Smarts: Wolfram's World of Knowledge Engines

Wow! Having toiled on my own start-up for the past 3 years in my hometown nestled in Upstate, NY, I've taken little time to appreciate other worldly developments of such potential magnitude as Wolfram Alpha. Ironically, Rochester has a relative abundance of academics and practitioners innovating in the related field of computational mathematics among others. Until recently, I had yet to tap into that community but am confident I will continue to follow related developments as I am thoroughly intrigued -- and you should too because it will undoubtedly have a significant impact on our lives.


It was only until about 6 months ago that I learned of systems like WebWork which attempted to deliver automatically individualized math-based homework problems over the web. The system, though limited, provided students with real-time feedback as to the correctness of their answers, encouraging rapid problem-solving iterations to help students more efficiently understand the problem logic and collect statistics for instructors to develop more customized lesson plans. The system also discouraged cheating by individualizing, or creating comparable variations of, problems. 


This system was originally brought to my attention by a frustrated yet aspiring entrepreneur. "Don't people get it?" he exclaimed, prefacing his disappointment with the project not taking off and speculations of the broader future. He went on to explain that while the world is complex and there are many ways to identify and solve a problem, they are in fact computable. And his claim is reinforced by cellular automata, which are closely related to the scientific and philosophical underpinnings of Wolfram Alpha. In lay man's terms, cellular automata "proves that anything that may be computed — and potentially anything that may exist in nature — can be generated from very simple building blocks and rules that interact locally with one another." He convinced me rather quickly that because everything can be mathematically reduced and related, that applications such as WebWork can be applied to all learning not just math courses. Although my friend was more so referencing individualized learning, it seems highly feasible for us to apply this system and others to achieve the following:

• semantic results by contextually-recognizing search queries;
• semantic recommendations by narrowing on strong correlations between individual profiles and activities; and 
• semantic predictions based on our collective historic and present knowledge.

Granted, I am over-simplifying in part due to my immaturity on the subject and in part to make the topic more palatable for others as well. But I will definitely revisit the topic as more information becomes present and I refine my thoughts. As a side note, the only competitive solution to Wolfram that I could find was True Knowledge, which does a rather good job for basic questions that you might type into Google. In any case, hopefully Wolfram Alpha lives up to the hype and I'll try to cover it's release in May 2009. You can learn more about Mr. Wolfram's research and expect to see the new computational knkowledge engine soon!