Enterprise Dark Data

Abstract

The increasing amount of digital data and the declining cost of data storage have led to the fact that companies began collecting any the data possible, regardless of its adequacy and usability. This results in increasingly diverse data, in terms of its structure, quality, availability and the source of origin. Dark data is one type of data that increases significantly as the volume of data expands. Scientific literature does not precisely define the term “dark data”, while its interpretation among scientists is ambiguous. The aim of this article entails an attempt to define the dark data occurring in an enterprise, by identification of its essential features. The article presents an overview of the definitions of the term dark data, a proposal of its interpretation, and a classification of data in a company with regard to: usability, availability and quality. The analysis of the concept of dark data was carried out via a review of international journals and articles published on the Internet by Data Science practitioners. As part of the research, four universal features of dark datasets have been indicated (unavailability, unawareness, uselessness, and costliness). Based on data availability and its quality, four groups of enterprise data have also been distinguished. The data classification developed in this way allowed systematization of the term “dark data”.