Databases and Information Retrieval (IR) have followed distinct paths,
mainly due the fact that databases used to store only rigidly structured data 
and handle rigidly structured queries, which serve the purposes of 
well-designed applications. In contrast, IR is employed for information 
discovery and primarily studies how (unstructured) documents are ranked 
according to their relevance to an unstructured query (typically a set of keywords). 

However, due to the increasing availability of database data, 
the increasing popularity of XML and the increasing amount of text 
stored in databases, it has become imperative to allow unstructured 
queries on structured and semistructured databases, in addition to 
documents. The most popular type of unstructured query is keyword queries, 
which eliminate the requirement of knowing a query language and the structure 
of the database, and whose success is proven by Web search engines. 
The main focus of this talk is proximity keyword queries, which discover 
the associations between keywords in a database 
(http://www.db.ucsd.edu/xkeyword/). In addition to that, we briefly 
present how link-based ranking techniques (similar to PageRank) can 
be applied to improve the quality of single-object results of keyword 
queries (http://www.db.ucsd.edu/objectrank/).