Recommend content based on agent-based preference indications

Abstract

In one aspect, a method includes receiving an indication that a user requests to view content at a social networking site, identifying one or more genres associated with the user, wherein each genre serves as a genre of interest for the user an agent for grouping of content, determining one or more entities mapped to the one or more genres, identifying a plurality of posts associated with the one or more entities, and providing the one or more of the plurality of posts for display to the user. Other aspects can be implemented in corresponding systems and apparatuses including computer program products.

Description

Background technique [0001] Information overload is a common problem in content-based systems. Social networking services are populated with posts and photos from friends, family, and acquaintances with varying levels of relevance to the user. Filtering through the available content to find the most relevant content can be a daunting task for users and can lead to information fatigue. To address this issue, recommender systems have been employed to analyze signals in order to determine the most relevant content for users. Such signals provide an indication of relevance based on contextual clues such as current time, user location, strength of connection with other members of the community, or content topics shared in the post. These signals can be useful in a general social networking setting, but in domain-specific communities the signals can be very different. For example, a social networking community built around clothing and fashion styles requires a collection of sign...

AI Technical Summary

Problems solved by technology

Filtering through the available content to find the most relevant content can be a daunting task for users and can lead to information fatigue

Images