Yesterday on its Engineering Blog, Twitter shared an intriguing peek into how its real-time search is refined with the help of humans. The company offered an in-depth look at the process of how it handles real-time searches and events. Twitter lays this two challenges these sudden events pose from a search and advertising perspective:
1) “The queries people perform have probably never before been seen, so it’s impossible to know without very specific context what they mean. How would you know that #bindersfullofwomen refers to politics, and not office accessories, or that people searching for “horses and bayonets” are interested in the Presidential debates?
2) “Since these spikes in search queries are so short-lived, there’s only a small window of opportunity to learn what they mean.”
To teach its systems how to find out what these spiking queries mean, Twitter uses a “real-time human computation engine.” In other words, as search queries are trending, they’re sent to live human “judges” who are asked several questions to determine things like what category a query refers to, who it refers to and if they have a Twitter handle, and if there are pictures of the query.
An interesting part of the post is how Twitter refers to this process helping the company not only to deliver but relevant ads as well. This kind of intricate and human-aided process of improving context and relevance should, ideally, be a boon for both users and advertisers.
Now if only we could get those pesky humans out of the way and hand everything over to the robots.
Network
