Data quality in a big data context

"In each of the phases of a Big Data analysis process, data quality (DQ) plays a key role. Given the particular characteristics of the data at hand, the traditional DQ methods used for relational databases, based on quality dimensions and metrics, must be adapted and extended, in order to capture the new characteristics that Big Data introduces. This paper dives into this problem, re-defining the DQ dimensions and metrics for a Big Data scenario, where data may arrive, for example, as unstructured documents in real time. This general scenario is instantiated to study the concrete case of Twitter feeds. Further, the paper also describes the implementation of a system that acquires tweets in real time, and computes the quality of each tweet, applying the quality metrics that are defined formally in the paper. The implementation includes a web user interface that allows filtering the tweets for example by keywords, and visualizing the quality of a data stream in many different ways. Experiments are performed and their results discussed."