This paper presents the results of the State University of New York at Buffalo (UB) in the Mono-lingual and Multi-lingual tasks at CLEF 2004. For these tasks we used an approach ba...
The supremacy of n-gram models in statistical language modelling has recently been challenged by parametric models that use distributed representations to counteract the difficult...
Statistical language models estimate the probability of a word occurring in a given context. The most common language models rely on a discrete enumeration of predictive contexts ...
John Blitzer, Kilian Q. Weinberger, Lawrence K. Sa...
We present a novel OCR error correction method for languages without word delimiters that have a large character set, such as Japanese and Chinese. It consists of a statistical OC...