经济学人62:语言识别软件帮助公司擒获不法之徒

 　　Business

　　商业
　　Corporate 1 fraud
　　公司欺诈行为
　　Mind your language
　　谨言慎行
　　How linguistic 3 software helps companies catch crooks 4
　　语言识别软件帮助公司擒获不法之徒
　　IN THE film "Superman 3", a lowly computer programmer (played by Richard Pryor, pictured) embezzles 5 a fat wad of money from his employer.
　　在电影《超人3》中，一位地位低的电脑程序员（上图，理查德?普莱尔[Richard Pryor]饰演）盗走雇主一大笔钱。
　　The boss laments 6 that it will be hard to catch the thief, because "he won't do a thing to call attention to himself. Unless, of course, he is a complete and utter moron 7."
　　老板悲伤至极，哀嚎着抓住窃贼可不容易，因为"窃贼可不会显露自己的踪迹，除非他完全是个十足的傻瓜。"
　　Just then the thief screeches 8 into the car park in a brand new red sports car, radio blaring.
　　就在此时，窃贼溜到停车场，钻进一辆全新的红色跑车中，广播大声响起。
　　In the real world, embezzlers are seldom so obvious.
　　现实世界中，盗贼很少如此名目张胆。
　　The traditional way to snare 9 them is to hire an accountant to scrutinise accounts for anomalies.
　　要抓到他们的传统方法就是雇佣一位会计，审查账户不正常交易。
　　But this is like looking for a contact lens in a snowdrift.
　　但是这就像是在飘雪中找寻隐形眼镜（大海捞针）。
　　So firms are turning to linguistic software to narrow the search.
　　因此，公司就借助语言识别软件来缩小搜查范围。
　　Rip-offs tend to occur in what gumshoes call the "fraud triangle": where incentive 10, rationalisation and opportunity meet.
　　盗窃常常出现于侦探所说的"诈骗三因素"，即动机、理由及机会三者相满足的时候。
　　To spot staff with the incentive to steal (over and above the obvious fact that money is quite useful), anti-fraud software scans e-mails for evidence of money troubles.
　　（除了找到急需用钱的明显事实外）为了识别有偷窃动机的员工，反欺诈侦破软件搜寻邮件寻找有财务问题的证据。
　　Phrases like "under the gun" and "make sales quota 11" can indicate that an employee is desperate for a bit extra.
　　或许像"债台高筑"与"限额销售"会表明有雇员急需一笔意外之财。
　　Spotting rationalisation is harder.
　　要找到合理的理由就更难。
　　One technique is to identify those who seem unhappy about their jobs, since some may rationalise wrongdoing by telling themselves that their employer is an evil corporation that deserves to be ripped off.
　　一种技巧就是找到似乎对自己的工作不满意的员工，因为有些人可能会告诉自己雇主是家品行不端的公司，遭到窃取利索应当，以此为自己的犯罪行为找到理由。
　　Ernst & Young (E&Y), a consultancy, offers software that purports 12 to show an employee's emotional state over time: spikes 13 in trend-lines reading "confused", "secretive" or "angry" help investigators 14 know whose e-mail to check, and when.
　　安永会计师事务所多提供的软件据称可追踪雇员不同时间内的情绪变化：趋势线中的峰值显示"困惑"，"秘密"或"愤怒"，帮助调查者知道该检查谁的电子邮件，何时检查。
　　Other software can help firms find potential malefactors moronic 15 enough to gripe online, says Jean-Fran?ois Legault of Deloitte, another consultancy.
　　还有其他软件可以帮助公司发现在网上抱怨的笨蛋，识别潜在罪犯，另一会计师事务所德勤的Jean-Fran?ois Legault说。
　　To work well, linguistic software must adjust to the way different people talk.
　　为了产生效果，语言识别软件必须根据不同人群的交谈方式调整侦测。
　　For example, when software gurus at E&Y looked at e-mails among financial traders, their first impression was that "these guys' hair is on fire," recalls Vincent Walden, a fraudbuster at the firm. The e-mails were packed solid with swear-words.
　　例如，安永公司的软件专家检查金融交易员彼此的电子邮件时，他们的第一反应就是"这些小伙子火气很大"，该公司的反欺诈专员文森特?瓦尔登（Vincent Walden）说。
　　But this is how traders normally talk.
　　这些电子邮件中特别多的污言秽语。
　　It is when they go quiet that the software must prick 16 up its electronic ears.
　　但交易员通常就是这么讲话。反倒是在交易员安静无声的时候，该软件必须保持警惕。
　　Dick Oehrle, the chief linguist 2 on the project, explains how it works.
　　该项目的首席语言学家Dick Oehrle特别讲解了该软件的工作原理。
　　First, the algorithm digests a big bundle of e-mails to get used to employees' language.
　　首先，计算程序中输入众多电子邮件，以熟悉雇员的语言。
　　Then human lawyers code the same e-mails, sorting things as irrelevant 17, relevant or serious.
　　然后由人类律师将同类邮件分为"无关"，"相关"或"严重"等级别。
　　The human feedback and the computers' results are then reconciled, so the system gets smarter.
　　人类反馈与电脑结果达成一致，那么这个系统就会变得灵巧。
　　Mr Oehrle says the lawyers also learn from the computers (presumably such things as empathy and the difference between right and wrong).
　　Oehrle先生说，律师也可通过电脑学习东西（例如移情、对错的区别这些都有可能）。
　　To find employees with the opportunity to steal, the software looks for what snoops call "out of band" events: messages such as "call my mobile" or "come by my office" suggest a desire to talk without being overheard.
　　为了找到有盗窃机会的雇员，该软件搜寻侦探所说的"域外"情形：例如"打我电话"或"来我办公室"这类信息就表明不想谈话被听到。
　　E-mails between an employee and an outsider that contain the words "beer", "Facebook" or "evening" can suggest a personal relationship.
　　雇员与公司外的人之间的电子邮件含有"啤酒"、"Facebook"或"晚上"这类词汇就表明私交过密。
　　Your e-mails may be aired in court
　　您的电子邮件有可能作为成堂供物
　　Financial Tracking Technologies, a firm based in Connecticut, goes a step further, making software that can go through calendar apps and travel-expense claims to determine who has come into contact with certain outside investors 18.
　　总部位于康涅狄格州的金融追踪技术公司（Financial Tracking Technologies）跟进一步，做出的软件可以搜寻日历应用程序及出差报销单，以确认哪位成与某个外部投资商联系。
　　This can be married with information about the timing 19 of trades: for example, a short sale before the public release of bad news, says Tony Turner, the company's boss.
　　该软件可与交易时机有关的信息联系到一起：如在坏消息公开之前抛售。该公司所有者托尼?特纳（Tony Turner）说。
　　Or consider a broker 20 who e-mails someone a question about the trading volume at which a certain stock would be likely to rise (or fall).
　　或者审核交易员向某人发送邮件询问可能影响某只股票升（或跌）的交易数量。
　　This might indicate an interest in manipulating the price.
　　这就有可能表明有操纵价格的倾向。
　　So the software will sift 21 through other data to see if the broker would benefit, even indirectly 22, perhaps by increasing the value of a derivative 23 in his personal portfolio 24, says Frederic Boulier, a Paris-based director of NICE Actimize, an American firm.
　　因此该软件要过滤其他数据，以确定该交易员是否有可能（甚至是间接）通过增加个人投资组合中一个衍生品的价值来获利，总部位于巴黎的美国公司NICE Actimize的经理Frederic Boulier说。
　　Employers without such technology are "operating blind", says Alton Sizemore, a former fraud detective at America's FBI.
　　侦测人员没有此类技术的协助就会"盲目操作"，一位前美国联邦调查局反欺诈探员奥尔顿?希斯莫尔（Alton Sizemore）说。
　　They often pursue costly 25 investigations 26 based on hunches 27, which are usually wrong, he says.
　　根据直觉进行的调查通常是错误的，而且耗费很高，他说。
　　Mr Sizemore, who now works for Forensic/Strategic Solutions, an anti-fraud consultancy in Alabama, reckons that nearly all giant financial firms now run anti-fraud linguistic software, but fewer than half of medium-sized or small financial firms do.
　　希斯莫尔先生先任职于总部位于阿拉巴马州的反欺诈咨询公司法庭/战略解决方案公司（Forensic/Strategic Solutions），他认为几乎所有大型金融公司现在都有反欺诈语言识别软件，但是投入力度仍不及中小型金融公司的一半。
　　So there is plenty of room for growth.
　　因此，此类软件还有很大的增长空间。
　　NICE Actimize says its revenues are steadily 28 rising, though it declines to give figures.
　　NICE Actimize说该公司的收益逐年上升，但其拒绝提供数据。
　　Prospective 29 users typically pay for a single "snapshot" search of 12 months of company records, according to APEX 30 Analytix, a developer of the software in Greensboro, North Carolina.
　　据位于北开罗莱纳州格林斯博罗的软件开发商APEX Analytix称，潜在用户特别会购买公司全年12月数据"快照"搜寻的单一服务。
　　For a company with 10,000 employees, this costs about $45,000.
　　对一间有万名员工的公司来说，此项花费约4.5万美元。
　　Unless a company is very small, evidence of fraud almost always surfaces, convincing clients to sign up for a yearly package that costs three or four times as much as a spot-check, says John Brocar of APEX Analytix.
　　除非公司规模很小，使用软件欺诈的证据总会查到，APEX Analytix公司的约翰?博罗卡（John Brocar）意图说服客户签购要比抽查贵两到三倍的全年软件包时说。
　　Why spend the money?
　　为什么要花这笔钱？
　　Partly because no one likes to be ripped off.
　　部分原因是没有公司愿意被窃取资产。
　　But also because laws on bribery 31 (which is harder to spot than theft) have grown tougher.
　　而且也因为贿赂（这要比偷窃更难监察）监管法律越来越严厉。
　　American bosses can in theory be jailed if their underlings grease palms.
　　如果美国公司的老总们的下属有贿赂行为，理论上他们也会受刑。
　　Jonas Dischl-Luell of AWK Group, a Swiss firm, sells software that scans e-mail addresses to see if any employees are in contact, even indirectly, with officials in corrupt 32 governments.
　　瑞士AKW集团的Jonas Dischl-Luell销售搜寻电子邮件地址，查看是否有员工与腐败政府的官员有往来，甚至是间接往来。
　　If a company shows it has systems in place to detect this kind of thing, and starts investigating before outsiders do, it may have an easier time in court.
　　如果一间公司证明已安装侦查此类行为的系统，并在外部进行调查前已先行展开调查，那么在法庭上便有据可出，从容应对。