This paper describes the PASCAL Network of Excellence first Recognising Textual Entailment (RTE-1) Challenge benchmark1 . The RTE task is defined as recognizing, given two text f...
The Recognizing Textual Entailment System shown here is based on the use of a broad-coverage parser to extract dependency relationships; in addition, WordNet relations are used to ...
We present a novel approach to recognizing Textual nt. Structural features are constructed from abstract tree descriptions, which are automatically extracted from syntactic depend...
In adding syntax to statistical MT, there is a tradeoff between taking advantage of linguistic analysis, versus allowing the model to exploit linguistically unmotivated mappings l...
We study the problem of correcting spelling mistakes in text using memory-based learning techniques and a very large database of token n-gram occurrences in web text as training d...