【作者】 曹梅；

【导师】 朱学芳；

【作者基本信息】 南京大学 ， 情报学， 2010， 博士

【摘要】 用户信息行为研究历来是情报学领域中的一个重要研究分支,对信息的属性、传递方式、价值以及信息科学的功能、结构体系、工具系统的研究,均离不开用户信息行为特性研究这个中心。图像检索作为视觉信息资源服务的主流方式,其相关的技术发展、资源组织、界面设计等必然应该面向图像信息用户,发展有针对性的为用户设计和服务的方法技术。深入开展针对图像检索的用户信息需求、行为策略模式、心理机制研究,有助于有效克服图像信息服务系统重经济、技术效率而轻用户效率的做法,提高图像信息检索系统的服务质量和利用效果；系统研究图像检索的用户信息行为的特征和规律,可以为用户信息行为理论注入新的血液,为多媒体信息资源管理提供来自用户的依据。本文基于“信息需求—信息查寻”相互交融的基本理论观点,选择图像信息检索行为问题,从网络图像检索需求和检索行为过程两大方面展开,初步实证分析和研究了图像检索中用户图像信息检索需求的特点与规律、图像检索的行为习惯与倾向、图像检索过程中的行为策略与模式等一系列相关问题。本文的理论基础是用户信息行为研究领域所积淀的相关理论和方法,包括：“以用户为中心”的研究范式、威尔逊的用户信息行为模型、用户信息需求的表达理论、信息查寻过程标准和动态模型、一般信息检索策略等。具体而言,本文开展了四个针对性的研究工作：第一,图像信息检索行为研究的基本框架。文献调研了国内外在图像检索的用户信息行为研究方面的主要成果,较全面地勾勒出图像信息检索行为研究的“倒会字塔”框架,揭示了该领域的研究进展、研究特点、研究内容、主题分布和研究思路等,研究结果可以为国内相关研究提供直接的指导。第二,网络图像检索需求与行为习惯的调研。从日常活动的角度,采用结构化问卷调研了代表性网络用户群体(在校大学生)在网络图像检索方面的日常信息需求和行为习惯和检索经验,获得了关于大学生在网络图像信息需求的内容、网络信息需求的层次、行为习惯和行为能力方面的一些结论。第三,网络图像检索的用户行为策略。设计用户实验,在真实的互联网络环境下,借助行为观察技术采集用户的图像检索过程,采用内容分析法综合定性和定量数据来分析用户的图像检索行为,从不同角度刻画用户日常图像检索的行为策略。分析的角度有：网络图像检索入口的选择与变换、图像检索中的翻页行为,相关性结果判断,常见的行为模式等；研究观察并剖析了图像检索过程中网络图像检索的典型用户心理、图像检索的各种行为策略以及若干高频行为模式。第四,图像检索需求的表达与调整。研究围绕三个问题：用户图像检索提问式的构造,图像检索提问式的语言分布特征,检索提问式在过程中的动态调整,得到了关于图像检索提问式构造依据、目的和途径,提问式的长度、检索语法运用、提问式调整序列模式等一些行为规律；研究观察到了图像检索需求的一些表达习惯,考察了网络图像检索过程的演进性,归纳了图像检索提问式调整的若干行为策略与模式。本文的技术路线沿着文献调研、用户问卷调查和用户实验而逐步开展理论学习和实证研究,是在总结前人成果和失败的经验基础上,为网络图像检索的用户信息行为研究而确立的,在理论和方法两方面既有继承性又有创新性。文章最后从人机交互设计、图像资源组织和图像语义检索三个方面对当前网络图像检索服务提出了一些应用建议。更多还原

【Abstract】 User behavior has been one of primary research branches in the field of information science, which is concerned with varied studies such as information property, information transmission mode, values as well as function, structure and tool systems of information science. Image retrieval is the mainstream way for the future visual information services. Thus the related technological development, resources organization and interface design should be faced with visual information users and develop targeting methods and technologies for design and services for users. To promote the study on information needs, searching and browsing strategies and models, psychological mechanism with regard to visual retrieval will contribute to effectively get over the practices which focus on much more economic&technological efficiency than users efficiency in visual information systems, which is helpful to improve the service quality and utilization effects of image retrieval systems. The systematic study on user behavior characters and laws will input fresh blood into user behavior theory and provide more bases from the user’s view for multimedia information management.Based on the fundamental views of mingled information needs-information searching theory, the dissertation is closely centered on user information requirement and information retrieval process in image retrieval, preliminarily analyses and studies series issues as the characters and laws of image retrieval requirement, behavior and tendency of image retrieval, strategy and mode in image retrieval process.The theoretical foundation of this dissertation is mainly on the accumulated theories and methodologies in user behavior research field including user-centered research paradigm, Wilson’s information-seeking behavior model, expression of user information needs, theoretical models of information seeking process, theoretical characterizations of information seeking strategies, and so on.Specifically, there are four items of work in that regards:The first part is about the framework of user behavior study for image retrieval. The article investigates key results and ideas in the study at home and abroad, outlines the "inverse pyramid" framework for image retrieval behavior studies, and reveals the study development, characteristics, contents, topic distributions and research ideas in this field, all these results can be direct instructions for related domestic studies.The second part is the investigation on image retrieval request and behavior customs. In perspective of daily activities, by way of structural questionnaire, the article surveys the requirements, customs and retrieval experiences in image retrieval activities by representative Internet user groups (in-campus college students), acquires some results in retrieval requirement contents, requirement level, behavior customs and behavior ability by the college students.The third part is about user behavior strategy. User experiment is designed in which user’s image retrieval process was recorded by way of behavior investigation technology, comprehensively quantified and qualified data to analyze the user retrieval behavior, and described the behavior strategy of users in daily image retrieval activities in various perspectives. The various perspectives are covering selection and change of retrieval entrance; page flip activity in retrieval, relevance judgment, and common behavior modes and so on. The article studied users’ outstanding psychological factors, behavior characteristics and tactics such as high-frequency behavior modes, which users employ when seeking images on the Web.The forth part is around the description and adjustment of image request which is centered on three issues:structure of image query for image retrieval, linguistic distribution characters of image query, dynamic reformulation of image query in process, and obtains series of behavior regular patterns in issues as image query structure basis, targets and channels; length of the query, utilization of retrieval grammars, query reformulation sequence mode. The article observes the expression way of image information needs, discusses the evolution of image searching state, and sums up some query reformulation strategies and modes.The technological route for this dissertation is along the study on documents, user questionnaire investigation and users experiment to progressively carry out theoretical and practical studies. The study is based on achievements and failures from predecessors and is established for research on image retrieval user behavior, which is inheriting as well as creative in both theories and practices.At the end of the article, some application suggestions in the current image retrieval services are put forward from three aspects, i.e., user interfaces, image information resources organization and image semantic retrieval.更多还原