Class Summary for October 2 Group 2: Lauren Barton Martin Falck Nelson Kile Carolyn O'Hare Robert Ryan After our usual 40 minutes of fooling with the V-Tel system, we discussed the Ranking and Relevance feedback unit and the assignments for it as well as some administrative details. The next unit dealt with concepts of clustering. Topics discussed were: - Centroids - Low level and hierarchical clustering - Uses with inverted file systems - Visualized with dendograms - Computing document to document similarity matrix and analyzing its complexity - Algorithms using Minimal Spanning Tree and nearest neighbor schemes - Steps in applying clustering - Clustering applications in IR systems - indexes - Measures of similarity - heuristics, formulas used for computation, and some common measures - Space and time required for clustering algorithmss - Clustering algorithms - non-hierarchical vs hierarchical - Issues on updating the clusters - Evaluation results