Random Query Answering with the Crowd

Random data generators play an important role in computer science and engineering since they aim at simulating reality in IT systems. Software random data generators cannot be reliable enough for critical applications due to their intrinsic determinism, while hardware random data generators are difficult to integrate within applications and are not always affordable in all circumstances. We present an approach that makes use of entropic data sources to compute the random data generation task. In particular, our approach exploits the chaotic phenomena happening in the crowd. We extract these phenomena from social networks since they reflect the behavior of the crowd. We have implemented the approach in a database system, RandomDB, to show its efficiency and its flexibility over the competitor approaches. We used RandomDB by taking data from Twitter, Facebook and Flickr. The experiments show that these social networks are sources to generate reliable randomness and RandomDB a system that can be used for the task. Hopefully, our experience will drive the development of a series of applications that reuse the same data in several and different scenarios.