学生论文
|
论文查询结果 |
返回搜索 |
|
|
|
| 论文编号: | 7441 | |
| 作者编号: | 2120132502 | |
| 上传时间: | 2015/6/9 22:07:36 | |
| 中文题目: | 探索式搜索任务属性与信息搜索行为关系研究 | |
| 英文题目: | The Research On the Relationships between Exploratory Search Task Attributes and Information Search Behaviors | |
| 指导老师: | 李月琳 | |
| 中文关键字: | 探索式搜索;任务属性;搜索行为;任务复杂度;任务困难度 | |
| 英文关键字: | exploratory search;task attributes;task complexity;task difficulty | |
| 中文摘要: | 本研究希望在清晰界定、区分任务复杂度和任务困难度概念的基础上,对这两个概念进行操作化测量。然后确定高任务复杂度和高任务困难度取值范围。最后尝试建立行为对高任务复杂度和高任务困难度的预测模型。通过预测模型探索用户行为与探索式搜索任务及探索式需求的关系,分析用户行为对探索式搜索任务属性的预测程度,进而推断用户行为对用户探索式需求的预测作用。 本研究采用实验法,依据主题设计了五个不同搜索任务,选取了48名被试者参与实验,通过视频软件记录被试者的搜索过程。在搜索前问卷中设计了结构复杂性、过程复杂性和复杂性评估三个问题测量主观任务复杂度。在搜索后问卷中设计了查询难度、实施难度和相关性判断难度三个问题测量任务困难度。在实验视频中共抽取了7类行为指标,分别是检索词、全文下载、列表翻页、高级检索使用、精炼(含排序方式、检索途径和分组浏览3项具体行为指标)、打开摘要页(含题名链接和引文链接2项具体行为指标)和相关性判断(含关闭摘要页、全文浏览和删除全文3项具体行为指标)。 数据分析结果表明主观任务复杂度和任务困难度的测量问题都具有很高的会聚效度、区分效度和一致性信度。数据分析结果表明主题熟悉程度和背景知识熟悉程度等个体因素与主观任务复杂度和任务困难度之间不存在显著相关关系;工具方法熟悉程度与任务困难度之间不存在显著相关关系;工具方法熟悉程度与主观任务复杂度之间存在显著的正相关关系,但这与我们假设它们之间存在负相关关系相悖。数据分析发现,不同任务类型下主观任务复杂度评估存在显著差异;不同学历下主观任务复杂度和任务困难度评估不存在显著差异,任务类型维度下任务困难度评估不存在显著差异。 研究选取值在5及以上的主观任务复杂度作为高任务复杂度;选取值在4.67及以上的任务困难度作为高任务困难度。通过分析高任务复杂度和高任务困难度下行为指标之间的关系,发现行为指标在两任务属性下的关系较稳定。数据表明高任务复杂度和高任务困难度与搜索行为之间不存在相关关系。 通过逐步将单个搜索行为作为自变量放入回归方程,未发现搜索行为与高任务复杂度之间存在多元线性回归关系。在对搜索行为和高任务困难度进行多元线性回归分析中,发现全文下载、列表翻页、高级检索使用、打开摘要页和相关性判断五个搜索行为作为自变量的回归方程显著,五类搜索行为共同作用能够很好地解释高任务困难度值。通过控制学历和任务类型两变量后,五个自变量对高任务困难度的解释作用未发生显著提高。因此最终选取全文下载、列表翻页、高级检索、打开摘要页和相关性判断五个搜索行为作为自变量建立高任务困难度预测模型。标准化系数预测模型如下: 高任务困难度=0.439*全文下载-0.411*列表翻页-0.291*高级检索+0.324*打开摘要页-0.270*相关性判断 在模型中,全文下载、列表翻页和打开摘要页对高任务困难度的预测起主要作用,高级检索和相关性判断的预测作用较弱。 | |
| 英文摘要: | Based on a clear definition, this study aims to conceptualize and measure the task complexity and task difficulty. The researchers choose high scores as the measure values of high task complexity and high task difficulty. Constructing forecasting models in which behaviors as the independent variables to predict high task complexity and high task difficulty is the ultimate goal. It will help to explore the relationships between behaviors and exploratory search task, and the prediction effect of behaviors on user demand, especially on the exploratory need. This study uses the experimental research method. Five different search task is designed according to subjects. There are forty-eight participants in this study, and all the search process is record by video software. The study designs three questions to measure subjective task complexity in the pro-search questionnaire. These questions are structure complexity, process complexity and output complexity. Another three questions are used to measure task difficulty in the after search questionnaire, and they are query difficulty, search difficulty and relevance judgement difficulty. Seven kinds of behaviors are extracted from video, including query, download, page turning, advanced search, refining, opening the abstract page and relevance judgment. The data analysis results indicate that the measurement questions of subjective task complexity and task difficulty both display good convergent validity, consistent reliability and discriminant validity. Task types significantly affect the evaluation of subjective task complexity. High task complexity indicates those subjective task complexity whose value is not less than 5 point. High task difficulty indicates those task difficulty whose value is not less than 4.67 point. Behaviors maintain a stable correlation under the high task complexity and the high task difficulty. Through multiple linear regression, a forecasting model for the high task difficulty is built: High task difficulty=0.439*download-0.411*page turning-0.291*advanced search+0.324*opening the abstract page-0.270*relevance judgment In the prediction effect on high task difficulty, download, page turning and opening the abstract page contribute the most part, and advanced search, relevance judgment contribute the less part. | |
| 查看全文: | 预览 下载(下载需要进行登录) |