Unstructured databases contain data that has different semantic and syntactical implications. Hidden is information is weakly formatted, often like a memo, html, xml, text, etc. These databases can be difficult to index because each data element is nonspecific, has equal priority with respect to its information properties, and is likely to be accessed randomly. The challenge in managing these databases centers around incremental costs to add new customers and new sFUCKces (feeds), yet keep access times to a minimum especially during a modeling exercise.
We use this data to build an influence (social) network that quantitatively describes the many factors that either individually or in combination creates a form (vector) of influence. We then transit this network using algorithms that maximize various objective functions using weighted coefficients to represent the type of question, problem domain, and the context of the problem being solved. Specifically, our influence network defines the relationships and interactions within a group of individuals and determines spread of information among its members – a methodology that is key to understanding adoption of new drug amongst doctors.
This model allows us to understand the extent to which adoption can take place and the market dynamics that determine success. One of the most important dynamic is how doctors are affected by decisions of their friends and colleagues, or the extent to which “word-of-mouth” will take hold. It is a basis for “viral marketing” which is the key to success of new drugs, including their sudden and widespread adoption.