The quality of statistical machine translation performed with phrase based approaches can be increased by permuting the words in the source sentences in an order which resembles that of the target language. We propose a class of recurrent neural models which exploit sourceside dependency syntax features to reorder the words into a target-like order. We evaluate these models on the Germanto-Engl...

In this paper, we present a new texture representation of face image using a robust feature from the Trace transform. The masked Trace transform (MTT) offers “texture” information for face representation which is used to reduce the within-class variance. We first transform the image space to the Trace transform space to produce the MTT. Weighted Trace transform (WTT) identifies the tracing line...

The paper presents a formal lexicalized dependency grammar based on Meaning-Text theory. This grammar associates semantic graphs with sentences. We propose a fragment of a grammar for French, including the description of extractions. The main particularity of our grammar is it that it builds bubble trees as syntactic representations, that is, trees whose nodes can be filled by bubbles, which ca...

This contribution presents a dependency grammar (DG) account of control and raising in English. Due to the minimalism of DG analyses of sentence structure, the difference between control and raising cannot be captured in the syntactic structure alone. The situation forces the DG account to reach to some other aspect of dependency syntax other than the raw hierarchies of structure to account for...

This paper describes ICL’s participation at RTE-7. We chose the Main task. Textual entailment is a problem to predict whether an entailment holds for a given test-hypothesis pair. We built an inference model to solve this problem by means of using dependency syntax analysis (by Stanford Parser), lexical knowledge base (e.g. Wordnet), web information (e.g. Wikipedia) and probability method.

In this paper we describe a memory-based machine learning system that extracts biomedical events from texts relying on information from contextual and syntactic features. The main characteristics of the system are that it uses information from dependency syntax and that it integrates classifiers that learn event triggers and event participants jointly. The results show that this system is more ...

A characterization is given of the class of tree translations definable in monadic second-order logic (MSO), in terms of macro tree transducers. The first main result is that the MSO definable tree translations are exactly those tree translations realized by macro tree transducers (MTTs) with regular look-ahead that are single use restricted. For this the single use restriction known from attri...