Minimum Error Rate Training is the algorithm for log-linear model parameter training most used in state-of-the-art Statistical Machine Translation systems. In its original formula...
The translation quality and parsing efficiency are often disappointed when Rule based Machine Translation systems deal with long sentences. Due to the complicated syntactic structu...
Abstract. Model differences calculated by differencing algorithms contain the atomic changes made to a model. However, they do not capture the user’s intention of the modifica...
Today, model-driven development is getting more sophisticated and can be used for modeling enterprises and developing application systems. Since current interoperability solutions ...
We propose models for semantic orientations of phrases as well as classification methods based on the models. Although each phrase consists of multiple words, the semantic orienta...