We propose a new phrase-based translation model and decoding algorithm that enables us to evaluate and compare several, previously proposed phrase-based translation models. Within our framework, we carry out a large number of experiments to understand better and explain why phrase-based models outperform word-based models. Our empirical results, which hold for all examined language pairs, suggest that the highest levels of performance can be obtained through relatively simple means: heuristic learning of phrase translations from word-based alignments and lexical weighting of phrase translations. Surprisingly, learning phrases longer than three words and learning phrases from high-accuracy wordlevel alignment models does not have a strong impact on performance. Learning only syntactically motivated phrases degrades the performance of our systems.

We propose a new phrase-based translation model and decoding algorithm that enables us to evaluate and compare several, previously proposed phrase-based translation models. Within our framework, we carry out a large number of experiments to understand better and explain why phrase-based models outperform word-based models. Our empirical results, which hold for all examined language pairs, suggest that the highest levels of performance can be obtained through relatively simple means: heuristic learning of phrase translations from word-based alignments and lexical weighting of phrase translations. Surprisingly, learning phrases longer than three words and learning phrases from high-accuracy wordlevel alignment models does not have a strong impact on performance. Learning only syntactically motivated phrases degrades the performance of our systems.

Abstract:We describe our experience in developing a discourse-annotated corpus for community-wide use. Working in the framework of Rhetorical Structure Theory, we were able to create a large annotated resource with very high consistency, using a well-defined methodology and protocol. This resource is made publicly available through the Linguistic Data Consortium to enable researchers to develop empirically grounded, discourse-specific applications.

We present SEARN, an algorithm for integrating SEARch and lEARNing to solve complex structured prediction problems such as those that occur in natural language, speech, computational biology, and vision. SEARN is a meta-algorithm that transforms these complex problems into simple classification problems to which any binary classifier may be applied. Unlike current algorithms for structured learning that require decomposition of both the loss function and the feature functions over the predicted structure, SEARN is able to learn prediction functions for any loss function and any class of features. Moreover, SEARN comes with a strong, natural theoretical guarantee: good performance on the derived classification problems implies good performance on the structured prediction problem.