III: Large: Causal Databases

点击翻译按钮获取中文摘要

The commercial success of data mining, and the great research interestthat this area attracts, prove that there is a need for analyzing andunderstanding data that goes well beyond classical database queries.Users are often particularly interested in understanding the causalrelationship between data items and the reasons for observations.Current database systems cannot explicitly model the causal structurewithin data (although it is often implicit in the data), and thus offerno specific support for causal queries. In the absence of informationabout causal relationships, users have to rely on techniques for miningfor statistically significant patterns in data. Causal relationshipsare often simply concluded from statistical dependencies. This can leadto inaccurate conclusions; correlation does not necessarily implycausation.This project creates the foundations for a new breed of databasescalled causal databases. Causal databases can model causal information,and allow for queries regarding causality and explanations, which arebeyond the scope of current databases. They can also take advantage ofcausal information that is implicit, but unexploited, in some currentdatabases, such as those for large engineering projects. In theproject, new database models and query languages for representing andtransforming causal information are developed, with particular focus onlarge engineering databases and scientific databases. In addition,efficient and scalable techniques for processing causality andcomputing explanations in large causal databases are developed. Thisinvolves both work on integrating causality processing into traditionaldatabase query processing architectures and the development of specialdatastream techniques for scaling up to the most data-intensiveapplications. Further information on the project can be found at the project webpage: http://www.cs.cornell.edu/databases/causality/