Selecting content with an external link for presentation based on user interaction with external content
An online system receives posts that include links to various external pages and presents those posts to users of the online system. When the online system determines an opportunity to present a post to a particular viewing user of the online system, the online system determines a quality metric and an associated value score for the post. The quality metric is determined as a likelihood that the viewing user will view the external page for less than a threshold time period, and is used to adjust the associated value score. The online system compares the value score of the post to the value scores of other posts and selects one or more of the compared posts for presentation to the viewing user of the online system.
1. A method comprising:
accessing a post from a posting user of an online system, the post comprising a link to an external page that is outside the online system;
determining an opportunity to present content to a viewing user of the online system;
determining one or more attributes of the external page;
determining a likelihood of the viewing user viewing the external page for less than a threshold time period based on the attributes of the external page;
determining a quality metric of the post for the viewing user based on the attributes of the determined likelihood of the viewing user viewing the external page for less than the threshold time period, where the quality metric decreases with a higher determined likelihood of the viewing user viewing the external page for less than the threshold time period;
computing a value score presenting the post to the viewing user, the value score adjusted upward by the quality metric;
ranking the post by its value score along with a plurality of other posts;
selecting, based on the ranking, one or more of: the post and the plurality of other posts; and
sending the selected posts for presentation to the viewing user.
2. The method of claim 1 , wherein the determined quality metric of the post is a binary classifier based on the likelihood of the viewing user viewing the external page for less than the threshold time period.
3. The method of claim 1 , wherein the attributes of the external page comprise one or more of: historical dwell time at the external page, a user bounce-back rate at the external page, and user average dwell time at the external page.
4. The method of claim 1 , wherein the threshold period of time is computed as a time between the user navigating to the external page and the user closing the external page, and is independent of the user navigating away from the external page.
5. The method of claim 1 , wherein computing a value score comprises:
computing a plurality of value score components, one of the value score comprising an external page value score that is computed based on a likelihood of the viewing user viewing the external page and a value of the viewing user viewing the external page, where the value of the viewing user viewing the external page is adjusted upward by the quality metric; and
aggregating the plurality of quality score components to determine the value score for the post.
6. The method of claim 1 , wherein the post accessed by the online system may comprise: an organic post from an individual user, or sponsored content from a business or organization.
7. The method of claim 1 , wherein the link associated with the accessed post comprises one or more of: a URL to a web page, a deep link to a location within a mobile application, and a link to an app store to install an application.
8. The method of claim 1 , wherein the external page associated with the accessed post comprises one or more of: a web page, a mobile application, and a state or page within a mobile application defined by a deep link.
9. The method of claim 1 , wherein the quality metric is further determined based on attributes associated with the viewing user.
10. The method of claim 1 , further comprising:
training a machine learning model using historical data about whether users viewed external pages for a threshold period of time after being presented with a post with a link to the external page; and
using the trained machine learning model to compute a quality metric for a post containing a link to an external page.
11. A computer program product, comprising a computer-readable storage medium containing computer program code for:
accessing a post from a posting user of an online system, the post comprising a link to an external page that is outside the online system;
determining an opportunity to present content to a viewing user of the online system;
determining one or more attributes of the external page;
determining a likelihood of the viewing user viewing the external page for less than a threshold time period based on the attributes of the external page.
determining a quality metric of the post for the viewing user based on the attributes of the determined likelihood of the viewing user viewing the external page for less than the threshold time period, where the quality metric decreases with a higher determined likelihood of the viewing user viewing the external page for less than the threshold time period;
computing a value score presenting the post to the viewing user, the value score adjusted upward by the quality metric;
ranking the post by its value score along with a plurality of other posts;
selecting, based on the ranking, one or more of: the post and the plurality of other posts; and
sending the selected posts for presentation to the viewing user.
12. The computer program product of claim 11 , wherein the determined quality metric of the post is a binary classifier based on the likelihood of the viewing user viewing the external page for less than the threshold time period.
13. The computer program product of claim 11 , wherein the attributes of the external page comprise one or more of: historical dwell time at the external page, a user bounce-back rate at the external page, and user average dwell time at the external page.
14. The computer program product of claim 11 , wherein the threshold period of time is computed as a time between the user navigating to the external page and the user closing the external page, and is independent of the user navigating away from the external page.
15. The computer program product of claim 11 , wherein computing a value score comprises:
computing a plurality of value score components, one of the value score comprising an external page value score that is computed based on a likelihood of the viewing user viewing the external page and a value of the viewing user viewing the external page, where the value of the viewing user viewing the external page is adjusted upward by the quality metric; and
aggregating the plurality of quality score components to determine the value score for the post.
16. The computer program product of claim 11 , wherein the post accessed by the online system may comprise: an organic post from an individual user, or sponsored content from a business or organization.
17. The computer program product of claim 11 , wherein the link associated with the accessed post comprises one or more of: a URL to a web page, a deep link to a location within a mobile application, and a link to an app store to install an application.
18. The computer program product of claim 11 , wherein the external page associated with the accessed post comprises one or more of: a web page, a mobile application, and a state or page within a mobile application defined by a deep link.
19. The computer program product of claim 11 , wherein the quality metric is further determined based on attributes associated with the viewing user.
20. The computer program product of claim 11 , further comprising:
training a machine learning model using historical data about whether users viewed external pages for a threshold period of time after being presented with a post with a link to the external page; and
using the trained machine learning model to compute a quality metric for a post containing a link to an external page.