Software Testing for Big Data Analytics at National Software Testing 2017

The traditional approach to software construction and delivery provides data validation in the input routines to ensure that only valid data is entered into the system. It can also be used to enhance system security by e.g. preventing input buffer overflow problems. It has always been assumed that such systems will retain a high degree of data cleanliness.
In the world of the IoT and development of Big Data Analytics for all forms of Internet based data sources such input data validation is not feasible; often the sensors are of very uncertain calibration and accuracy. This led to J Easton’s observation that some 80% of all the data that we need to use is of uncertain veracity. As a result, it is now the case that traditional approaches to system development and testing are no longer sufficient.
The session will evaluate some of the practical and governance issues raised by current approaches to the development and delivery of IoT device Apps and areas such as Sentiment Analysis.
It also identifies critical issues that now arise because of the imminent arrival of the GDPR in areas such as testing Security and Privacy by Design and Default and the need to be able to process the withdrawal of consent to use of share personal data.