A method and apparatus for identifying items of information from an information system which is especially useful for identifying information from the World Wide Web. The information items have not been seen by the user before and are relevant and interesting to the user. A first record of items requested by each member of a group of users is obtained. A second record is obtained of items requested more than once by the same user. Pairs of users in the group are analyzed, and said to be "friends" if they have requested some of the same information from the second record. Information that has been requested by a user's friend but which has not been seen by the user before is identified for that user.