Association rules for Web data mining in WHOWEDA

Abstract

The authors discuss association rules which can be discovered from Web data. The association rules are discussed within the scope of our WHOWEDA (warehouse of Web data) project. WHOWEDA is supported by a Web data model and a set of algebraic operators. The Web data model allows a uniform and integrated view of Web data gathered using a user's query graph. A user's query graph describes the query by example (what the user perceives as the query) and the Web coupling query gathers instances of such a query graph from the Web and stores them in the form of subgraphs (called Web tuples) in a Web table. We discuss association rules within this domain. An association rule defines an association between the nodes and links attributes of Web tuples within a Web table. There are two different classes of association rules that can be developed from data in a Web table. There are two different classes of association rules that can be developed from data in a Web table. Node-to-node associations are those rules that relate the content (defined by metadata attributes) between two or more nodes within a Web tuple. Link associations are rules that show the connectivity of different URLs. Distinguishing the two types of associations provides a view of the structure of the Web data. The goal of performing Web association mining on Web data is to better organize searching patterns through hyperlinked documents.