The concept of Extensible Markup Language (XML) keyword searching is an interesting subset of query processing, as XML (along with its variants) has become the main web data representation format. This paper analyzes “how to efficiently evaluate and optimize XML keyword queries” through the semantic caching of materialized views. Traditional keyword searching deals with the metadata of an XML (or web) document. The authors not only extend the search across the entire document, but also apply optimizations to manage the huge amount of Internet data. Materialized views are queries that have been cached, so they are immediately available to speed up query processing. The twist is that queries can be reused for other than a duplication of the exact same query. The paper describes an innovative new approach using a search engine devised by the authors, which “can answer queries using materialized views and maintain [those views] incrementally upon XML data update[s].”

In traditional XML keyword search, a query is converted into an XQuery, but this approach “incurs excessively high time complexity” and requires existing data schemas. The authors describe a general framework where the relevant materialized views can be created and continually updated with respect to dynamic source data. Experimental evaluation results show the significant performance improvements of answering queries using materialized views combined with the efficiency of incrementally maintaining views.

In all, this paper is very detailed, with theorems for these new materialized view approaches and algorithms for implementing the authors’ approach. Combined with the experimental results, this could be an important addition to the XML keyword query research field.