Communication of thought has two layers – expressed and implied. With Sainapase’s Data Adaptation, expres-sion is addressed. Sainapse is capable of ingesting text in any format and normalize it for further processing. The next step is extracting the intent from text. In a Von Neumann world of if-then-else programming, intent is exactly what is expressed; there is no ambiguity. Human communication however, is far more unstructured and hence context variant. Even the shortest complete sentence in English, “I do”, has different meanings when said by a bride at church, or at office by the same lady in response to “Do you have your car with you today?” Humans have no trouble intuitively parsing contextual intent from words; machines find it near impossible. Sainapse converts text elements as vectors in extremely high dimensional non-linear space. Sainapse learns to do this over a sequence of steps:

  • Normalize inputs across heterogenous sources of data.
  • Expansion of dimensions through co-occurrence discovery.
  • Theme level discovery of vectors.
  • Index level distance function
  • Magnification of the distances to find sufficient distance between disparate concepts that represent “intent”.

These steps enable bucketing new queries to larger concept buckets. Drill down to sub-concepts within a concept bucket is achieved through representational learning.

Sainapse is built to work seamlessly with both unstruc-tured and structured data. A simple statement - “I’m locked out, please help!” - has very different implications based on context. A user trying to log into her laptop, versus when her new car’s electronics malfunctioned and the door is not opening at night, needs very different responses for the same apparent question. The ability to parse this intent is critical for Sainapse to provide the right answer to solve either of her issues.

Sainapse’s patent pending technology for intent extraction is key differentiator for delivering industry beating impact on time to solve and accuracy of response.

Talk to us; we will be thrilled to show you how seemingly esoteric technology components work seamlessly to solve your real world problems.