Discovering Data and Information Quality Research Insights Gained through Latent Semantic Analysis

Discovering Data and Information Quality Research Insights Gained through Latent Semantic Analysis

Roger Blake (University of Massachusetts Boston, USA) and Ganesan Shankaranarayanan (Babson College, USA)
Copyright: © 2012 |Pages: 16
DOI: 10.4018/jbir.2012010101
OnDemand PDF Download:
List Price: $37.50


In the recent decade, the field of data and information quality (DQ) has grown into a research area that spans multiple disciplines. The motivation here is to help understand the core topics and themes that constitute this area and to determine how those topics and themes from DQ relate to business intelligence (BI). To do so, the authors present the results of a study which mines the abstracts of articles in DQ published over the last decade. Using Latent Semantic Analysis (LSA) six core themes of DQ research are identified, as well as twelve dominant topics comprising them. Five of these topics--decision support, database design and data mining, data querying and cleansing, data integration, and DQ for analytics--all relate to BI, emphasizing the importance of research that combines DQ with BI. The DQ topics from these results are profiled with BI, and used to suggest several opportunities for researchers.
Article Preview


Research in data and information quality (DQ) crosses several research disciplines and is becoming a unified body of knowledge. Starting with the pioneering works of Wang and Strong (1996) and Redman (1998), DQ has borrowed and adapted theories and techniques from many other areas including information systems, operations management, cognitive psychology, and organizational behavior, to name a few. In doing so, DQ researchers have applied a variety of research methods. Quantitative methods have been used to propose and test measurements for DQ. Models and representations have been defined for managing quality in organizations. Qualitative methods have been used to identify dimensions of DQ from the perspective of users and their context of usage. In the recent past, DQ research has embraced the design science paradigm by validating the usefulness of models, artifacts, and techniques in real-life settings. In this paper we have used the term “DQ” interchangeably with “information quality”, consistent with earlier research (Ballou & Pazer, 1985; Pipino, Lee, & Wang, 2002).

Given the extensive growth of DQ research and its likely continued growth in the future, it is important for researchers to understand the key research themes in DQ research and the popular research topics within each theme. DQ clearly has an enormous impact on the effectiveness of business intelligence (BI) and concurrent with DQ research, research of business intelligence has also grown significantly. The capabilities BI offers organizations have sharply increased and fact-based decision-making has become not just the norm for many companies, but is considered critical for success. As BI assumes ever more significance, so too does the need for a conceptual understanding of this field, as has been investigated by researchers (Foley & Guillemette, 2010). Since DQ is fundamental to BI, it is important to understand how the topics and themes of DQ research have evolved over time. It is also important to target BI-areas, related to DQ, that have not been addressed in the literature, and to identify the BI topic(s) within DQ that can garner the attention of practitioners and academics, both in BI and in DQ.

Concepts analogous to those found in DQ research can be found in BI but, these are often constructed and represented differently. These concepts may overlap but are difficult to associate. For instance, accuracy and completeness are well-known to DQ researchers as two important DQ dimensions, each defined differently and distinctly from the other. Data mining research has investigated the same phenomena, but generally considered both as forms of “noise”. Some data mining studies have defined noise in a manner that is very similar to the definition of accuracy in DQ research. Other data mining studies examining noise have used definitions similar to those for completeness in DQ research (Blake & Mangiameli, 2011).

Finding the core concepts in DQ research and how they relate to BI in order to point to opportunities for researchers is an important motivation for this study. Although they have not been explicitly connected to BI, there have been many attempts to define the core concepts of DQ research which have proposed frameworks to summarize and/or classify this area (Ge & Helfert, 2007; Lima, Maçada, & Vargas, 2006; Madnick, Wang, Yang, & Zhu, 2009; Neely & Cook, 2008). By examining the literature, each defines the classification framework from the respective researchers’ point-of-view. Although they offer invaluable insights into DQ research, we posit that there is a more interesting point-of-view that comes not from the researchers but from the research itself. What if the body of literature can inform us about the core themes and in addition associate the dominant core topics within each theme? What if we could examine relationships between those topics and themes and how they relate to business intelligence? The existing literature does not answer these questions. We believe that our methodology can answer these questions and more. Further, our methodology can be replicated to define the status of this (any) research field at any time in the future.

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 8: 2 Issues (2017): 1 Released, 1 Forthcoming
Volume 7: 2 Issues (2016)
Volume 6: 2 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing