US8661069B1
patent
An automated method may include clustering documents into a plurality of clusters, selecting a representative document, respectively, for two of the plurality of clusters, determining a final redirect target document for each of the representative documents in the two of the plurality of clusters, determining whether the final redirect target documents are duplicate documents of each other, and re-clustering the two of the plurality of clusters together if it is determined that the final redirect target documents are duplicate documents.
22 claims. 59,027 characters of text.