Behavior-based item review system

摘要

Systems and associated processes for generating user attribute interest data and using that data to present and/or develop reviews of items are described herein. These systems and processes can dynamically collect data on a user's interest in specific item attributes from the user's interaction with an electronic catalog and can store this data in association with the user. This data may be used to provide users with a personalized set of reviews which will provide the user with information relevant to interesting item attributes. Further, this data may be used to display a set of dynamically generated statements and/or questions that prompt users to write an item review on specific item attributes. Accordingly, user attribute interest data can potentially both present and elicit more thorough and useful feedback than existing review systems in some embodiments.

主权项

1. A method of obtaining behavior-based review feedback, the method comprising:
programmatically identifying an interest of a first user in an attribute associated with a plurality of items represented in an electronic catalog by analyzing behavior of the first user with respect to the electronic catalog; outputting a review creation user interface that provides functionality for the first user to create a review of a selected one of the items, wherein the review creation user interface prompts the first user to provide feedback about the attribute; storing a response input by the user as a review in association with the selected item in physical computer storage; receiving a request from a second user to view information about the selected item; determining whether the second user is interested in the attribute based at least partly on:
identifying explicit attribute interest cues from behavioral data associated with the second user, the behavioral data representing behavior of the second user with respect to the electronic catalog and the explicit attribute interest cues comprising detected user behavior with respect to the review created by the first user, the detected user behavior indicative of the second user manifesting interest in the attribute,building a map of the attribute interests of the second user based at least partly on analyzing the explicit attribute interest cues and based at least partly on adjusting data representing the attribute interests based on one or more of item category, time, and ranking of the attribute interests, anddetermining whether the map of the attribute interests includes the attribute; in response to determining that the second user is interested in the attribute, selecting a set of reviews including the review of the first user for output to the second user in a review presentation user interface, wherein the set of reviews is sorted based at least partly on the attribute so that the review of the first user that mentions the attribute is presented before other reviews that do not mention the attribute; determining that at least two reviews of the set of reviews indicate a quality of performance of the selected item with respect to the attribute; and outputting an indication to the user that at least two users agree regarding the quality of the performance of the selected item with respect to the attribute; wherein at least said selecting the set of reviews to output for presentation to the second user is implemented by a computer system comprising computer hardware.