彭博要不要提防Twitter

    彭博社（Bloomberg）长期以来一直以身为财经数据、新闻与分析的首要提供商而自豪，但它可以在提供新闻的速度上跟胜过社交媒体吗？

    投资者根据行业新闻而迅速作出相应的决策，数百万美元就会因此而转手。这就是为什么投资者愿意每年支付20,000美元的费用来使用彭博终端系统。构成彭博社核心业务的这项服务聚合了来自1,000多家新闻机构和90,000家网站的信息。这些信息对那些希望为自己的投资策略抢占优势的客户而言至关重要。一些公司——比如总部位于纽约的初创公司Dataminr——声称，他们可以获取、分析社交媒体上提供的信息，以比各大通讯社更快的速度来提供影响市场走势的重要信息。

    4月23日，黑客入侵美联社（AP）官方Twitter帐户，诈称白宫受到攻击。随后数秒钟之内，道琼斯工业平均指数便暴跌146点。虽然在这个骗局被揭穿之后股市迅速回升，但这起事件表明，现在有许多华尔街交易员依靠Twitter获取交易信息。

    Dataminr表示，它能够在重大新闻发布前两分钟就提醒客户小心骗局。这家初创公司与Twitter建立了合作关系，可以获得Twitter平台上每日近4亿条Twitter短信的全部“流水”记录。它设置了事件检测软件，可以查阅整个Twitter圈（Twittersphere），来识别与其客户设置的监视名单上列出的话题、行业或公司相吻合的活动热点或新出现的趋势。这家公司表示，通过这种方式，它可以向政府机构或金融公司提供有用的信息。

    这家公司在发现前美国国防部官员发布的一条Twitter消息之后，成为最先报道奥萨马•本•拉登被击毙消息的实体之一。在又出现19条涉及这个话题的Twitter短消息之后，该公司便向其客户发出了提醒消息，在这些客户收到这个消息的20分钟之后，传统新闻网站才发布了这个消息。（Dataminr发言人没有对就此事予以置评的要求作出回应。）

    在纳秒必争才能获得优势的新闻报道行业里，从一个没有多少地域限制的庞大网络中获得实时新闻和数据至为关键。比如，一家炼油厂发生火灾会促使汽油价格短暂飙升。虽然通讯社会在事件发生数分钟之后报道这个新闻，但来自事件实发现场的Twitter短消息可以实时提醒投资者，从而让他们可以在市场上完成一笔赚钱的交易。

    除了速度之外，Twitter数据分析还有望为各大公司提供所谓的情绪分析——即对人们的感觉进行估量。这种分析可以实时进行，不仅估量消费者对整个市场的“情绪”，而且还可以估量人们对具体的市场营销活动、产品发布及个人代言作出的反应。2010年，美国印第安纳大学（Indiana University）信息学教授约翰•博伦发表了一份研究报告。报告发现，Twitter上的社会情绪可以预测道琼斯工业平均指数的波动，而且准确率惊人地高达87％。博伦说：“Twitter平台绝对是一个非常有影响力的环境，社交媒体网站上的用户成为传感器。确实有群体智慧这回事。”

    Bloomberg LP has long prided itself on being the premier provider of financial data, news, and analytics, but can it deliver news faster than social media?

    Millions of dollars can change hands through the rapid decisions investors make based on industry news. It's why they're willing to pay an average of $20,000 a year to use a Bloomberg terminal. The service, which makes up Bloomberg's core business, aggregates information from more than 1,000 news organizations and 90,000 websites, vital to clients looking for any kind of advantage for their investment strategies. Companies like New York-based startup Dataminr claim they can deliver market-moving information faster than the newswires by accessing and analyzing information available on social media.

    Yesterday the Dow Jones Industrial Average dropped 146 points within seconds after a hacked AP account falsely claimed the White House was under attack. While the market recovered quickly after the hoax was exposed, it showed just how many Wall Street traders were now relying on Twitter for trading information.

    Dataminr says it was able to alert its clients to the hoax a full two minutes ahead of major news. The startup has partnered with Twitter to access the entire "firehose" of nearly 400 million daily tweets on the platform. It implements event detection software that trawls the Twittersphere to identify hotspots of activity or emerging trends that match its client's watchlist of topics, industries or companies. In this way, it says it can provide useful information to governments or finance firms.

    The company was one of the first entities to report the news of Osama Bin Laden's death after spotting a tweet by a former Defense Department official. It only took 19 more tweets about the subject before the company sent an alert out to its clients, who received the news 20 minutes before it broke on traditional news sites. (Dataminr representatives did not respond to request for comment.)

    Access to real-time news and data from a massive network with few geographic restrictions is crucial in a practice where an edge is gained by nanoseconds. For example, a fire at an oil refinery could briefly spike the price of gasoline. While it could take newswires several minutes to report the story, tweets from the actual incident could alert investors in real time and allow them to make a profitable trade.

    In addition to speed, Twitter analytics promise to provide companies with so-called sentiment analysis -- a measure of how people feel. This can be done in real time, gauging not just the broad market "mood" of consumers, but also people's reactions to marketing campaigns, product introductions, and personal endorsements. In 2010, Indiana University informatics professor Johan Bollen published a study that found social sentiment on Twitter could predict swings in the Dow Jones Industrial Average with a startling 87% accuracy. "It's definitely a very powerful environment, you have people acting as human sensors," says Bollen. "There is such a thing as the wisdom of the crowd."