Quality has a link to the notion of zero defects in process modelling (Crosby 1995, Wing 2019), and data quality is no exception to this observation (Nagle 2018). Information systems can be viewed from at least perspectives: the data perspectives and the application process. Of these two, the data perspective is more important as many application domains are data intensive and changes in data influence the application process logic, but not necessarily vice versa.
Data quality problem undermines the usability of data. Data only matters if it is correct and useful (Webster 2014). Research or business based on data with errors do not meet the requirements of good research or business in term of accuracy and consistency (Anderson 2008), and this phenomenon will likely incur biased or wrong conclusions and vulnerability of causing ‘big errors’ brought by big data. The grandiosity of data needs to not only refer to its large volume, complex data structure and fine granularity but also the significance of data quality and usage in data (Firmani 2016, Liu 2016).
Data modelling techniques were developed for the specification of the data perspective in a way of providing concepts for the description of data types and structural relations between them (Frederiks 1997). Although many data modelling techniques have been proposed in the literature (Allemang 2008, Batini 2018, Hingorani 2017, Mancas 2021, Mkhinini 2020, Poblet 2019), the question of how to achieve a good data quality remains yet to be answered. There has been speculation that the lack of a management science for data quality may contribute to low data quality levels (Fox 1994, Redman 1998).In fact, the lack of a management science contributed to data quality issues adversely.
According to HBR study (Nagle 2017), only 3 percent of companies meet data quality standards, 47 percent of newly created data contain at least one critical error and half of the data quality scores are below 57 percent. It is important to realise that data are different from other resources and require different management techniques, thus they need a science, at a minimum, that can provide a solid foundation for addressing data quality issue. Nevertheless, even data science failed to address the issue or guideline of data quality correctly (Cao 2017, Newman 2016, Priestley 2019), since it instead mostly focuses on extracting useful information from datasets. Unlike vast array of literature on software-related requirement engineering within the context of software engineering, literature devoted to data-related conceptual modelling quality is limited regardless of ER model, object-oriented model or whatever (Cherfi 2002).