Towards advanced data management systems

本研究旨在开发先进的数据管理技术，该技术使用和/或扩展现有的数据库理论和技术来建模数据和处理许多新兴应用程序所需的查询。此类数据的示例包括逻辑数据、不确定性数据、网络日志、序列数据等。我们的重点是逻辑数据和序列数据。我们考虑了信息集成中的逻辑数据，并使用视图研究了查询重写和合查询的求值，其中查询和视图是可能扩展的逻辑规则，约束是线性算术比较。文献讨论了属性比较，并为某些类别的约束开发了实用的解决方案，以找到最大限度地包含重写，当存在时，除了不等式比较的查询。我们将在上面的上下文中为重写算法开发约束解析和优化模块。这扩展了我们在CoopIS 2005中的结果，在那里我们考虑了具有相等的线性算术表达式的约束。在本文的另一个研究主题中，我们研究了不具有任何结构的序列数据的建模和处理。新兴的应用程序产生并使用了大量这样的数据。我们更关注生物序列：DNA和蛋白质。主要问题是需要使用的存储空间和索引构建过程较长。此外，对精确匹配和近似搜索的有效支持是此类应用程序的核心。因此，理想的解决方案应该是空间高效、可扩展、合理的索引构建时间和高效的各种搜索技术。虽然字符串搜索并不新鲜，但不断增长的生物序列的巨大规模以及需要支持的查询的种类和数量构成了新的研究挑战。越来越需要更复杂的索引结构和技术。基于我们在CIKM 2005上的结果，本工作的预期贡献包括生物序列数据管理的理论和实践进展。

This research aims at development of advanced data management techniques, which uses and/or extends existing database theory and technology for modeling data and processing queries required in numerous emerging applications. Examples of such data includes logical data, uncertainty data, web logs, sequence data, etc. Our focus is on logical data and sequence data. We consider logical data in information integration and study query rewriting and evaluation of conjunctive queries using views, where the queries and views are logical rules possibly extended with constraints which are linear arithmetic comparisons. Literature discusses attribute comparisons and for certain classes of connstraints practical solutions are developed to find maximally contained rewritings, when exists, for queries with all but inequality comparisons. We will develop constraints resolution and optimization modules for the rewriting algoritms in our context above. This extends our results in CoopIS 2005, where we considered constraints which are linear arithmetic expressions with equality.In another research theme in this proposal, we investigate modeling and processing sequence data, which do not have any structure. Emerging applications generate and use enormous amounts of such data. We focus more on biological sequences: DNA and proteins. The main issues are the required uge storage space and lengthy process of index construction. Further, efficient support for exact match and approximate searching is central to such applications. Hence a desired solution should be space efficient, scalable, with resonable index construction time and efficient various search tecniques. While string searching is not new, the enormous size of ever-growing biological sequences, and the variety and volume of queries to be supported pose new research challenges. There is a growing need for more sophisicated indexing structures and techniques. Based on our results in CIKM 2005, the expected contributions of this work include both theoretical and practical advances in the management of biological sequence data.