基于预训练的谷歌搜索结果判定

Google Search Result Classification Based on Pre-training

对搜索引擎返回的结果进行初步判定有利于优化语义搜索过程,提高搜索的准确性和效率。谷歌搜索引擎在所有的搜索引擎中占据主导地位,然而其返回的结果往往非常复杂,目前并没有有效的方法能够对搜索页面的结果做出准确的判断。针对以上问题,该文从数据特征和模型结构设计出发,制作了一个适用于谷歌搜索结果判定的数据集,接着基于预训练模型设计了一种双通道模型(DCFE)用于实现对谷歌搜索结果的判定。该文提出的模型在自建数据集上的准确率可以达到85.74%,相较于已有的模型拥有更高的精度。

The preliminary judgment of the results returned by the search engine is of substantial significance to optimizing 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.