The preliminaryjudgmentof the results returned by the search engine is of substantial significance to opti-mizing the search process.As a dominant search engine,Google often returns very complex results,for which there is no effective way to make accurate judgments on the results of search pages.This paper first constructs a data set suitable for Google search result classification,and then,proposes a dual-channel model(DCFE)based on the pre-training model to determine the Google search results.The accuracy of our model on the self-built dataset reach 85.74%,which has higher accuracy the existing models.