Is the big data really reliable?

The popularity of the internet, so that in recent years the information explosion in the world, and the arrival of the Internet of things, but also increased the network of information load, these information knot weaving every node of the network, become a data, "big data" with the entry into People's view.

In the eyes of the left-wing http://www.aliyun.com/zixun/aggregation/7331.html "> Tech Manager, Big data is like the four-dimensional pocket of a robot cat--omnipotent, and now, many big data technologies, such as mathematical models, predictive algorithms and artificial intelligence software have been widely used. From daily life to business, sports, medical, and even military areas, all data-related problems, as long as the "pocket" from the corresponding information tools to analyze, can be resolved.

However, as the ironic quip of Mark Twain, Will Rogers and Charles Caitlin said: "It is not what we do not know that leads us into the predicament, but what we know, but not the real." The big data that we think we know well may not be as real as we seem.

Before large data, the Internet field has been leading the human trapped in the "true and False event": The 90 's true and false Internet makes a variety of internet concept stocks rampant, the 2003-2005 true and false ISP, the concept of mixed up the whole industry; True and False e-commerce and so on. Therefore, some experts pointed out that from 2012 to 3721.html ">2014 year will be a big data field of melee." The identification of large and false data should not be overlooked.

The current data algorithm and analysis for our digital world is too simple and not smart enough. Mathematical model is a simplified model, based on natural science, physical law, particle behavior in the fluid and so on these data prediction reliability may be considerable. But human beings are a complex organism, the randomness of human behavior is very large, so the probability of "Non sample Error" In the analysis of human behavior increases greatly.

Therefore, large data is important, but intuition is still essential, and many, including technical limitations and human factors, constrain the reliability of large data analysis. The ability of human beings to recognize social cognition can not be replaced by artificial intelligence at present, social relationship analysis, contextual analysis, etc.

In addition to the imperfect technology and the uncertainty of human behavior, resources such as large data cannot be fully shared and are an important factor affecting their reliability. Each person looks at the matter angle and the attention point is different, to the information gathering point also different, but each person also is imperfect, the flaw inevitably exists. Recently, Huang, vice president of Beijing Micro-venture Investment Management Co., was interviewed by reporterssaid: "Each company has to do a large set of data, in fact, the cost is very high, as you have the data I buy data, or you have data we docking data, so, in fact, is the real cloud data." The sharing of data resources between enterprises can better avoid one-sided information. "

can be seen, in a large data era, the explosion of information," signal and noise coexist, to use large data to make more accurate predictions must learn to distinguish "signal and noise", do not let "noise" misleading. At the same time, "signal" sharing between, also can better for the data information "positively".

Of course, the benefits of large data for human life are now well known, and the co-evolution of human, dataset and analytic algorithms can make large data create new values and wealth.