User preference correlation for web-based selection

摘要

A database of user preference information is extracted and compiled from multiple websites by web-crawling robots without cooperation or specific participation by users. Users who interact with a website are frequently required to register and create a login or userID name that uniquely identifies them. Thereafter, when an individual rates an item, it is often recorded and published under their userID name such that other users can see how a specific individual rated the item. Although there is no requirement that a specific user register on different websites utilizing the identical userID, it is extremely common that this practice occurs and the use of identical userIDs on multiple sites is used herein to expand preference analysis beyond a single site. Once the database exists, users can request or be passively offered suggestions that result from preference associations across multiple websites as performed by a preference analysis and suggestion function.

主权项

1. A system for assisting users with the selection of items on the Internet, comprising:
one or more hardware processors; at least one web robot operable on the one or more hardware processors for crawling multiple websites to determine published userIDs and associated published preferences for items that are available on the web; a database accessible to the one or more hardware processors for storing the userIDs and associated preferences; software operable on the one or more hardware processors for performing an analysis and suggestion function; wherein the system assumes until determined otherwise that a userID used by a first user on a first website represents the same user as the same userID used by a second user on a second website; wherein associated preferences stored for a particular userID include preferences for items available from a plurality of websites; wherein the system provides a third user with a suggested first item based on the expressed preferences of at least the first and second users across multiple websites relative to the first item and items that at least the first and second users have previously reviewed; wherein the suggested first item provided to the third user is based on both positive and negative preferences of the first and second users; wherein a weighting profile is created for weighting rating values that represent positive and negative preferences; and wherein the weighting profile is deployed by the system with respect to combining preferences of the first and second users in suggesting the first item.