In other words, users with similiar search needs tends to click similar ads. Segmenting users by means of clustering is therefore a good strategy for increasing the revenues of a search engines. Users are profiled by means of the search queries and the clicked URLs. Standard clustering algorithms (such as K-Means, min-wise hashing and CLUTO) are then used to segment the users.
Obtained results are pretty impressive.
No comments:
Post a Comment