U.S.-Germany Cooperative Research on the Integration of Textand Database Systems

该奖项将支持麻省大学W. Bruce Croft博士和德国达姆施塔特国家计算机科学研究中心综合出版与信息系统研究所Erich Neuhold教授在计算机科学领域的合作研究。该项目的目标是改进文本检索模型。对能够提供访问复杂多媒体数据的信息系统的需求日益增加。传统的数据库系统提供了许多有用的数据管理功能，但是处理非结构化数据（如文本）的能力有限。Croft博士目前的研究重点是基于文本的信息检索，它使用来自搜索者的相关反馈来智能地修改后续查询并提高整体检索效率。建议的合作工作将探索将基于文本的信息检索中使用的相关反馈和概率检索模型与面向领域的信息检索中使用的关系数据库模型相结合的方法。该方法是将文档和查询表示为单独的有向图；文档将是静态的，查询图将与文档图匹配，并根据搜索者的反馈动态地重建。该研究项目将受益于两位研究者的互补专业知识。Croft博士将致力于检索模型，并研究该模型对超文本检索和集成查询语言的影响。Neuhold教授将集中讨论对数据模型的影响、与数据库系统的实际集成以及为集成系统设计用户界面。本研究的结果有望从理论角度和通过构建原型获得的实践经验中更好地理解整合问题。

This award will support collaborative research in computer science between Dr. W. Bruce Croft, University of Massachusetts, and Professor Erich Neuhold, Integrated Publishing and Information Systems Institute, German National Research Center for Computer Science, Darmstadt, Germany. The objective of the project is to improve text retrieval models. There is an increasing need for information systems that can provide access to complex, multimedia data. Conventional database systems provide many useful data management functions, but have a limited capacity for handling unstructured data, such as text. Dr. Croft's current research focuses on text- based information retrieval that uses relevance feedback from the searcher to modify intelligently subsequent queries and improve overall retrieval effectiveness. The proposed cooperative work will explore ways to incorporate relevance feedback and probabilistic retrieval models that are used in text-based information retrieval with relational database models used in field-oriented information retrieval. The approach is to represent documents and queries as separate directed graphs; the document will be static and the query graphs will be matched to the document graphs and reconstructed dynamically, based on feedback from the searcher. The research project will benefit from the complementary expertise of the two investigators. Dr. Croft will work on the retrieval model and study the implications of this model for hypertext retrieval and integrated query languages. Professor Neuhold will concentrate on the implications for data models, the actual integration with a database system, and the design of user interfaces for the integrated system. The results of this research are expected to lead to a better understanding of the issues of integration, both from a theoretical perspective and from practical experience gained through building a prototype.