Human-machine Interactive Machine Translation Technology Research Based On Internet

With the increased international exchanges, on the basis of the computer, Internet and other related technical change rapidly, machine translation technology and application have made vigorous development. Machine translation is the use of computers to achieve the translation between different languages, divided into rule-based machine translation method and corpus-based machine translation method. Otherwise, based on the hybrid machine translation method is bank wins, that is:the combination of the two methods. Such as, the rule-based technology is introduced in the corpus approach, which is based on example method, statistical method, by corpus preprocessing to make it into a natural language knowledge base method. In recent years, corpus-based machine translation technology rapid ly develops, and obtains the widespread application, according to different areas and the demand of the market, and appears the diversiform machine translation product having certain practical value, making machine translation technology into the development journey of science and technology.In this paper based on the Internet human-machine mutual machine translation method is a new translation method, the method is different from the previous methods of machine translation, it is not entirely dependent on the automatic machine learning, but through the web crawler technology from Internet to obtain bilingual parallel corpus, and then in the artificial intervention to complete the whole translation process, compared to the independent automatic machine translation, translation quality is better, compared to separate artificial translation, translation is more efficient.On the research process of the translation technical, basically did the following work:(1) corpus-based machine translation method can be divided into two types: statistical-based machine translation method and example-based machine translation method, which take the corpus as a source of translation knowledge. In this paper, based on statistical machine translation method. Bilingual parallel corpora is through the network crawler technology obtained from Internet.(2) after obtaining corpus, considering the complexity of language, the fully automatic machine translation method can not achieve satisfactory results, therefore, the paper puts forward the computer aided translation method to the manual intervention text, which is a text weights method that based on artificial scoring to make the translation optimal results.(3) in statistical machine translation process, phrase tables in the parameters training can effectively filter the tables, a more reasonable.(4) nowadays, the main use of language model is N-gram model, this paper uses the tri-gram grammar model, this model is simple and direct, but also because of lack of data taken smoothing technology to smooth it.(5) finally, in this paper, adding artificial score and C-value method, the translation model is improved. Experimental results show that, the improved machine translation method is superior to the traditional methods of machine translation, phrase translation probability tables have certain reduction, BLEU has also improved.