Alibaba Innovative Research (AIR) > Machine Learning (algorithm)
User Growth Problem in Taobao Search

Theme

Machine Learning (algorithm)

Topic

User Growth Problem in Taobao Search

Background

As the number of internet user in China grows to 900 million,new internet user becomes very difficult to touch.  Since user is the basic of all internet companies, how to increase current user’s engagement so as to increase each user’s ARPU(average revenue per user) is an important problem for companies. Companies spend billions on acquiring new users, and it’s much more important to keep those users active in their apps. 

Taobao is a marketplace which attracts hundreds of millions of users to visit each day. Search is important in Taobao to help users find products and more than 100 million users will use search every day. Our target is to find ways to increase users’ engagement in Taobao Search. There are many metrics of engagement. Active days of one month mean the average visit days of one user in one month. Average page views/clicks/queries for one user are another metrics. 

There are some methods to increase user’s engagement. Group people and give different strategies to each group. For example, for new users we can give some discount coupons to stimulate them to buy. The problems are how to group people and what can we do to increase such group people’s engagement.

How to balance the current reward and the long-term reward is another challenge. Current reward means clicks/queries/paid money for this visit. Long term rewards are those rewards which will happen in the future. The average gap for the user to visit the app again is an example. Total paid money of this user for the next 6 months is another example.

User growth in search or recommendation is a new problem which has few researches in academic community. We hope this problem can attract more attentions.

Target

  • Method of increasing user’s engagement
  • Research framework of user engagement
  • Method of grouping people
  • Algorithms for balancing current and long-term reward

Related Research Topics

  • metrics of user engagement
  • short and long reward of search(recommedation) system
  • framework of user growth

 

Suggested Collaboration Method

AIR (Alibaba Innovative Research), one-year collaboration project. 

Scan QR code
关注Ali TechnologyWechat Account