Thursday, June 28, 2007
Faceting and clustering
Faceting is a "simple" grouping of data from the results, like say, grouping all results from a certain database, or for a certain publication date, or for a certain author. However, clustering actually goes deeper into the semantic meaning--or ideas--expressed within titles, abstracts, publication names, etc. so that you have results grouped into subjects, even though the actual subject words didn't appear in the original result's information (example: 3 articles talking about the bombing of Hiroshima, Pearl Harbor and Nazi Germany might--if you're lucky enough--to be clustered under "World War II" even though they didn't specifically mention it in their abstract or metadata.