Facebook NLP Group Postings

August 30, 2011

ZDNet reports on the technical side of Facebook’s newest adjustment to the ubiquitous news feed in, “Facebook using natural language processing to group posts, link to Pages.”

Facebook has added a new type of story to its News Feed today: if more than one of your friends post about the same topic, and it has a Page on the social network, the posts will be grouped under a Posted About story, even if your friends don’t explicitly tag the Page. The story is posted in the following format: “[Friend] and [x] other friends posted about [Page]” where the last part is a link to the Page in question. It turns out Facebook is using natural language processing on status updates as well as the headlines of posted links to figure out if a topic mentioned has a corresponding Page, and then searches to see if your other friends have done so as well.

This is Facebook’s first attempt to match the “trending” ability of Twitter. I’m just not sure that it’s effective yet. As a librarian and a Facebook user, there’s a personal and a professional side to this story for me. The privacy aspect of the social network prevents a metasearch that would allow a user to find relevant posts by subject across the platform. So in its place, it seems Facebook is attempting to show you common themes among those within your friend group, those to whom you have access.

A sure complaint will be the mile-long list one has to scroll through when a friend’s birthday is being universally acknowledged. Sometimes the groupings don’t even make sense, as is the case when two people on different days post to the same person’s wall, but the news feed is grouped together. The only common denominator in that equation is the individual whose wall was being written on. The topics can be completely separate and yet somehow wind up grouped together.

The real power that Facebook should leverage with this feature is the ad revenue potential. The ZDNet article shows an example of several friends discussing the new Harry Potter movie, with a news grouping headline of “So and so and two other friends also posted about Harry Potter.” “Harry Potter” then acts as a link to the product page, in this case, the movie promotion page. So here there is potential to take casual conversation, use NLP to pick up on products or services that also use Facebook, and then direct those users back to the product pages for that particular product or service. If Facebook could find a way to do this effectively, and get companies to buy in on this type of targeted marketing, it could be much more effective than sidebar ads. Expect to see more experimentation from Facebook with natural language processing.

Emily Rae Aldridge, August 30, 2011

Comments

Comments are closed.

Search the site
Subscribe to Beyond Search
Feature archive
News archive

Stephen E. Arnold monitors search, content processing, text mining and related topics from his high-tech nerve center in rural Kentucky. He tries to winnow the goose feathers from the giblets. He works with colleagues worldwide to make this Web log useful to those who want to go "beyond search". Contact him at sa [at] arnoldit.com. His Web site with additional information about search is arnoldit.com.