Towards Context-sensitive Information Retrieval Based on Quantum Theory: With Applications to Cross-media Search and Structured Document Access

基于量子理论的上下文敏感信息检索：跨媒体搜索和结构化文档访问的应用

• 批准号: EP/F014708/1
• 负责人: Dawei Song
• 金额: $ 41.25万
• 依托单位: The Open University
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2008
• 资助国家: 英国
• 起止时间: 2008 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FF014708%2F1
• 关键词: Towards; Context; sensitive; Information; Retrieval

Consider the following search scenario: I am looking for information about children activities around Cambridge; this information is usually listed at the top of documents. I also want to see images of the listed places, and here is an example image of what sort of activities I am looking for. The weather for this weekend is predicted to be rain, so I prefer indoor activities . This information need contains contextual components: local search, weather. It also contains multimodal components: it specifies where the relevant information can be found in the document structure; it requests text and non-text results; and it requires a mixture of image-, text- and structure-based querying. This type of common and realistic information needs cannot be satisfied with today's search technologies.Context-sensitive search and multimodal search are two major challenges to developing new search technologies that will allow information access systems to be truly usable and valuable and make a fully satisfying impact on today's everyday life. Current IR research cannot deal with these challenges sufficiently because of its traditionally ad-hoc and incremental nature and the lack of a unified theoretical framework and mechanisms to seamlessly integrate context and multimodal search. To address these new challenges, it is essential that a radically new IR theory is developed, leading to a revolutionary shift of the IR paradigm. The growth of context-sensitive information access systems and the proliferation of multimedia and structured information sources, where multimodal access is essential for the efficient and effective e.g. learning and teaching through well-structured multimedia learning resources, searching and browsing of web repositories, cultural heritage collections and multimedia digital libraries, etc., make this revolutionary shift increasingly urgent.This proposal aims to make such a revolutionary shift of the IR paradigm by developing a novel and unified information retrieval (IR) theory based on the Quantum Theory (QT) framework to address the emerging challenges of context-sensitive and multimodal search. The proposed theory will be applied, evaluated, validated and refined in two important scenarios: multimedia and structured document retrieval.It has been shown that there exist peculiar relationships between formal methods in IR and QT, suggesting that a non-classical approach based on quantum theory can potentially resolve the aforesaid challenges. The hypothesis that this project builds on is that the QT theory provides innovation and inspiration into circumventing the emerging context and multimodality issues that current IR and search technologies cannot deal with. It will bring new light into IR research and force us to think about the problem from a different but more revolutionary way. This offers tantalizing possibilities and out of the ordinary implications, some of which, if realized, can lead to genuine breakthroughs and frontier technologies. The project is also innovative in the sense that we will conduct pioneering research in not simply building a generic QT-based IR theory but also applying and evaluating the proposed theory in the practical settings.This is a highly adventurous investigation into a largely unexplored area based on the intriguing connections between QT and IR. The success of this research will make significant and far-reaching impact on both information retrieval and quantum information processing: a completely new paradigm and underlying theory of IR for developing context-sensitive and multimodal search technologies that previously could not be brought about by incrementally extending classical IR models; and an expansion of the territory of quantum information processing.

考虑以下搜索场景：我正在查找有关剑桥周围儿童活动的信息;这些信息通常列在文档的顶部。我也想看看列出的地方的图像，这里有一个什么样的活动，我正在寻找的例子图像。这个周末的天气预计会下雨，所以我更喜欢室内活动。这种信息需要包含上下文组件：本地搜索，天气。它还包含多模态组件：它指定在文档结构中可以找到相关信息的位置;它请求文本和非文本结果;它需要基于图像、文本和结构的混合查询。这种常见的和现实的信息需求不能满足今天的搜索技术。上下文相关的搜索和多模式搜索是开发新的搜索技术的两个主要挑战，这将使信息访问系统真正可用和有价值，并对今天的日常生活产生完全令人满意的影响。目前的IR研究不能充分应对这些挑战，因为其传统的特设和增量的性质，缺乏一个统一的理论框架和机制，无缝集成上下文和多模态搜索。为了应对这些新的挑战，它是一个全新的IR理论的发展，导致革命性的转变IR范式至关重要。对上下文敏感的信息访问系统的增长以及多媒体和结构化信息源的扩散，其中多模式访问对于高效率和高效果至关重要，例如通过结构良好的多媒体学习资源进行学习和教学，搜索和浏览网络知识库、文化遗产收藏和多媒体数字图书馆等，该提案旨在通过开发基于量子理论（QT）框架的新颖且统一的信息检索（IR）理论来应对上下文敏感和多模态搜索的新兴挑战，从而实现IR范式的革命性转变。该理论将在多媒体和结构化文档检索这两个重要场景中得到应用、评估、验证和改进，结果表明，IR和QT中的形式化方法之间存在着特殊的关系，这表明基于量子理论的非经典方法可以潜在地解决上述挑战。这个项目的假设是，QT理论提供了创新和灵感，以规避当前IR和搜索技术无法处理的新兴上下文和多模态问题。它将为红外研究带来新的曙光，并迫使我们从一个不同的，但更革命性的方式来思考这个问题。这提供了诱人的可能性和不同寻常的影响，其中一些如果实现，可以导致真正的突破和前沿技术。该项目的创新之处还在于，我们将进行开创性的研究，不仅是建立一个通用的基于QT的IR理论，而且还将在实际环境中应用和评估所提出的理论。这是一个高度冒险的研究，基于QT和IR之间有趣的联系，对一个基本上未开发的领域进行了研究。这项研究的成功将对未来的研究产生重大而深远的影响。对信息检索和量子信息处理的影响：一个全新的范式和基础理论的IR开发上下文敏感和多模态搜索技术，以前不能带来增量扩展经典IR模型;和领土的扩展量子信息处理。

项目成果

Beyond Redundancies: A Metric-Invariant Method for Unsupervised Feature Selection

超越冗余：用于无监督特征选择的度量不变方法

• DOI: 10.1109/tkde.2009.84
• 发表时间: 2010
• 期刊: IEEE Transactions on Knowledge and Data Engineering
• 影响因子: 8.9
• 作者: Yuexian Hou

Database Systems for Advanced Applications - 14th International Conference, DASFAA 2009, Brisbane, Australia, April 21-23, 2009. Proceedings

高级应用数据库系统 - 第 14 届国际会议，DASFAA 2009，澳大利亚布里斯班，2009 年 4 月 21-23 日。会议记录

• DOI: 10.1007/978-3-642-00887-0_60
• 发表时间: 2009
• 作者: Huang Z

Quantum Interaction

量子相互作用

• DOI: 10.1007/978-3-319-28675-4_14
• 发表时间: 2016
• 作者: Coecke B

Advances in Information Retrieval

信息检索的进展

• DOI: 10.1007/978-3-319-06028-6_2
• 发表时间: 2014
• 作者: Wilkie C