Large data and affective analysis: words and Chatterbox

Smogfarm engaged in emotional analysis based on large data, measuring, tracking, and aggregating millions of of people's emotions, and then making the world's first quantitative indicator of mass psychology-the pulse of the Earth. Why would you do that? Because the data is the future, emotion is a big deal, elections, survey results, popular news stories and even the stock market will be affected by "crowd" sentiment.

More words must be

Kredstreet, by contrast, has found another vertical business model that seems more likely to be realizable and has begun a closed test: Social stock traders rank (the Social Juno rankings). The Social stock Juno rankings An emotional analysis of the stocktwits data, sampling the Twitter stream to determine whether the trader is bullish or bearish. It will also compare the current performance of the stock market with that of the trader in the past, and then grade and rank it according to the trader's accuracy rate.

The idea of this ranking system is really good. Because it's all about strength and results, and nobody can overwhelm a bombastic celebrity with a high rate of accuracy. In fact, this idea can be extended to many areas, such as advocating a strong voice is not the need to be famous public opinion network start-up State practice. To tell you the truth, this set of things does work better than diplomas and brick houses sometimes.

Of course, smogfarm/kredstreet and the emotional analysis of the flow, as well as summly engaged in natural language processing, as well as the Palantir engaged in data mining is still in the early stages, and very immature. But imagine, 5 years from then. If their big data, high-energy searchlight hits something like Facebook timeline, how much will they be able to discern?

A few years ago, EFF (Electronic Frontier Foundation) found that even something as simple as a browser setup would make your anonymity less than imaginable. Last week, a new study found that the human movement trajectory is highly unique, according to the study, from each anonymous cell phone in the direction of the data can be identified by only four points of time to identify the user, the recognition rate of up to 95%. Good software can dig out many treasures from these seemingly sparse, blank data.

Chatterbox

With the evolution of language and image processing software, once applied to something like Facebook timeline, it is possible to produce unexpected results. The words you say, the pictures you send out, the games you play, and so on, will gradually outline a horrifying portrait of yourself that may have been something you never wanted to disclose.

Worse, it could be surprisingly easy. Bosses may not need to read your Facebook timeline at all, just the file analysis software to your head, half a minute later, your work habits, neurasthenia, life frustration, emotional instability, attitude to the boss are all inferred-your photos, comments, likes, top, etc., This information is meaningless from the individual, but it can be pieced together to form your jigsaw puzzle.

It's a big deal. Not long ago, new security tools developed by IBM were able to use large data to identify security threats and disgruntled employees. Facebook and its ilk are also apparently counting on such large data applications to unearth large gold mines for their targeted ads.

But this can be reversed by the user. We can assume that once this tool matures, the performance on the web will not be as straightforward or as cautious as it used to be. Or even just look at it. Such networks may end up with only two outcomes: lifeless, insecure, or just talking about romance and duplicity.

Zuckerberg's law holds that information shared by Internet users doubles every year. But as soon as everyone realizes that chatterbox, can this rule continue?

Tools have no so-called good or bad, say more will or chatterbox, to see the user do not do evil, for users and businesses, the result is only two: win or lose, the use of the emotional become a better business, abuse of the data has no future.

(Responsible editor: Fumingli)