Method and system for analyzing user interaction to identify documents associated with dissimilar items that may contain synonyms

摘要

Tools and techniques are described for analyzing interactions to identify dissimilar items that may contain synonyms. Methods described herein may retrieve activity records that represent interactions between a visitor and a server-based system, and may identify within the activity records inputs that the visitor provided during the interaction. The methods may identify items within the activity record that are associated with the inputs, and may access additional activity records that also contain the same inputs. The methods may then identify additional items within the additional activity records that are associated with this same input, and may establish similarity ratings for the two items, with the similarity ratings indicating a likelihood that documents respectively associated with the items contain synonyms.