检测邮寄地址相似度
我们遇到了会员试图注册多个帐户以利用推荐奖励和免费试用奖励的问题。我们需要一个邮寄地址并根据 USPS 对其进行验证，但发现这仍然无法提供足够的保护，并且用户仍然很容易更改该地址上的元素（例如“第 2 行”）并使其仍然有效。
检测其他用户是否已使用某个地址注册的最佳方法是什么？
地址变体示例：
59 Grant Ave 2 楼
格兰特大道 59 号 Suite 2
任何类似于精确比较的问题是街道地址可能在不同的公寓/办公室等中有多个租户。
我们正在使用 Rails 3、MySQL、Memcache、Sphinx
We have problems with members trying to sign up multiple accounts to take advantage of referral rewards and free trial incentives. We require a mailing address and validate it against the USPS but have found this still doesn't provide enough protection and it's still too easy for a user to change an element on there address like "line 2" and have it still be valid.
What is the best away to detect if a different user has already signed up with an address?
Example of address variations:
59 Grant Ave
Floor 2
59 Grant Ave
Suite 2
The problem with anything resembling exact comparisons is that the a street address may have multiple tenants in different apartments / offices etc.
We are using Rails 3, MySQL, Memcache, Sphinx
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（2）
我们面临着类似的问题，尽管是在不同的背景下。您无能为力，我们基本上编写了一种方法，尝试根据标准化电话号码、标准化街道地址、Leveshtein 距离、对地址进行地理编码时的物理距离等。找出某个东西是否真的重复仍然需要人工，但至少可以显着减少可能的候选者数量。
We face a similar problem, although in a different context. There's not that much you can do, we basically wrote a method that tries to identify possible duplicates based on normalized phone number, normalized street address, Leveshtein distance of name and address, physical distance when geo-coding the address etc. Finding out if something really is a duplicate still needs a human, but at least you can reduce the number of possible candidates significantly.
它需要智能才能可靠地完成，因此您可以忘记算法。我认为您能做的最好的事情就是报告与已提交的地址相似的地址，然后将其留给某人来决定。
It requires intelligence to do it reliably, so you can forget about an algorithm. I think that the best you could do is to report addresses which are similar to ones already submitted and then leave it to a person to decide.