Maximum Likelihood Estimation using Word Based Statistical Machine Translation

Machine Translation is the programmed interpretation of one natural language to another natural language. Human languages can be processed by computers via grammars/rules or via statistics/data. There are many machine translation techniques which are either Rule-based or Knowledge-based. In this paper we will discuss the Statistical Machine Translation (SMT) techniques. The Statistical Machine Translation makes use of statistics or data to find the translation of a given source text. In word primarily based SMT, the simple unit of translation is a word. As a consequence words in an input sentence are translated phrase by phrase individually and these words are than sooner or later organized in a particular way to get the target sentence. The paper discusses numerous fashions for the word based Statistical Machine translation. The paper is written with reference to Kashmiri Language, as not much work has been done in this regard.