Rehaman Shaik 01 03 The distinct designs of Rule Based Machine Translation and Statistical Machine Translation make itdifficult for a pure machine translation solution to handle all translation obstacles Once lexical and syntacticanalyzers are dealt with using Rule Based the ExpectationMaximization algorithm is run to resolve certainuncertainties with maximum likelihood We have designed and explained an integrated hybrid machine translationsystem in this paper Fusing the most effective features from different strategies is what is pursued In the first stepRBMT converts Arabic text to English and then EM is run on the output to improve it and create the final version Todo well and become better the EM algorithm must be able to accurately translate a frequency from one language toanother The results prove that the new approach works better than the common Rule Based approach and the EMalgorithm as proven by the BLEU system The survey suggests that scores from the HMT system are regularly greaterthan those from SMT Copyright (c) Journal of Engineering Technology and Sciences. All rights reserved 2026