Web usage mining for the recommendation of materialized webviews

Abstract

In this paper, we propose an approach, which is based on web usage mining techniques, to recommend webviews to be materialized. The webview materialization is a term used to represent the transformation of dynamic web data into equivalent static web data. That is the creation of a static instance of a dynamic web page, at a certain point in time. In this work we will extend our previous approach [6] which concerns the use of sequential patterns to recommend the materialized webviews. Firstly we analyze the DIWS historic to extract the frequent sequential patterns and the frequent association rules. Then we use these repetitive behaviors to calculate a materialization weight for each webview. The webviews with high materialization weights are the most recommended for the materialization. Our experiment results show that our approach reduces the materialization risk more than those using only the recent time period to select the materialized webviews.

Cost Model for the Recommendation of Materialized Webviews

“…It consists of maximizing the QoS under the constraint to respect a degree of QoD which is specified by θ. The techniques of data mining have recently been used to the selection of materialized data either on data warehouse [4,5,6,22] or on the web [10,11,23]. In [23], S. Saidi et al have proposed an approach that firstly analyzes the historic of user navigation and then selects the set of materialized webview which improve the query response time and respect maintenance and storage constraints.…”

“…After searching the association rules, we have developed an algorithm, which is published in [11], to filter the webviews. This filtering is based on the materialization weight ( ).…”