Facebook grouping similar posts
Tue, Aug 9 2011 15:40 PM | 
balajik Facebook yesterday officially announced that it is grouping similar news feed stories based on keywords.So if any of your friends speak about iPhone ,then they will be grouped under the topic "iPhone" with a link to the Fan Page.
Facebook has said that it will be using Natural Language Processing to match words in status updates to brand pages, but it will not be using human editors or sentiment analysis to filter out or cluster negative or positive stories, or corral unrelated elements. That makes me worried because if you accidentally post something that has the word “iPhone” then it may be grouped together with a completely random other story about an iPhone.To make things worse, same topic with different sentiment would be grouped together.
What do you guys think? Should it be more than just 1 topic? Do they have to consider other features than just topic/keyword.Should they be using more sophisticated way of grouping the related posts?
Any thoughts?
| 
Tags: 
Dangers of mis-using NLP
By bringing together a lot of content clustered around a single topic with out any other filters I fear there will inevitably be examples of bad content juxtapositions.
Possible example:
Cheap Airline Tickets Here!
Airline Crash Kills Dozens!
Would I want my post about a certain subject put in the same page/news feed with other content that just happens to have the same content? Not sure as it could lead to some unsavory connections.