新版有的一些变动.

New Features

We begin with a discussion of relational design, rather than E/R design. There are a number of technical improvements to the discussion of functional and multivalued dependencies, including the synthesis algorithm for 3NF and the use of the "chase" algorithm to make inferences of dependencies.
We have added coverage of high-level design using UML.
Coverage of index selection and use of materialized views has been added.
There is new coverage of the database aspects of PHP.
There are two new chapters on XML-related material, including XML Schema, XPath, XQuery, and XSLT.
Coverage of information integration has been expanded. There is new coverage of algorithms for processing mediator queries. Also added is a treatment of "local-as-view" mediation and the techniques necessary to process queries in this setting.
There is greatly expanded coverage of data mining for large-scale data. Improvements on the a-priori algorithm for association rules are explained. We also added material on similarity search (shingling, minhashing, and locality-sensitive hashing) and clustering of large-scale, high-dimensional data.
The PageRank algorithm is explained, along with generalizations to topic-specific PageRank and TrustRank for detecting link spam.
Entity resolution (merging of similar records) is covered.
Material on peer-to-peer databases has been added, including an algorithm for distributed hashing.


从ullman的个人页面上抓取下来的,方便大家阅读.
http://infolab.stanford.edu/~ullman/dscb.html