After taking the course of Analytics and Visual of Big Data, we are all aware of the great ability and advanced functions of date mining technology. From the blogs, we’ve seen a great deal of applications of various fields, such as in Sports, weather, business and et cetera. However, today I’d like to write something about the shortness and drawbacks of the data. Absolutely there must be something that big data can not do or only can do poorly.
Thing first come to my mind is that data struggles with social. Compared to human being, computer-driven data analysis are pretty good at measuring the quantity of social interactions rather than the quality. Unlike people, data analysis rarely can do few things in mirroring each other’s emotional states, detecting uncooperative behavior and assigning value to things through emotion. Thus one would draw a wrong conclusion about how close social relationships between people rarely based on the number of days they meet each other or calls they make.
Plus, data mining somehow might “favor memes over masterpieces”. Due to employing dimensionality reduction approach like singular-value decomposition (SVD), data mining software have to filter and ignore relatively large data, in which useful information might exist. For instance, data analysis are most likely to detect when great numbers of people take an instant liking to some cultural product. However many important products of great value are hated at the first beginning only as result as the factor that they are unfamiliar.
Data also obscures values. There is saying that, in a sense, the data is never raw, which means that it’s always structured in order to achieved people’s predispositions and values. In fact, from construction to interpretation, there are always value choices exist.
Well, this is not to say that data mining isn’t a great technique. But like every coin has two sides, it has strong merits in some aspects, in the meantime, it is supposed to have a little shortness as well.