Data X-Ray: A Diagnostic Tool for Data Errors

By Xiaolan Wang, Xin Luna Dong, Alexandra Meliou. 2015

96 citations recorded by the source corpus.

A lot of systems and applications are data-driven, and the correctness of their operation relies heavily on the correctness of their data. While existing data cleaning techniques can be quite effective at purging datasets of errors, they disregard the fact that a lot of errors are systematic, inherent to the process that produces the data, and thus will keep occurring unless the problem is corrected at its source. In contrast to traditional data cleaning, in this paper we focus on data diagnosis: explaining where and how the errors happen in a data generative process. We develop a large-scale diagnostic framework called DATA X-RAY. Our contributions are three-fold. First, we transform the diagnosis problem to the problem of finding common properties among erroneous elements, with minimal domain-specific assumptions. Second, we use Bayesian analysis to derive a cost model that implements three intuitive principles of good diagnoses. Third, we design an efficient, highly-parallelizable algorithm for performing data diagnosis on large-scale data. We evaluate our cost model and algorithm using both real-world and synthetic data, and show that our diagnostic framework produces better diagnoses and is orders of magnitude more efficient than existing techniques.

Subfields: Information Systems, Information Systems and Management, Management Science and Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Information Systems

In Information Systems and Management

In Management Science and Operations Research

Citations in the corpus

Automated Machine Learning: State-of-The-Art and Open Challenges (2019, neutral): “Data X-Ray [127] has been introduced as a diagnostic tool for data systematic errors that are inherited from the process of producing the data itself.”
MacroBase: Analytic Monitoring for the Internet of Things (2016, methodology): “Data X-Ray [46] uses a Bayesian-inspired metric to find interesting subgroups within hierarchical datasets.”
Overview of data quality challenges in the context of Big Data (2015, background): “All the above discussed techniques apply data cleansing techniques upon the data itself, but do not attempt to correct the cause which create the errors and dirtiness in the data; as most of those data quality issues are reported to be systematic, thus inherent to the process of data creation, it is”
Combining Quantitative and Logical Data Cleaning (2015, methodology): “In contrast, when errors are introduced systematically, for example by a faulty information-extraction process, then other approaches, like the Data X-Ray [25], are more appropriate than ours.”
MacroBase: Prioritizing Attention in Fast Data (2016, methodology): “The literature contains several recent explanation techniques leveraging decision-tree [23] and Apriori-like [39, 80] pruning, grid search [68, 82], data cubing [69], Bayesian statistics [78], visualization [17,62], causal reasoning [81],”