Logical Model of Relationship for Online Social Networks and Performance Optimizing of Queries - WISE 2012 Challenge - T1: Performance Track Scalability Winner

Sina Weibo is currently the microblogging web service with the highest number of registered users in China. As in any large social network, the relationship representation is so huge that executing queries over the network is a very challenging problem. The WISE 2012 conference proposed a challenge based on Sina Weibo with two tracks: performance testing and repost prediction. This paper focuses on the first track challenge, which goal is to implement 19 queries with the highest throughput and the lowest latency, using a scalable parallel paradigm. In the input database, there are 265 millions of relations among more than 60 millions of users and more than 400 millions sent messages. This paper formalizes the logical model of the relationship in order to present the queries in precise and simple manner. Some optimization techniques were also proposed, such as the aggregate-rank-delete procedures, which can be applied to some of the queries for improving the performance. The proposed model and optimizations were implemented in a very scalable parallel system and the experimental results show that our solution can obtain high throughput and low latency for most of the queries.