A New Era of AI Enabled Web

Code: AiCMSWorld001

A New Era of AI Enabled Web

Hello Everyone,

           This post is not just a concept but through it, we are forwarding some serious questions to the current web industry with some sort of answers as well. We have evolved a lot since our starting, but I'm going to stick with the web and online business only. Today, the market is ripe with Inbound Marketing, Big Data, Big Data Analysis and a more such concepts. As we get huge amount of data every second, it is analyzed and our digital presence is designed based on the collected data.

          We are about to introduce a new era in the Web Technology, but first lets get some things straight.

  Are we on the right path?

          To this day, the digital presence of a user is designed as per the analysis of the received data. But is it the right way? I doubt that.

The norm is to design A/B testing, landing pages and several other things to give the best experience to the visitors. But in my view, we are judging more then 7 billion people with a few types of landing pages (if we expect every one to come on site ). That is similar to defining your future of every one using the Twelve Zodiac Signs, which is not a reliable process. We know that every person is different, and moreover, every person is different at every given time. Their needs are different at different moments.

          Web Giants, like Google or Amazon, are doing much better in this field and are also working hard on better analysis to provide you with the best possible experience.

 Content is the King?

           Of course it is. But it only leaves a big impact if it is in a proper sequence of What, When and Where.

          Google gives hundreds of results based on any search query that you make. Do you consider every result listed to be fit for you? No, you don't. And why is that?  so, not going in the reasons we put the future first.

  Can it be a thing of the future?

           Google kept the forwarding of results good. But those results, sometimes, do not give the "Wow" effect to the visitor. 

          Lets suppose that you are running a web designing company and provide all the features that a person is looking for. But what if a website itself can provide all features which that person is looking for on the top and the rest of the things below? Or perhaps, even a better scenario, where it can be decided whether a visitor is a designer looking for some spot requirement of SEO, or an SEO expert looking right now for any specific topic in the designing on his site.

          Furthermore, what if a visitor is more interested in a dealership rather than purchasing and is looking for how you deal with your dealers (in case you own any industrial website).

          It could be that you have a separate page for that but, what if the whole site itself represents what a visitor wants to see?

          If colourful websites attract a particular visitor, then on visiting your website, they might find that it looks dull and boring besides all your impressive content, or that it is not of their particular taste. In such a case, after skimming through a few pages, they might exit.

   Any Existing Answers?

          Yes, you've got a lot of analytical tools, heat maps, Visiting Path analysis tools etcetera. But I think they are not that much sufficient. Lets suppose that if your site is in red, you get 10% conversion and if it is in blue, you get 5% conversion. What would you do? Obviously, you will want to turn this 10% into 15%. But leaving 5%?

          Big corporates are analyzing quite well, but how many SMEs are doing analysis of tools which they have included in their website and making changes based on it? Not every one can hire a Data Mining Engineer.

          Even if somehow you got traffic, you will have to make a person retained and then convert his visit into a goal.

 Our Approach:

          That's what we are doing at XAVOC Technocrats (xEpan.org). We have designed a CMS that can learn. Infact its a totaly new algorithms set, we put into a CMS. It is  still in beta phase, but so far, not only does it provide a good analytical result but also learns itself and provides the best content to the visitors. 

          We have designed Intelligent Blocks on the website rather than simple content. You can define various dimensions of visitors, like for a web designing company, these can be:  
  • Designing Oriented Customers
    • They always want to know what you have made in the past, what your best designs are, etc.
  • SEO Oriented Customer
    • What type of SEOs you do? Do you deal in PPC or do you do Directory Submissions, etc.
  • Technical Customers
    • What platform you will use? Will it be jQuery enabled, will the site, which you will make, be responsive, etc.
          You might think at this point that you can have separate pages for each type but that's a problem according to us. Every body might have that but What if a person is a mix of all those fields and wants to hear all the good things (as per them)?

           For example, what kind of slides should be included in the slide-show? (Must be focused more towards the visitor's mindset.)

          So, you design the website with the intelligent blocks each with various dimensions.  

          Now, the system starts giving them to various visitors in a  random way or one by one method and starts recording events like it does in an analytical system. But the interesting part is that it learns what is converting your visitor into a customer and starts increasing the weights of the concerned Blocks and dimensions based on the visitor's visiting pattern that includes not only the search keywords and site references but also the country and its time of visit, etc.

          Still, it manages to keep itself random for better learning. You can teach it a few tips in the beginning to get a better start.

           The system learns by itself that a visitor from a particular blog, where you mentioned your product, was a discussion on quality or quantity or a rate-fight and provides the content to the visitor in the same fashion. Now that's a 'Wow' effect.

           We want our customers to achieve the "That's what I was looking for!" satisfactory statement. This system tries not to lose any customer just because of the fact that what they wanted to see was present on your site but you could not solicit their attention to it.
How this was achieved?

          We had started with a simple probability rule, with weighting the content based on past experiences of the system. It did worked like a charm but not the way we had planned. We wanted a Pro Sales Man working as a website, dealing with every visitor in its own fashion and learning its experiences, and on the next visitors visit, cloning itself with the newly learnt facts.

          Therefore, we moved to Neural Networks. Somehow, we designed
 new techniques
for better working/learning (we are about to provide them as open source as well), but a number of simple true and false conditions are no more precise than a ton of zodiac signs.

          We needed a more human approach. We needed a live, digital but still human-like salesman behind the veil of a website. So we moved to Fuzzy Logic. In combination of Neural Networks we are designing a technique that can create rules based on various factors including a Current-Path-of-a-visitor-so-far on the website. As the person moves ahead, the system gets more information about a person's behaviour and updates the upcoming content itself in a better way. It also learns that whether what was served by it did well in converting a visitor into a customer or not.

It is tricky and complex, yet fun! And you must read our next two posts in the same blog to get the complete picture.

   What's the difference between Neural Networks, Fuzzy Logic and our approach?

           Usually, pattern recognition works to predict the output of a sequence. At most, we can get missing values, but what we tried is to get the next best content selection by analyzing the visitor's behaviour that can convert that visitor into your customer, all the while learning through the visitor's behaviour.

 Do we have a firm ground?

          Yes, we have. We are about to launch a new version of our CMS that includes all this. You can subscribe to it at http://xepan.org to get more information about it.

          To know more about xEpan, CMS & AI, please refer to few more blogs pertaining to xEpan.
Date: 2014-10-10