Machine Translation Essay

1467 Words6 Pages
Abstract-Machine translation industry tries to achieve the translation excellence and research in this area is in a booming stage. All most all areas can utilize the advantages of machine translation because various languages are in the world and majority of global communications are in English which cannot understand by the ordinary people. Translation software saves money and time in order to translate languages. Marketing, customer support, research, software localization, message translation in email and instant messaging, learning foreign languages, website translation, communications, and educational purpose etc. are some of the applications of machine translation system. Whatever it is perfect translation of languages is a challenging…show more content…
Machine translation can be done using various methods. Mainly they are divided into four categories. Following section discusses about these methods:
a. Rule Based Machine Translation (RBMT)
In RBMT method source language structure is translated into target language structures with the help of linguist information. Dictionaries are prepared for rules and use these dictionaries for translation of two different languages. Predictability and easy customization are the two main advantages of this method. Lack of good dictionaries, dictionary building and ambiguity in the rules are the drawbacks of RBMT [1]. UNIversal TRANslator (UNITRAN) is an example of RBMT tool.
b. Statistical Machine Translation (SMT)
SMT uses the statistical models and this concept is coming from the information theory. Statistical analysis is done with the help of bilingual corpora in order to learn the method at the time of translation. Corpus is the basis of SMT model but resource availability for the corpus creation is limited. Earlier prediction of result cannot be possible with SMT and low efficiency with languages which have different word orders [2]. N-gram based SMT is an example of statistical machine
…show more content…
Example Based Machine Translation (EBMT)
As the name implies translation of EBMT system is based on examples which are stored in corpus. It generates the rapid output because examples are used to train the system. It works well with small data sets. One main drawback of this method is inefficiency in deep linguistic analysis. Example based method is mainly used to translate two totally different languages like Japanese and English [3]. Eg.
Open Document