Discovering the most important data quality dimensions in health big data using latent semantic analysis

Big Data quality is a field which is emerging. Many authors nowadays agree that data quality is still very relevant, even for Big Data uses. However, there is a lack of frameworks or guidelines focusing on how to carry out big data quality initiatives. The starting point of any data quality work is to determine the properties of data quality, termed ‘data quality dimensions’ (DQDs). Even these dimensions lack precise rigour in terms of definition in existing literature. This current research aims to contribute towards identifying the most important DQDs for big data in the health industry. It is a continuation of previous work, which, using relevant literature, identified five DQDs (accuracy, completeness, consistency, reliability and timeliness) as being the most important DQDs in health datasets. The previous work used a human judgement based research method known as an inner hermeneutic cycle (IHC). To remove the potential bias coming from the human judgement aspect, this research study used the same set of literature but applied a statistical research method (used to extract knowledge from a set of documents) known as latent semantic analysis (LSA). Use of LSA concluded that accuracy and completeness were the only similar DQDs classed as the most important in health Big Data for both IHC and LSA.

© 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.