The avalanche of (both user- and device-generated) multimedia data published in online social networks poses serious challenges to researchers seeking to analyze such data for many different tasks, like recommendation, event recognition, and so on. For some such tasks, the classical “batch” approach of big data analysis is not suitable, due to constraints of real-time or near-real-time processing. This led to the rise of stream processing big data platforms, like Storm and Flink, that are able to process data with a very low latency. However, this complicates the task of data analysis since any implementation has to deal with the technicalities of such platforms, like distributed processing, synchronization, node faults, etc. In this paper, we show how the RAM
S framework could be profitably used to easily implement a variety of applications (such as clothing recommendations, job suggestions, and alert generation for dangerous events), being independent of the particular stream processing big data platforms used. Indeed, by using RAM
S, researchers can concentrate on the development of their data analysis application, completely ignoring the details of the underlying platform.
This is an open access article distributed under the Creative Commons Attribution License
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited