TDM: joining tensors and data model to avoid inconsistencies in Big Data analytics

Analytics tasks require multiple kinds of algorithms based on different theoretical foundations: linear algebra, statistics, graph theory, information theory, etc. Algorithms are implemented using different computing paradigms such as GPU based computing, map-reduce, concurrent programming, parallel programming, functional abstraction. Tensors are abstract and powerful mathematical objects used in multiple data analytics tools \[1\], including deep learning to represent multi-dimensional data or data mining to analyze latent relationships in complex data sets with tensor decompositions (\[2\], \[3\]). So, tensors cover a wide range of applications in different domains, for example in earth sciences to study climate change \[4\], in medical research to determine the orientation of directional diffusion in brain imaging data \[5\].