XRumer 8.0: the birth of artificial intelligence Most users know that the scheme works with virtually any resource - it is "Registration -> Input -> posting." Sometimes the possibility of posting without registration. But the essence is the same everywhere. If we represent a schematic, it looks like this: Websites that work on this scheme - the hundreds of millions . It is important to cover a greater share of this huge resource, and therefore need to carry out a set of measures for improvement. According to our observations, many relatively simple resources, XRumer-in to send a message to prevent only 1-2 Obstacles or unknown field in the registration form, Unknown or graphic or text captcha or some JavaScript, or program does not find a link where you can register / log / post it. Such complexity - associated with rigid, fixed schemes of XRumer-a. Increased flexibility, self-learning, including with the assistance of all users of the program - will give the opportunity to cover a large part of the Internet for posting. Therefore, XRumer 8.0 is designed to be six new destinations in the sum solve this problem: 1. What is the essence of artificial intelligence in this system? main feature - a new improved self-fields, forms, and self-featured unknown links / posts. The essence of the process that the program can draw conclusions on the success / failure produced steps: if the attempt fails repeatedly received "skill" will be fixed in the brain of the program, if the attempt has never led to positive results, then automatically decides about that this "skill" needed to weed out. The stages of this process are described below. 2. Stages of learning a new algorithm here will require a new base - not 10-50 million resources, and more than a billion, after binding to any specific engines should not be there. 1.) At the first stage of a simple crawler traverses the list of domains and sabdomenov, checking is there any resource anywhere on the at least one form to be sent. Most likely, this procedure will be split into several dozen servers. The process will take place rather quickly, as GET-requests are processed immediately, especially since it is implemented based on Linux - in the end, this drop-out could take no more than a month. As a result, we have all the Internet sites on which there is generally some form to send. 2.) The next step - learning itself, it will be launched on the fact of availability of training algorithms. The base will be split into parts for the 10-50 million hits, after which the parts are distributed to all clients wishing to participate in the process. 3.) Each member of the learning process is going to a special database of "new knowledge of the program" - a chain of logical connections that all participants must send to us (perhaps the process will be automated, so as not to bother clients). The process of purely voluntary: those who do not have time to waste, certainly should not be involved. But those who take part - will have the opportunity to test the first XRumer 8.0 4.) The collected information is filtered, thoroughly tested on a variety of resources. It is very important to work out all the details - from all the correct action depends on what kind of share will encompass a new software version 3. What do we give it? This approach can increase the piercing - from 100 ... 300 000 resources (at a conservative estimate) to + tens of millions (according to optimistic estimates). In any case, you can say for sure: the effectiveness of XRumer 8.0 will be many times higher than that of XRumer 7.0. 4. How do other areas of the project? Of course, in parallel will develop plug-ins (it's possible that will add new ones - to work with the same YouTube, Twitter, Facebook), is directly dependent on the efficient operation of dop.kadrov. 5. Prospects of this mechanism when all steps in the automation of training will be the most polished, this procedure can be repeated ~ once a year / six months, adding "knowledge" of the program changes on the Internet. This will help maintain the relevance of a very long time. Of course, in addition to Auto-learning will continue and manual training is very specific resources, such as in the past month will be added new captures, tekstkapchi, engines. Just Artificial Intelligence will give an additional push in the development of the program. 6. As can be expected to yield XRumer 8.0? Approximate readiness training algorithms - June of this year. Accordingly, the release of XRumer 8.0 can be held in the autumn, in September / October. 7. Maybe it's a grand illusion? No, it's more than real plans for us this - the importance of a logical continuation of the project. In order for you to visually monitor the entire process will be opened separately under a private blog design - 2012, which will highlight the work to promote, with screenshots and videos. This blog design will be available from February. --- If you have any thoughts, suggestions, ideas - write all Comments.
