信息类、导航类与事务类查询的网络动态性分析

发布时间：2018-02-28 08:04

本文关键词： 信息类查询事务类查询导航类查询查询动态信息需求动态文档内容动态　出处：《数据分析与知识发现》2017年04期 　论文类型：期刊论文

【摘要】：【目的】分析信息类、导航类与事务类查询随时间的网络动态性特征,以期为搜索引擎性能优化提供相关依据。【方法】利用相关评测指标分别从查询动态文档内容动态和信息需求动态三个角度出发,分析不同意图类别查询随时间变化所呈现的特征;针对不同意图类别查询,分析在不同查询流行度特征中,其文档内容以及信息需求的变化情况。【结果】在查询流行度分布方面,信息类查询通常包含波峰,事务类查询更可能包含多个波峰且具有周期性,导航类查询通常保持平滑趋势;信息类查询随网页内容与信息需求变化幅度均比其他两类查询的要大。【局限】观察时间段只有29天;未对不包含波峰与包含多个波峰的查询流行度分布图中波峰进行归类与自动识别。【结论】对于信息类查询来说,搜索引擎尽可能地对其查询结果进行多样化展示;对于导航类查询来说,搜索引擎需要保证与之相关权威网页在查询结果中的靠前性;对于与用户交互行为相关的事务类查询,应长时间保持相关网页排序不变;对于一些与娱乐相关事务类查询,在网页排序中需考虑网页的新颖性。
[Abstract]:[objective] to analyze the network dynamic characteristics of information class, navigation class and transaction class query over time, in order to provide relevant basis for search engine performance optimization. From three angles of document content dynamics and information demand dynamics, this paper analyzes the characteristics of different intention category queries with time, and analyzes the characteristics of different query popularity in different intention category queries. [results] in terms of the distribution of query popularity, information class queries usually contain peaks, while transactional queries are more likely to contain multiple peaks and have periodicity. Navigation queries usually keep a smooth trend, information queries vary more with the content and information requirements of the web pages than the other two types of queries. [limited] observation time period is only 29 days; Without classifying and automatically identifying the peaks in the popularity distribution map of the query without or containing multiple peaks. [conclusion] for information query, search engine displays the query results as diverse as possible; For navigation class query, search engine needs to ensure that the relevant authoritative web pages are in the forefront of query results, and for transaction class queries related to user interaction behavior, the ranking of relevant pages should be kept unchanged for a long time. For some entertainment-related transaction queries, the novelty of web pages should be considered in the ranking of web pages.
【作者单位】：西南大学计算机与信息科学学院;
【基金】：国家社科基金青年项目“融合用户个性化与实时性意图的查询推荐模型研究”(项目编号:15CTQ019) 西南大学博士启动基金“查询意图自动分类与分析研究”(项目编号:SWU114093)的研究成果之一
【分类号】：TP391.3

【相似文献】