As writers, we all know what semantics is. But there are some interesting intersections between fractals, semantics, computer programming and the science of information retrieval. These topics have, of course, been studied in depth by logicians and AI theorists. However, as Web publishing is becoming increasingly more important in today's writers market, an understanding of search engine semantics has very practical applications for us as professional writers.

Fractal semantics pertains to repeating, predictable patterns found within the way we derive meaning from ordinary language. Although meaning-assignment may seem arbitrary (e.g., we could have just as easily placed verbs before nouns in simple sentences) there appears to be a predictable set of rules - a definable logic - to the way meaning is derived or assigned.

So what does fractal semantics have to do with search engines? Well, I don’t have a full understanding yet, but it appears that search engines must recognize semantic rules in the midst of the seemingly random and chaotic mass of text information existing in cyberspace. They must do this in order to deliver query results that are relevant to the searcher’s meaning-intentions.

For example, take the phrase “Kentucky Fried Chicken.” To the ordinary person, that phrase refers to the popular fast food chain. However, search engines don’t “think” the way ordinary humans do. A search engine has difficulty determining whether the phrase means “Kentucky Fried Chicken” (the fast-food chain) or “fried chicken (in) Kentucky.” As a result, an end-user searching for “fried chicken in Kentucky” will receive search results for “Kentucky Fried Chicken.”

Colloquialisms, dialect and idiomatic peculiarites are also particularly difficult for SQLs (Structure Query Languages) to interpret. (For instance, in the South, folk use the phrase "car wreck"; in the North, "car accident." Both mean the same thing, but try getting a search engine to understand that!) But all this is slowly and inexorably changing. Currently search engine algorithm developers are prograaming search engines to “understand” language in ways that are closer to the way ordinary humans actually understand language.

The Kentucky Friend Chicken example is just one among many which highlights the complex semantic structure of ordinary language. As SEs are beginning to “think” more like humans, they are getting better at looking at important factors such as word-order, regional influence, prepositional placement and co-occurence of words.

In short, SEs are beginning to recognize mean in much in the same way that ordinary humans do. The thought is that, if search engines can process language similarly to the way we do, then the science of information retrieval will change dramatically. Such important research may even change or augment the current standards of composition which contemporary writers utilize.

Eric Bryant, UT Alum '98
Chief Executive Officer
Gnosis Arts Writing Services