学术界：大数据关键是“落地”

    他们说：大数据将拯救世界，改变我们做生意的方式，还会告诉我们很多不了解的事。

    这些说法中有些可能会成为现实，有些可能不会——但关键在于，围绕“大数据”所开展的大肆造势铺天盖地，亟需做点剖析。这项技术确实有前景，但滥用这个说法就是另一码事了。

    不管实际情况如何，“大数据”仍然是个新兴领域。对各行各业来说，要从自己的数据中心及美国政府的数据库新开放的数据宝藏中发掘各种商业机会还为时尚早。伯克利大学（Berkeley College）教授达尔尚•德赛称，看看“谷歌趋势”（Google Trends）就会发现——“大数据”直到2011年还很少被用作搜索词，直到2012年才真正开始流行。

    不过德赛说：“我还是新技术的超级拥趸，我深信有很多技术可以用来改善生活。”

    德赛是在一个研讨会上发表这些观点的。这个研讨会名为“理解大数据”，主要由学者参加，是由纽约大学城市科学与进步中心（Center for Urban Science and Progress）的康斯坦丁•康托可斯塔组织的，与会者还有来自普拉特学院（the Pratt Institute）、圣弗朗西斯学院（St. Francis College）和纽约技术学院（the NYC College of Technology）的专家。它是“科技三角U”（Tech Triangle U）的一个组成部分，后者是由布鲁克林科技三角（Brooklyn Tech Triangle，纽约新兴科技创新中心——译注）发起的一个项目，旨在推动布鲁克林科技界和学术圈建立联系。

    每位与会专家都就大数据未来将对自己的研究领域产生什么样的影响阐述了自己的看法。康托可斯塔领导着一支开展建筑信息学研究的团队，主要研究如何用数据科学来分析城市能源消费。他在会上预先透露了对《第84本地法》（Local Law 84）推行后一些公开数据的研究成果。这个法案要求纽约市大型建筑必须披露对能源和水的利用情况，但它只要求公布相关数据——并没有要求这些建筑改变能源运用方式。不过他表示，它确实为这些建筑做出改变带来了两种压力。

    第一种就是竞争压力，因为这些建筑将密切关注同类建筑的情况，同时开始根据能源利用效率争取住户。康托可斯塔说：“我们能利用这一点比较纽约不同地区的能源利用情况，还能了解住户能源消费水平的差异。”

    它意味着像Radiator Labs这类公司将有望打开市场。这家公司的产品基于无线局域网工作，能解决老旧建筑中蒸汽供暖水平不均的问题。随着基准数据开始让物业经理明白他们和同类建筑在能源利用水平上存在差距，他们可能就会对这种产品产生兴趣。

    康托可斯塔称，第二种压力将来自上层。纽约公用事业公司Consolidated Edison已开始设法通过新的方式来管理峰值需求，也就是在用能高峰期间向大客户付钱，让他们自愿减少能耗。有了更多能源使用数据后，能源基准数据城市的大型公用事业公司可能就能开发出更好的节能战略。它包括和EnergyHub这样的公司合作，招募消费者参加自愿项目，在能耗趋于峰值时通过设备和互联网将家里的空调调整几度，或者干脆断电。

    圣弗朗西斯学院的埃米莉•霍洛维茨表示，光收集一大堆数据还远远不够。要想更好地改变现状，搜集数据只是必需的一步，但还很不充分。她说：“在我看来，大数据最大的问题在于，通过它能看到太多东西。”凭借大数据技术，研究者终于能首次把众多不同数据放在一起，但这并不意味着他们就能揭示什么新东西。这样确实能揭示相关性——但要证明因果关系就要难得多。

    与会专家还有普拉特学院的杰西•布雷登和纽约市立大学的杰森•蒙哥马利，但大数据什么时候，以及怎样才能催生大家能感受到的与众不同的客户服务，他们并没有解决这个问题。众多新技术公司都宣称更多数据将带来更好的服务——个性化电子商务就是其中一个领域，也就是说，如果你在某个电商平台上评论了更多产品，这个网站就会产生更好的产品改进建议——但现在还不清楚这种说法到底有多大的真实性。

    不过，各类提供数据服务的公司正在大张旗鼓地展示，它们正在帮助各类企业积聚、利用它们业已收集到的数据。同时各种数据代理商也正在收集数量惊人的关于消费者身份及（大致）需求的各类信息。

    不过与会专家都认同的一点是，细节决定成败，获得这么多数据是一回事，要真正理解它们、用好它们却完全是另一码事。大数据可能会越来越流行，但要真正获得重大成功仍然需要努力。（财富中文网）

    译者：清远

    Big data will save the world, they say. It will change the way we do business, they say. It could tell us things we didn't know we didn't know, they say.

    Some of that might be true and some of it may not -- but the point is that the hype around the term "big data" is thick enough to require a chainsaw to cut through. The technology is promising; the semantics are another story.

    Whatever the reality, "big data" remains a nascent field. Businesses are a long way from seeing all the opportunities that could come of the newly opened troves of data in their data centers and in those of the U.S. government. Just look at Google Trends, Berkeley College professor Darshan Desai said -- "big data" was barely used as a search term until about 2011 and didn't quite take off in popularity until 2012.

    Still, "I'm a big fan of the new technology," Desai said. "I believe there are ways to make lives better."

    Desai made her remarks as part of a panel of academics led by Constantine Kontokosta of the NYU Center for Urban Science and Progress. The panel, called "Making Sense of Big Data" also featured scholars from the Pratt Institute, St. Francis College, and the NYC College of Technology. It was part of Tech Triangle U, an initiative by the Brooklyn Tech Triangle to connect the technology and academic communities in Brooklyn.

    Each scholar demonstrated how they thought big data would impact their fields of study. Kontokosta, who leads a building informatics research group that focuses on the application of data science to the analysis of urban energy consumption, gave a preview of some of his work on data released under Local Law 84, an energy and water usage disclosure law for large buildings in New York City. The law itself is pure data -- nothing in it forces buildings to change their practices around energy. But it does create two kinds of pressure on buildings to change, he said.

    The first is competitive pressure, as buildings monitor their peers and begin to compete for residents on efficiency. "We can use this to compare how much energy use is varying across the city," Kontokosta said, "but also how much the difference is in how much people are paying."

    Which means companies like Radiator Labs, which offers a Wi-Fi-enabled product that addresses uneven steam heating in old buildings, may start to see property managers taking interest in their wares as benchmark data starts to show that they aren't keeping up with peer structures.

    The second pressure will be from the top, Kontokosta said. The New York utility company Consolidated Edison is already working to manage peak demand by paying large customers to voluntarily reduce energy consumption during peak demand. With more usage data, big utilities in energy benchmarking cities may be able to develop better demand reduction strategies. Those could involve partnerships with companies like EnergyHub to recruit customers into voluntary programs where home thermostats can be adjusted, by the utility and over the Internet, a few degrees when demand is spiking and a blackout is possible.

    But assembling a bunch of data isn't enough, said Emily Horowitz of St. Francis College. It's necessary but not sufficient to make change for the better. "The big problem with big data, in my view, is you can see all kinds of things," she said. Researchers may be able to put many different kinds of data side by side for the first time thanks to big data technology, but that doesn't mean that they have revealed anything. Correlations, sure -- but proving causation is much more difficult.

    The panelists, which also included Pratt's Jessie Braden and CUNY's Jason Montgomery, did not address the questions of when and how big data will deliver appreciable differences in consumer services. Many new technology companies have claimed that more data leads to better results -- e-commerce personalization is one such area; if you rate more products on a retailer's website, you are told that it will lead to better product suggestions on that site -- but it's unclear how true that statement actually is.

    Nonetheless, companies offering data services are making a strong showing helping businesses aggregate and make use of the data they're already collecting. And data brokers are collecting an alarming array of information about who we are and what we (presumably) want.

    But the devil is in the details, and making sense of all that information is an entirely different proposition from merely accessing it, the panelists agreed. Big data may be increasingly popular, but it's still looking for its first big hit.