CAREER: Knowledge Representation and Re-Use for Exploratory and Collaborative Search

现代搜索系统在寻找特定信息时工作得很好，但对跨越多个搜索会话并将调查和学习作为主要目标的探索性搜索几乎没有什么支持。当前的搜索系统几乎没有为用户提供捕获、重新使用和恢复他们的搜索工作的方法，几乎没有对发现关于主题的结构信息的支持，并且几乎没有用于搜索者之间的协调和知识共享的方法。因此，用户开始搜索时通常与其他人已经发现的关于域和有效搜索过程的知识隔离，合作进行搜索的团队必须投入大量的人工精力来使用搜索系统之外的渠道进行沟通和协调他们的工作。这项研究将开发和评估新的技术和界面，以允许用户捕获、保存、共享和重复使用有关搜索任务、搜索过程和领域信息的结构化信息。这项研究的一个中心创新将是将模板结构(包括列表、层次结构、二维网格和概念图)整合到用户界面和底层搜索系统中，以帮助用户保存和组织他们发现的信息。这些结构化的信息将被系统用来帮助支持协作工作的个人和用户组的未来搜索。这项研究将回答关于搜索系统如何结合方法来捕获、共享和重复使用在搜索过程中开发的知识的问题，以帮助改善从事类似任务的个人、协作小组和其他搜索者的未来搜索活动。这项研究将探索新的搜索算法和界面，利用关于搜索的结构化信息来帮助用户恢复任务、协调协作努力、发现主题结构和知识共享。结果将提供关于用户对探索性搜索的需求以及系统如何最好地支持这些需求的见解。具体地说，将解决五个目标。(1)将进行用户研究，以了解信息结构是如何在不同类型的搜索任务中自然创建和使用的。(2)设计和实现一个实验性的搜索系统，以支持结构化搜索信息的创建、捕获和使用，以帮助用户执行探索性搜索任务。(3)将利用试验性搜索系统进行研究，以了解不同类型的模板结构的益处，以便为不同类型的搜索任务(例如，学习、规划、决策、收集)保存信息。(4)将进行更多的研究，以了解如何利用现有的结构化搜索信息来帮助支持个人用户恢复任务、协作用户群体以协调他们的努力，并作为从事类似任务的用户的一种搜索协助形式。(5)这项工作的最终目标是研究基于结构化搜索信息和其他搜索交互数据来识别和排序与当前搜索相关的搜索的方法。本研究将显著改进探索性和协作性搜索的工具和方法，并将提供经验结果来指导未来的搜索用户界面和系统开发。该项目的结果将在帮助用户发现新的信息和主题结构、理解他们找到的信息以及从其他人以前的搜索工作中构建方面产生重大的变革性影响。作为项目一部分开发的成果和软件将通过在顶级会议和期刊上发表的论文传播，并将在项目网站(http://ils.unc.edu/searchstructures/).)上提供

Modern search systems work well when looking for specific information, but do little to support exploratory searches that extend across multiple search sessions and have investigation and learning as primary goals. Current search systems provide few ways for users to capture, re-use, and resume their search efforts, little support for discovering structural information about a topic, and few methods for coordination and knowledge sharing among searchers. As a result, users often begin searches isolated from knowledge about the domain and effective search processes that others have already discovered, and groups working collaboratively on a search must invest considerable manual effort to communicate and coordinate their efforts using channels outside the search system. This research will develop and evaluate novel techniques and interfaces to allow users to capture, save, share, and re-use structured information about search tasks, search processes, and domain information. A central innovation in this research will be the integration of template structures (including lists, hierarchies, two-dimensional grids, and concept maps) into the user interface and underlying search system in order help users save and organize the information they discover. This structured information will then be used by the system to help support the future searches of individuals and groups of users working collaboratively.This research will answer questions about how search systems can incorporate methods to capture, share, and re-use knowledge developed during searches to help improve the future search activities of individuals, collaborating groups, and other searchers working on similar tasks. The research will explore new search algorithms and interfaces that leverage structured information about the search to help users with task resumption, coordination of collaborative efforts, discovery of topic structures, and knowledge sharing. The results will provide insights about users' needs for exploratory searches and how systems can best support them. Specifically, five objectives will be addressed. (1) User studies will be conducted to understand how information structures are naturally created and used in different types of search tasks. (2) An experimental search system will be designed and implemented to support the creation, capture, and use of structured search information to help users with exploratory search tasks. (3) Using the experimental search system, studies will be conducted to understand the benefits of different types of template structures for the purpose of saving information for different types of search tasks (e.g., learning, planning, decision-making, collecting). (4) Additional studies will be conducted to understand how existing structured search information can be used to help support individual users in task resumption, collaborating groups of users in coordinating their efforts, and as a form of search assistance for users working on similar tasks. (5) The final objective of this work will investigate methods for identifying and ranking searches that are related to the current search based on the structured search information and other search interaction data. This research will significantly improve the tools and methods for exploratory and collaborative search, and will provide empirical results to guide future search user interfaces and system development. The outcomes of this project will have substantial transformative impact in helping users to discover new information and topic structures, make sense of the information they find, and build from the prior search efforts of others. Results and software developed as part of the project will be disseminated through papers published in top-tier conferences and journals, and will be made available on the project website (http://ils.unc.edu/searchstructures/).