联系我们

论文查询结果	返回搜索

论文编号：	13179
作者编号：	1120180973
上传时间：	2022/6/7 10:29:10
中文题目：	面向科研任务的知识服务平台个性化推荐研究
英文题目：	Personalized Recommendation for Research Tasks in Knowledge Service Platform
指导老师：	李月琳
中文关键字：	用户研究；个性化推荐；任务特征；知识服务平台
英文关键字：	User research, personalized recommendation, task characteristics, knowledge service platform
中文摘要：	个性化推荐作为互联网平台（或系统、网站）主动为用户提供的一项信息服务，长期受到学术界和实践领域的关注，在个性化推荐算法、个性化推荐系统等相关研究方面取得了丰富的成果。相较于此，用户期待的个性化推荐如何为其提供信息服务、用户如何与个性化推荐展开交互等用户视角下的个性化推荐研究并未得到应有的关注。本研究以任务为切入点，开展用户视角下的个性化推荐研究，探讨面向科研任务的知识服务平台个性化推荐相关问题。 本研究的目的和意义在于，从现实出发，厘清知识服务平台个性化推荐服务的特征，洞察信息用户的个性化推荐需求，进而分析科研任务的特征与个性化推荐之间的关系，构建用户视角下的个性化推荐模型。本研究一方面有助于平台更好地了解用户的个性化推荐需求和个性化推荐交互行为，以便更好地为用户提供个性化推荐服务。另一方面也丰富了个性化推荐理论研究、用户研究和任务研究，也为个性化推荐算法设计、平台交互设计提供用户视角下的指导。本研究搭建了用户研究与个性化推荐算法研究之间的桥梁，为进一步开展用户导向的个性化推荐研究奠定理论和实证基础。 本研究采用多种研究方法，通过定性、定量相结合的方式开展研究。首先，本研究采用内容分析法厘清了知识服务平台在个性化推荐的展示形式、推荐内容、推荐解释、推荐类型、推荐时间节点5个方面的服务特征，并发现88.89%的平台提供的推荐类型为静态推荐，即推荐没有考虑用户与平台的交互行为，这一发现为用户视角的个性化推荐研究提供了现实依据，为识别学术界和实践领域个性化推荐的研究鸿沟提供了参考。其次，本研究采用半结构化深度访谈法识别了科研信息用户的个性化推荐需求，包括5个类别：内容需求、交互功能需求、界面布局需求、效能需求和情感需求，不同类别的需求具有不同的特征，并进一步构建了信息用户的个性化推荐需求层次模型，识别了影响信息用户个性化推荐需求的因素。上述发现为知识服务平台更具针对性地设计个性化推荐提供了指导，也对未来开展用户导向的个性化推荐研究具有一定的启示意义。最后，本研究采用实验法开展了科研任务的特征与个性化推荐关系研究，根据用户的页面交互路径识别了4种类型用户，分析了不同类型用户在与平台交互中潜在浏览的个性化推荐服务形式，发现用户的交互行为具有习惯性特征，影响着他们对个性化推荐的使用；厘清了用户与个性化推荐交互的3个阶段，9种个性化推荐交互行为，识别了24种个性化推荐交互路径，并阐述了用户的个性化推荐交互路径链，为平台的个性化推荐交互设计提供了强有力的指导；本研究进一步分析了任务特征与用户个性化推荐交互行为、个性化推荐交互路径之间的关系，研究表明任务影响着用户的个性化推荐交互行为和路径。这一发现为构建用户视角下的个性化推荐模型提供了实证依据，即任务作为情境因素影响着用户对个性化推荐的关注和使用。本研究分析了任务特征与用户个性化推荐需求之间的关系，构建了个性化推荐对用户任务完成的支持模型，研究进一步加深了学术界对用户个性化推荐需求的理解，尤其是相关性、有用性在个性化推荐中的意义，从理论上解释了个性化推荐如何帮助用户任务的完成。 总之，本研究构建了用户视角下的个性化推荐模型，模型以用户的信息需求和信息搜索为出发点，包含了影响用户个性化推荐的两大因素“情境、交互”，强调用户视角下的个性化推荐需要关注用户的信息搜索情境和用户与平台的交互行为特征，模型也展示了它们与用户个性化推荐需求之间的影响。在此基础上，本研究进一步构建了用户视角下的个性化推荐算法、系统设计指导模型，该模型突出了用户使用个性化推荐之前的交互行为和情境的重要性，进一步结合用户的个性化推荐需求，有助于指导平台进行个性化推荐算法及个性化推荐系统的设计。
英文摘要：	Personalized recommendation (PR), as an information service proactively provided by Internet platform (or system, website) to users, has attracted the attention of academia and industry for a long time, and has got rich achievements in PR algorithm, PR system and other related research. In contrast, the research from the user perspective, such as how users expect PR to provide information services, how users interact with PR, has not received much attention. This study carries out the research on PR from the user perspective, and focuses on the problems related to PR of research task in knowledge service platforms (KSP). The purpose and significance of this study is to clarify the characteristics of PR service of KSP, identify the user’s PR needs, then analyze the relationship between task and personalized recommendation of KSP, and develop PR model from the user perspective. this study can help KSP understand users' PR needs and PR interaction behavior better, provide better PR services for users. It also enriches the theoretical research, user research, task research, and provides guidance from the user perspective for PR algorithm design and platform interaction design. This study adopts qualitative and quantitative methods to analysis data. Firstly, this study uses the content analysis method to clarify the PR service characteristics of the KSP in five aspects: the display form, recommended content, recommendation explanation, recommendation time node, and recommendation type. It is found that 88.89% of the recommendation types provided by the platforms are static, the recommendation does not consider the interaction between user and the platform, this discovery provides a practical basis for PR research based on user perspective, and provides a reference for identifying the research gap of PR in academia and industry. Secondly, this study uses the semi-structured in-depth interview method to identify the user’s PR needs, including 5 categories: content needs, interactive functional needs, interface layout needs, efficiency needs and emotional needs, develop PR needs hierarchical model, identify the factors affecting the user’s PR needs. These findings provide guidance for the PR design in the KSP, and also have certain enlightenment significance for the future research on PR with equal emphasis on technology and humanities. Finally, this study uses the experimental method to study the relationship between tasks and PR in KSP. 4 types of user are identified according to the user's page-interaction path. It is found that the user's interaction behavior has habitual characteristics, which affects the use of PR. This study clarifies 3 stages of interaction between users and PR, 9 kinds of PR interaction behaviors, identifies 24 kinds of PR interaction paths, and expounds the user's PR interaction path chain, it provides a strong guidance for the PR interaction design of the platform. This study further analyzes the relationship between task characteristics and users' PR behavior and PR interaction path. The result shows that user tasks, as context factors, affect users' attention and PR use, this discovery provides an empirical basis for proposing PR model from the user perspective. This study also analyzes the relationship between task characteristics and users' PR needs, develops the PR support model for users' task completion, especially the significance of relevance and usefulness in PR, and theoretically explains how PR helps users to complete their tasks. Based on all the research findings, this study develops PR model from the user perspective. The model takes the user's information need and information search as the starting point, and includes two factors "context and interaction" that affect the user's PR need and behavior. The model also shows the relationship between them with users' PR needs, and further develops the PR algorithm and system design guidance model from the user perspective. This model highlights the importance of users' interactive behavior and context before using PR.
查看全文：	预览  下载(下载需要进行登录)