This paper deals with machine translation on the Web. In the first section we outline the problems of machine translation, paying special attention to the web-oriented systems. In the second section we present a method of using multidomain corpora for the proper style output. The third section turns to text-level analysis and briefly discusses anaphora and coreference resolution. The fourth section turns to other NLP systems that may improve the quality of machine translation on the Web. In conclusion, we set the range of goals for the nearest future.
S. Kulikov, Moscow State Regional University