Interview - Dr. Berkan, hakia Mastermind

We had an exclusive interview with Dr. Riza C. Berkan, CEO of hakia today. For those of you who don't know, hakia is building the Web's new “meaning based” search engine. The vision of the company is to improve search relevancy and interactivity to take web search beyond its current boundaries and into the future. We have covered hakia on a couple of occasions and have been loocking forward to talking with Dr. Berkan.  

Dr. Berkan is a nuclear scientist who specializes in artificial intelligence and fuzzy logic. He is also a notable authority on high level computing in natural language processing, author of the book “Fuzzy Systems Design Principles”, and author of numerous articles on artificial intelligence.

I have to tell you guys what a rare pleasure it was to talk with Dr. Berkan. Coming off the interview with Jimmy Wales of Wikia, I have been fortunate to be able to talk with two of the nicest and most innovative Web 2.0 people out there. Dr. Berkan personally led me on a guided tour of what hakia's capabilities are now, and will be before the end of the year. Dr, Berkan said something on the hakia site that I find very true: “Internet Search is the single most important phenomenon for global unification and progress of human civilization.”

After about 40 minutes of having all my questions answered without having asked them (a little like being with the Oracle in the Matrix), I just had to ask some questions, well what was left of them any way.

Profy: How or when did the vision for hakia begin Dr. Barkam?

Dr. Berkan: Well, Victor Raskin (Imminent professor-Purdue) and I worked together 16 years ago for the US Government in the nuclear field. We waited around for a while for someone to come up with a language based search engine, and when no one did, we decided to do it.

Profy: That is so funny, (laughs all around). How does QDEX differ from old indexing systems?

Dr. Berkan: Old indexing methods essentially used one index table, while QDEX dissects the data and stores it to gateways for any possible queries one might ask. The information density in the QDEX system is much higher and therefore full semantic analysis can be undertaken more easily. The QDEX is distributed over a network of very fast servers that can individually access bits of data more rapidly than from a single source.

Profy: So this is like a Raid configuration for hard drives in a way?

Dr. Berkan: Well there is a slight similarity, but a better analogy would be comparing old indexing to a balloon, and thinking of QDEX as bubble wrap. A balloon fills up and expands as data is added, while bubble wrap maintains its cohesive nature but spreads the data into many smaller entities. Scaling active data sets makes better use of bandwidth, CPU usage, and storage capacity.

Profy: These are fascinating innovations; can you give a simple example of how this works?

Dr. Berkan: Sure, take a system like Google uses, and ask the math to decompose the sentence structure of a sentence with say 10 words. The result would be hundreds of queries resulting in no active data sets. Trying to break down a 10 word sentence with mathematics would produce billions of permutations. However, a person would break that same sentence down into perhaps 5 or 6 permutations. QDEX combined with the SemanticRank Algorithm bridges that huge gap and produces far fewer permutations.

Profy: This looks like a lot more than a search engine to me?

Dr. Berkan: Actually hakia is much more. Search is reduced now to people looking relevant queries for pages and a fairly limited range of data. People have real problems that search engines cannot address. They have health, money, relational and other personal issues that search does not address. Eventually, hakia will be able to address these real world issues with tangible, valuable and actionable results.

Profy: It would appear that SEO companies will have to hire a whole new job description in order to attempt to manipulate hakia as they have Google?

Dr. Berkan: (laugh) Yes, I am not currently aware of any method that might be employed in an effort to circumvent the relevancy of hakia. In essence, hakia results render the actual content that was asked for with no ambiguity as to the content. Any attempt to optimize a page's placement would effectively render the actual page.

Profy: So a junk company wanting top ranking for the search “War and Peace”, would have to write the book “War and Peace” to be ranked in the top 2?

Dr. Berkan: (laugh) Yes that is essentially it. They would have to turn their site into someone else's to be seen as relevant.

Profy: Thank you so much for your time and my lesson in Web 4.0 search Dr. Berkan!

Dr. Berkan: It is my pleasure, we are happy that Profy is so interested in what we are trying to achieve.

It is a wonderful thing to be able to glimpse the future. I feel terrible that I cannot adequately display here just how cutting edge and advanced hakia's work is. The list of people associated with this project and their openness speaks volumes about the character of this company.

There are a couple of things that we need to focus on in evaluating hakia at this time. First and foremost, hakia is still in Beta development. If you go there, please look at the examples and try to guide yourself through what is actually taking place there. This effort is a lot more than the neat organization of some facts. As an engineer myself, I grasp what this is going to be when complete but it is nearly impossible to represent in a blog form. What you are looking at in hakia is the way people will browse and search the net in the years to come. This is as unavoidable as the trend towards color TV from black and white 50 years ago. Whatever you are using to search the web folks, that will be a dinosaur in a little while, seriously.

Result of keyword search. Note: Question search and phrase search are possible and revealing!