The main element of Google's real-time results are tweets. Tweets are the real-time messages that Twitter users post on Twitter.com. Google's Amit Singhal, who led the development of Google's real-time search, recently revealed how Google ranks tweets in the real-time results.
There's some kind of PageRank just for tweets
Google's PageRank algorithm looks at the link structure of a web page. The more websites link to a website and the more websites link to the linking websites the more relevant is the linked website.
Tweets are not about links but about followers. On Twitter.com, people "follow" the comments of other Twitter users. The more followers a Twitter user has, the more reputable are the tweets of that user. If Twitter users that have many followers follow another Twitter user then these users will have a larger impact on the reputation of that user.
"It is more than a popularity contest", said Google's Amit Singhal. "One user following another in social media is analogous to one page linking to another on the Web. Both are a form of recommendation.
As high-quality pages link to another page on the Web, the quality of the linked-to page goes up. Likewise, in social media, as established users follow another user, the quality of the followed user goes up as well."
There are additional filters and algorithms
The follower reputation rank is only one of Google's methods to rank tweets:
- Hashtags
Twitter users often use "hashtags" in their comments. Hashtags are symbols that start with a # followed by a popular topic, for example #earthquake.
If such a hashtag is included in a tweet, the tweet will show up in the real-time results when other Twitter users click the hashtag's topic word elsewhere on the site.
- Spam
While hashtags can be useful to maximize the exposure of a tweet, they are also often abused for spamming. The wrong hashtags can serve as a red flag that triggers Google's spam filters.
Amit Singhal didn't go into the details but he said that Google modeled the hashtagging behavior in ways that tend to reduce the exposure of low-quality tweets.
- The signal in the noise
There can be thousands of tweets that contain a very popular word such as "Obama". To find the relevant tweets, Google looks for "signals in the noise". Such a signal can be an increasing number of tweets that mention other words near mentions of "Obama", for example "Cambridge police". The tweets with the signals will be chosen for the real-time results.
The problem with Google's real-time results is that they don't last. The time and efforts that you have to invest in getting listed in Google's real-time results is better spent on optimization for Google's regular results.
No comments:
Post a Comment