摘要
[目的/意义]用户搜索过程中的检索式行为是其信息需求表达和搜索交互过程中的关键环节,因此检索式构建和重构行为分析也是信息搜索行为研究的重点。[方法/过程]本研究以用户参与百度搜索竞赛题目的搜索过程的检索式为研究对象,从检索式重构转移概率矩阵和重构转移序列聚类两个角度展开分析,探讨搜索任务难度与用户重构转移模式之间的关系。[结果/结论]研究发现随着搜索任务难度增加,用户的检索式序列长度显著增加,用户重构模式的转移也更具多样性和复杂性。本文还基于用户检索范围上下位类的变化对用户检索中的重构序列进行聚类,依据聚类结果,提出了稳定型、波动型和逐步型3类重构序列结构。且随着任务难度增加,波动型和逐步型的重构序列比例显著上升,而稳定型的比例下降,任务难度的增加给用户确定检索需求的范围带来了更大的困扰。研究说明任务难度确实对用户的检索式重构模式产生影响,依据对用户检索式重构行为的观察和分析有助于系统预测用户是否在检索中遇到困难,并为用户提供适合用户当前情况的检索词推荐帮助。
[ Purpose/significance ] Query formulation is the key process in the expression of users' information need and inter- active search. Therefore, query formulation and query reformulation analysis are of great importance to information retrieval behavior research. [ Method/process] Data in this study are collected when subjects are participating in an online search competition, or- ganized by Baidu. corn and Wuhan University. From the perspectives of query reformulation transfer probability matrix and clustered query reformulation sequences, the paper investigates the relationship between task difficulty and user query reformulation transfer mode. [ Result/conclusion I The results show that the length of query reformulation sequence increases greatly and user query refor- mulation transfer becomes more diverse and complex when the tasks become more difficult. Users' query reformulation sequence pat- terns are clustered into stable, wave and gradual types based on users' retrieval range change. The results demonstrate that in more difficult tasks, the ratios of wave type and gradual type increase significantly, but the ratio of stable type goes down. The increase of the task difficulty brings more trouble on users' retrieval range. The study indicates that task difficulty has an influence on query re- formulation strategies, and monitoring users' query reformulation behaviors could help search system to detect whether the user has difficulties, and whether any assistance is needed.
出处
《情报理论与实践》
CSSCI
北大核心
2018年第2期22-27,13,共7页
Information Studies:Theory & Application
基金
国家自然科学基金青年科学基金项目"基于用户检索行为和搜索任务情境的个性化信息检索系统研究"的成果之一
项目编号:71303015
关键词
检索式重构
任务难度
序列聚类
重构转移
query reformulation
task difficulty
sequence clustering method
query reformulation transfer