I graduated from Tsinghua university and got my bachelor degree in 2004. I am working on computational advertsing, audience targeting and data mining. I have a very strong interest in algorithms, recommender system, ads, data mining and machine learning.
As a data science team member, I focused on data collection (Sina Weibo, Tencent Weibo, Facebook, Linkedin…), data analysis and data mining. I delivered some backend data process pipelines and a celebrity classifier which can be found here, 必应影响力 for BingScore project, and also some algorithms to detect hot tweets, detect moments for celebrities. Now I am working on Revolve, which is an upcoming app for Windows phones which melds aspects of a calendar and contact manager, and presents you with information about people you’re going to meet with that it’s collected from multiple sources.
As the tech lead, I focused on the whole E2E scenario about recommending “People You May Know”. In order to have a higher accuracy and more coverage, besides the 2-degree social graph, I expanded the signals to interest tags, demographic information, including the education and career information, and the communication frequency between users. For example, if you are interested in ML, weibo will recommend ML experts to you. If you graduate from Tsinghua University, weibo will recommend some schoolmates to you, and so on. We also have a pipeline to calculate the interest distribution according to the follower-followee relationship. We have 25M requests per day with 5000qps and 100ms latency.