been carried out against SemCor. We show that, with the alignment algorithm employed, potential precision is high (74.3%), however the coverage of the method is low (2.7%), due to alignments being far less frequent than we expected. Contrary to our intuition, precision does not rise consistently with the number of alignments. The coverage is low due to several factors; there are important domain differences, and English and Spanish are too close languages for this approach to be able to discriminate efficiently between senses, rendering it unsuitable for WSD, although the method may prove more productive in machine translation.

Consider the setting where a panel of judges is repeatedly asked to (partially) rank sets of objects according to given criteria, and assume that the judges' expertise depends on the objects' domain.  Learning to aggregate their rankings with the goal of producing a better joint ranking is a fundamental problem in many areas of Information Retrieval and Natural Language Processing, amongst others.  However, supervised ranking data is generally difficult to obtain, especially if coming from multiple domains.  Therefore, we propose a framework for learning to aggregate votes of constituent rankers with domain specific expertise without supervision.  We apply the learning framework to the settings of aggregating full rankings and aggregating top-k lists, demonstrating significant improvements over a domain-agnostic baseline in both cases.