A content recommendation engine for content creators and consumers
CHALLENGE AND GOAL
In today’s connected world, one of the biggest challenges businesses face is ensuring that their content is delivered to the right audiences at the right time. Although social network operators have robust content targeting solutions, they are not accessible to the general public.
SOLUTION AND OUTCOME
In order to democratize content analysis, researchers at SRI International developed a multimodal embedding based generalized recommendation model, named MatchStax. It connects content to users in two ways.
First, when provided with a piece of content, MatchStax identifies the social media users that would find the content valuable. Next, and conversely, when a user profile is chosen, MatchStax suggests content that is of interest to the user.
Today, there’s no shortage of content on the web. Anyone that has used social networks such as Twitter, Facebook or Pinterest is likely familiar with the fact that those sites have robust targeting functions for advertising and general content. While the systems are effective, they are not intended for public use. Instead, the social networks use the solutions to power their advertiser services.
Given that content is continually added to social networks, companies need to ensure that their content is not being lost in the crowd. To help content producers get more engagement from their audiences, researchers at SRI created MatchStax.
MatchStax is a multimodal embedding based generalized recommendation model that matches content to users and vice-versa. When MatchStax is supplied with a piece of content — such as an image or text — the model then indicates which social media users are likely to find the content valuable.
Conversely, if a specific user is chosen, MatchStax indicates content that is of interest to that user.
MatchStax technology is designed to be used across industries and is compatible with social media networks such as Facebook, Twitter, Pinterest and Instagram. A use case where the technology adds value is retail. For example, a shoe manufacturer could use MatchStax to predict the designs that are most attractive to users.
MatchStax can also be used for a variety of other purposes, such as helping to improve awareness around public health campaigns and for general entertainment purposes.
This research was developed with funding from the Defense Advanced Research Projects Agency (DARPA).
The views, opinions and/or findings expressed are those of the author and should not be interpreted as representing the official views or policies of the Department of Defense or the U.S. Government.
Distribution Statement “A”: Approved for Public Release, Distribution Unlimited