The collaborative development of cloud computing and edge computing has become an important trend of future development. However, at present, cloud edge collaboration is in the initial stage of development. We should accelerate standardization construction, guide and improve the service level of cloud edge collaboration, and promote the healthy development of cloud edge collaboration. In natural language processing, the processing of coordinate structure is a very important work. However, the traditional parallel structure processing method is inefficient and complex. Based on Natural Language Processing (NLP) technology, this paper adopted Support Vector Machines (SVM), Decision Tree (DT) and Bayesian methods, and analyzed the effectiveness of the three methods. This paper proposed to improve the accuracy of parallel structure conversion by using the word order adjustment technology based on statistical model, which provided a certain reference for parallel structure translators. From the perspective of syntax and semantics, this paper analyzed the translation skills of coordinate structure and word order, and put forward corresponding translation strategies and rules. In the experimental analysis part, the accuracy of SVM, DT and Bayesian methods reached 97.65%, 88.94% and 90.64% respectively among all the tested data; the time spent by SVM, DT and Bayesian methods reached 9.21s, 10.84s and 10.33s respectively. To sum up, SVM outperformed the other two methods in terms of accuracy and time. English translation is difficult, so translators often use a lot of translation techniques. In the demonstration of the example sentence, examples without corresponding skills were also provided to illustrate the advantages of this method through comparison. In short, translation must meet the following three points: clear narrative logic, accurate technical content, and fluent language.
