Temporal Graph Warehousing for Big Data Analytics

Data warehouse has been considered as a kernel technology of deriving business intelligence from big data, and used for creating multidimensional on-line analytical processing (OLAP) (or big data analysis) reports for administrative decision-makings (Inmon, 2005; Kimball & Ross, 2013). An indispensable need is tracking the timing of changes in dimensions (Tansel, Clifford & Gadia, 1993; Ozsoyoglu & Snodgrass, 1995; Snodgrass, 2000; Kulkarni & Michaels, 2012), together with related business activities, to create business intelligence reports more accurately. One of the common excruciations in maintaining or utilizing cubes is the fact that many dimensions, except for the time dimension, usually change over time. Such dimensions are called slowly changing dimensions (SCDs) by Kimball & Ross (2013), as they change slowly and unpredictably. In this paper, we would like to propose an emerging research roadmap regarding the cross product of {non-temporal, temporal} x {data, graph} warehousing, which inspires the following four kinds of on-line analytical processing models; i.e., traditional data warehousing (Inmon, 2005; Kimball & Ross, 2013), temporal data warehousing (Golfarelli & Rizzi, 2009), non-temporal graph warehousing (Sakr et al., 2021), and temporal graph warehousing. The evolution of these models makes a more subtle and precise big data analytics in cloud event tracing applications. For example, digital contact tracing for COVID-19-related applications, or digital footprint summarization involved between connected data, connected people, and connected computers in contemporary AIoT applications