Data is information/observations about the physical word, sometimes recorded in a temporal domain (across time). It is an abstraction/representation (model) of the universe/reality. This inherent feature introduces the concept of 'errors' and 'inaccuracy'.

Our perception of the universe and reality, which is reflected in the form of recording of data, looks over these initially introduced inaccuracies. For example, a meter rule may be able to measure up to 1 mm (10-3 m) and the data is recorded as so. However, the discovery of the elements and the size of their atoms have changed the then held views of how accurately humans perceived the real world. Existing and upcoming research on subatomic particles further boggles the mind about the whole concept of ''size/dimensions' and 'time'. Thus, the human perception of the universe is getting more and more 'fine-tuned', which is reflected in the recording system/precision/perception/meaning of data. This error can be thought of as error due to inaccuracy in human perception of the universe.

Thanks Nahian.

If I get it right, any recorded/abstract value which is different from a real-life value becomes an 'accuracy' error. It converges with some explanations forwarded throughout the data quality literature. My next challenge would be how to potentially identify those errors if there is a lack of knowledge about those 'real' life values!!

Hai Suraj,

In normal scenario inaccurate data will be certainly outlier and some sort of missing noisy data.As u feel, it become definitely challenging in big data to handle inaccurate data. It could be combated by proper integration in advance before gathering data in the initially stages to ensure usefulness. short meant effective cross checking during data collecting phase. later using anomaly detection we can track inaccuracies or exceptions. ample algorithms available to tackle anomalies.