Shulou(Shulou.com)06/02 Report--

Internet automation programs have penetrated into all aspects of our lives! Zhengzhou Darnett has sorted out some knowledge points in the network and shared them as follows, hoping to help you understand the Internet!

At present, the Internet has penetrated into all aspects of our lives, but it is still only the projection of the real physical world encoded by bit information in the virtual cyberspace.

So as long as you customize the corresponding automated program to imitate human behavior, at the same time, because the machine is faster and tireless, it will be used to publish marketing messages in bulk in forums, websites, and app. Moreover, in the case of inadequate regulation, industries with higher profits tend to have a lower bottom line, and spam posted by automatic robots is often related to grey industries such as gambling, fraud, sex and so on. Some automated programs will also try to "hit the library" to steal user accounts and passwords, bringing huge security risks to the website.

Therefore, the CAPTCHA arises at the historic moment. As the same automated program, the purpose of QR code is to distinguish whether the user is a robot or a real person.

The most common CAPTCHA is automatically generated distorted text and patterns, although it can effectively identify a large number of automated programs, but it is not a good experience for human users. And with the development of machine learning, it becomes easier and easier to crack it.

Design pattern of ▲ CAPTCHA style

Google's CAPTCHA team has various innovative experiments, such as the creative use of CAPTCHA in the digitization of paper books. In addition, in addition to distorting the mainstream lines of text and pictures, Google's team also tried new ideas, using tracking users' click behavior to identify whether they were real people. Users only need to click the check box of "I am not a robot" to verify.

In the latest version of Google CAPTCHA reCAPTCHA v3, you don't even have to do anything, and the system is quietly checking whether the current user is a robot. Technology is making the "CAPTCHA" more and more invisible, and humans no longer have to do the "reverse Turing test" in order to prove their identity. however, this progress has also brought a lot of new problems.

Primary CAPTCHA CAPTCHA: crooked text

In 2000, Louis Fengan (Luis von Ahn), who graduated from Duke University's Department of Mathematics and came to Carnegie Mellon University for a doctorate in computer science, together with his tutor, put forward the concept of CAPTCHA, which is a fully automatic public Turing test that distinguishes computers from human beings (English: Completely Automated Public Turing test to tell Computers and Humans Apart, referred to as CAPTCHA).

The Turing test was proposed by Alan Turing, a computer pioneer and "father of artificial intelligence", to pass the Turing test on the basis that a computer can talk to humans without being identified as a robot. CAPTCHA is also a kind of Turing test, but its purpose is not to create AI, but only to identify real human users.

One of the most common CAPTCHA codes is the distorted text generated by the algorithm to prevent it from being automatically recognized by the optical character recognition program (OCR).

There are some ways to add a curve to a letter or to stack different letters together, and there are ways to add a complex background.

There is also a picture verification code, which requires the user to identify the object of the picture and drag the missing parts to the correct position and puzzle.

But regardless of the form, these CAPTCHAs have a common principle: it is very difficult for computers to make it easy for humans to identify. Some researchers believe that in order to avoid the loss of users due to the difficulty of CAPTCHA, human users are usually required to pass the test in less than 30 seconds and the user pass rate is more than 90%.

There is another point that is not known by ordinary people. CAPTCHA is called a "Turing test", so it has the original intention of promoting the development of artificial intelligence at the beginning of its design.

According to the definition, the algorithm of CAPTCHA must be made public, so that the process of cracking CAPTCHA is to solve the corresponding artificial intelligence problems, such as image recognition, OCR with higher accuracy, etc., and the cracker does not have to work hard to deduce the algorithm through reverse engineering.

Digitize paper books by using CAPTCHA

At present, CAPTCHA has been widely used in major websites and app. Data show that 200 million CAPTCHAs have been used every day in just five years after the launch of this technology.

Soon, CAPTCHA inventors proposed a new project, reCAPTCHA, mainly to digitize paper books before the advent of the Internet. The idea is like this: the CAPTCHA system will show users two words, the first is to automatically generate distorted text normally, and the other is from a scanned version of a paper book, which is often difficult to be recognized by OCR programs because of its age or stains on the paper.

Therefore, when the user enters the CAPTCHA, as long as the first word is entered correctly, it can be identified as human, and the second word entered is only "voluntary labor". This is because the system will default that the second word input is correct, and the input result will only be compared with the input results of other users. If multiple users have the same answer, the word will be digitized.

You might think that such a word recognition doesn't make much difference compared with the huge books to be digitized, but at the beginning of its launch, reCAPTCHA could enter 30 million characters. In 2011, it completed all the digitization of the New York Times, an old newspaper that has been published since 1851 with a large amount of pure paper content.

In 2009, Google saw the value of the project and bought reCAPTCHA, which was also used by Facebook, Twitter, CNBC, etc. While helping the most popular websites resist the harassment of automated programs, scanned versions of ancient books in Google books that are difficult to identify automatically are also digitized with reCAPTCHA.

In addition, reCAPTCHA is also used to help machine learning systems improve the image recognition rate, which works in the same way as the digitization of ancient books, using house numbers and photos of cats and dogs that cannot be recognized by machines as verification codes for human recognition.

At the same time, users are actually tagging training sets for machine learning systems, so you may have taken credit for the artificial intelligence technology behind the powerful AlphaGo.

NoCAPTCHA: verification method that does not need to enter characters

After Google acquired reCAPTCHA, it improved it in the way of Google.

In 2014, Google launched a new CAPTCHA system, NoCAPTCHA reCAPTCHA. Although the name is a bit of a mouthful, it is still a CAPTCHA system. Its core is that you do not need to enter a CAPTCHA. Users only need to click on the "I am not a robot" check box, and Google can tell if you are really human.

ReCAPTCHA's slogan also goes from "stop spamming and read some books" (Stop Spam. Read Books), became the original purpose of CAPTCHA "simple for humans, difficult for robots" (Easy on Humans, Hard on Bots).

NoCAPTCHA tracks the behavior before, during, and after the user clicks on the verification box, such as the time spent on the web page, to determine whether it is human.

If you are misjudged as a robot, there is also a chance to "appeal", like image verification, to choose the right target from a pile of pictures.

Sites that use reCAPTCHA v3 put reCAPTCHA v3 code on every page of the site, not just on the login page. The reCAPTCHA system tracks all browsing behaviors of users for analysis.

In this way, Google can get almost all of the user's behavior. Google also confirmed that the hardware information used by the user, that is, the software on the device, would be sent back to the Google server, but said the results were "only used to analyze user behavior, not for personalized advertising recommendations." However, the fact that privacy has been mastered is here. Do you want to be verified more quickly or do you want privacy in exchange for quickness?

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.