NYI: Parallel Algorithms for Information Retrieval

在今后几年内，全世界的信息系统将处理对数以千计的个别地点每天产生的数千兆字节的数据的收集和解释。在这样一个异构的分布式数据库系统中定位数据，相当于在世界范围内大海捞针。为了完成这项工作，正在开发一种适当的数据模型，该模型基于对所收集数据的类型和规模的理解，以及对数据执行查询的类型和相对频率。需要对多面数据进行并行搜索、索引和压缩的算法。还需要在构成整个搜索空间的许多节点之间进行通信的协议。这些协议必须允许在各种平台之间高效、可靠和安全地分发查询和结果。在每个数据存储站点开发多计算机搜索引擎。一组合作者使用原型进行评估，他们的反馈将被纳入最终实现。本研究结果可为世界范围内刚性和松散结构数据库的发展提供指导。

Within the next several years, worldwide information systems will deal with the collection interpretation of several gigabytes of data generated daily at thousands of individual sites. Locating data in such a heterogeneous distributed-database system, is tantamount to funding the proverbial needle in a worldwide haystack. To accomplish this endeavor, an appropriate data model based on an understanding of the type and magnitude of the data collected and on the type and relative frequency of the queries to be performed on the data are being developed. Algorithms for conducting parallel searches, indexing, and compression of the multi-faceted data are required. Also necessary are the protocols for communicating among the many nodes that constitute the entire search space. These protocols must allow for efficient, reliable, and secure distribution of queries and results among a variety of platforms. Multicomputer search engines at each data storage site are developed. A set of collaborators employs the prototype for evaluation and their feedback will be incorporated into the final implementation. The results obtained in this research may serve as a guideline for the development of worldwide both rigidly and loosely structured databases.