Big Data and AI Pipeline Framework: Technology Analysis from a Benchmarking Perspective

Abstract

AbstractBig Data and AI Pipeline patterns provide a good foundation for the analysis and selection of technical architectures for Big Data and AI systems. Experiences from many projects in the Big Data PPP program has shown that a number of projects use similar architectural patterns with variations only in the choice of various technology components in the same pattern. The project DataBench has developed a Big Data and AI Pipeline Framework, which is used for the description of pipeline steps in Big Data and AI projects, and supports the classification of benchmarks. This includes the four pipeline steps of Data Acquisition/Collection and Storage, Data Preparation and Curation, Data Analytics with AI/Machine Learning, and Action and Interaction, including Data Visualization and User Interaction as well as API Access. It has also created a toolbox which supports the identification and use of existing benchmarks according to these steps in addition to all of the different technical areas and different data types in the BDV Reference Model. An observatory, which is a tool, accessed via the toolbox, for observing the popularity, importance and the visibility of topic terms related to Artificial Intelligence and Big Data technologies has also been developed and is described in this chapter.