Web Usage Mining: Application to an Online Educational Digital Library Service

Аннотация

This dissertation was situated in the crossroads of educational data mining (EDM), educational digital libraries (such as the National Science Digital Library; http://nsdl.org), and examination of teacher behaviors while creating online learning resources in an end-user authoring system, the Instructional Architect (IA; http://ia.usu.edu). The knowledge from data/database (KDD) framework for preparing data and finding patterns in large amounts of data served as the process framework in which a latent class analysis (LCA) was applied to IA user data. Details of preprocessing challenges for web usage data are included. A meaningful IA activity framework provided four general areas of user behavior features that assisted in the interpretation of the LCA results: registration and usage, resource collection, project authoring, and project usage. Four clusters were produced on two samples (users with 5–90 logins and those with 10–90 logins) from 22 months of data collection.