Liao Ruoxue, lead architect of Baidu revealed this week that a 100+ people team in the company has been working on a recommendation system since last year. The system has been implemented into many services like Baidu Ting (online music radio), Baidu Zhidao and Baidu News as underlying technology.

Baidu’s Dragon Boat Festival Doodle

Acquiring information online has gone through 3 different stages. The first stage was to browse by categories when the information online is still in scarcity; The second stage was the era of search engine like Google and Baidu; And the next one belongs to the recommendation system that could predict a user’s ‘mind’ and push the right thing to them. No more hard time Googleing through tons of web pages. Liao also said that Baidu’s goal is to “Define and seek users’ hidden demand, especially when these demands are sometimes too complicated to be explained by users themselves.”

The technology of recommendation has become very popular these years, companies like Google, Youdao, Bing, Amazon, Pandora and Netflix all joined. For example, when viewing a product on, you must have seen items recommended by the etailer, which is all based on a matrix of what other shoppers bought or viewed that have association with what you’re viewing right now.

Baidu’s recommendation system features interdisciplinary, platformization and intelligence. Take Baidu Zhidao as an example, it usually takes more than 12 hours before someone eligible coming across a new question on the Q&A service. While with the recommending mechanism, the time can be shortened to less than 5 minutes.

Currently this technology is supported by the users database of Baidu, which stores hundreds of billions webpages and processes 100PB data everyday. “When the traditional mode does not work, I think we might make it a product focusing on recommendation.” Liao also predicted that, “When the technology is ready, it may cooperate with vertical websites like”

Although Baidu leads its peers in this area, it still has a long way to go. “This area is still new, Douban FM got started early, but none has achieved much.”

Liao pictured an interesting scene for us: in the future, when you enter “nice movies” in the search engine, you will be linked to your favorite Korean movie (because the engine knows you are a big fan of Korean drama); and when you enter “how to propose to a girl”, you will see a customized plan just for you and your girlfriend. Isn’t that cool for you?

动点科技驻湾区记者. Charlie is an entrepreneur based in San Francisco and Hong Kong who calls herself the undefeated caffeine champion. You can reach her at charlie.sheng (at)

Leave a comment

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.