Abstract The explosion of the Web leads to production large amounts texts and inevitably influences their quality. Errors that tend occur more often can distort results, especially when are used for scientific purposes, in language teaching or learning. Hence, there is a need examine existing corpora based on web clean up data, which may contain such “noisy” fragments. In our study, we deal wit...

The size and quality of the parallel corpus used for training, greatly impacts the quality of translation of an SMT system. But, there are very few sources of parallel corpora for many language pairs. This is a major hurdle in the development of good SMT systems. To alleviate this problem, comparable or non-parallel corpora, which are largely available, can be exploited to extract parallel data...

In the development of corpus linguistics, the creation of corpora has had a critical role in corpus-based studies. The majority of created corpora have been associated with English and native languages, while other languages and types of corpora have received relatively less attention. Because an increasing number of corpora have been constructed, and each corpus is constructed for a definite p...

User simulation is used to generate large corpora for using reinforcement learning to automatically learn the best policy for spoken dialogue systems. Although this approach is becoming increasingly popular, the differences between simulated and real corpora are not well studied. We build two simulation models to interact with an intelligent tutoring system. Both models are trained on two diffe...