资 源 简 介
The technique of discriminative reranking is one way of incorporating larger contextual information into the parsing process. It has also been shown to be effective in improving parsing performance over 1-best parsing. For instance, the best performing constituency parser for English is Charniak-Johnson"s reranking parser. This distribution contains reranker models for the well-known Stanford and Berkeley parsers for English trained using Mark Johnson"s reranker code. All these models are freely available under GPL and can be applied to the k-best output of either of the two parsers. These models were built by Sudheer Kolachina (jointly with Prasanth Kolachina) as part of his MS by Research (in Computer Science and Engineering) thesis on "Non-local features in Syntactic Parsing", done at Language Technologies Research Center, IIIT-Hyderabad. If you are using these models in academic work, it would be great if you can include the following references in your documentation/r
