Big data is an wide term, considering recent flight disasters , we feel Big Data can help us to trace the causes, so my question is, can Big Data efficient for Flight Data Mining? is there any recent survey paper on this area?
Big data contain structure or unstructured data and flight data mining is also a unstructured data. So mining such huge data is efficient for flight data but it needs some tools to make it faster like hadoop map deduce
Here are a few papers that are emerging around this topic. This is going to be a huge play for big data since many of the "things" in the IoT are going to need to be tracked as they move in multiple dimensions. We are about to see an explosion of "things" that can (and need to be) tracked at rest and in flight (drones will escalate the urgency for this due to security concerns - watch for the escalation in launches of military satellites). Here are a few articles that you may find useful:
here's an interesting perspective from the farming community, which will make heavy use of drones and big data as agribusiness become further immersed with technology (on multiple levels)
Big data implies that you have a really large amount of data. 20-50 TB is not big data. If you have you have the data flight recorder data and some radar data you have a limited amount of data. You can acquire meteorological data and a 3D terrain model of the flight path so you can generate a simulation environment and do monte carlo simulations varying variables you don't have or that are imprecise, i.e. micro-meteorological occurrences. This is a massively parallel problem for which map-reduce may or may not be appropriate. Your problem is not within usual domain but first define your problem, then define your first solution approach and then see what's the best way to do it. You have data and you can construct a model so that is likely not the issue. If the simulations don't show anything, you'll look further out. A fuselage crack can be defined for the model by an aeronautical engineer define the possible behavior that would result.