本文研究了针对大规模查询日志进行时效性实体词挖掘技术,实体词挖掘在搜索引擎中应用广泛,而时效性实体词的识别,对于网络新增热门实体词的检索体验优化具有至关重要的作用。对此,本文通过query聚类、提取实体词上下文组成长文本的方式,提出了一种改进贝叶斯的分类算法进行时效性实体词的识别。实验结果证明本文方法在时效性实体词识别方面,达到了90%以上的准确率。 In this paper, we study the time-efficient entity word mining technology for large-scale query logs. Entity-word mining is widely used in search engines, and the recognition of time-sensitive entity words is crucial for the retrieval experience optimization of new hot entity words in the network effect. In this paper, this paper proposes a modified Bayesian classification algorithm to recognize time-sensitive entity words by means of query clustering and extracting the text of entity word context group. Experimental results show that the proposed method achieves over 90% accuracy in time-sensitive entity word recognition.