信息类、导航类与事务类查询的网络动态性分析

[Purpose] This paper aims to provide some basis for the performance optimization of search engines through analyzing web dynamic of informational, navigational and transactional queries. [Methods] This paper firstly analyzes the web dynamic of query intent from the following three aspects: query dynamic, web dynamic and the change of information need by using some evaluation metrics. And then, for each category of query intent, this paper also analyzes the change of web document and the change of information need for different type of query dynamic. [Results] ] Results show that, the distribution of query popularity of Informational, transactional and navigational queries is more likely to contain the crests, multiply crests as well as having thecharacter of periodicity and trend of flat respectively. The change of web content of informational queries over time is greater than that that of navigational and transactional queries respectively, and the information intent of informational queries also changes over time is greater than that of another two types of queries respectively. When containing the same number of crests, the change of information need and web content of informational queries is greater than that of informational and navigational queries respectively. When containing the same number of crests, the change of information need of informational queries is greater than that of another two types of queries respectively, and the change of web content of navigational queries is smaller than that of informational and transactional queries respectively. [Conclusion] Search engines need to diverse the search results of informational queries; Search engines need to keep the related and authority pages in the top-ranking of results for navigational queries; Search engines need to keep the ranking of related pages unchanged in the long time for the user behavior-related transactional queries, while search engines need to consider the novelty when ranking pages for the entertainment-related transactional queries. [Limitations] The observing period of our research is only one month; The query popularity distribution was not classified or automatically identified when it contains zero or multiple peaks.