The Real-Time Experience-based Predictive Simulator (REPS) is a next generation, system that learns in real-time from multiple sources and provides up-to-the minute, situation-based predictions and prescriptive recommendations with explanations.
The Semantic Insights Research Assistant (SIRA) is a suite of applications that apply [and grow] domain knowledge to extract meaning from large natural language data sets in real-time.
3rd Wave AI supports; contextual adaptation, explanation of results, natural language understanding, incremental learning from experience, works with low volume data, learns and reponds in real time, predictive and prescriptive analytics; all with minimal supervision.
Natural language understanding (NLU) applies both world and domain knowledge along with natural language syntax and grammar to capture the various interpretations of what is happening within and across a corpus of documents. NLU is concerned with identifying how concepts and things are related to each other in time.
Autonomous, Training, Reasoning, Explaining, Changing Contexts, Changing Situations, New Situations, New Results, Missing Data, Wrong Data, Real-Time Learning, Real-Time Predictions and Real-Time Prescriptions; Reading, Understanding, Learning and Applying Natural Language Texts.
Computer technology has advanced to the point where real-time simultaneous recording and processing of multiple complex experiences is possible. To meet the needs of our dynamically changing world, statistical averaging of experience is no longer sufficient to adequately predict or prescribe what will happen next. Statistical methods most often fail when each new situation presents facts that are in some way unique. This is the real world. This is the world of natural language understanding. This is the world of 3rd Wave AI.
Digital Twin refers to a digital replica of a Physical Twin (e.g. external systems, processes and devices). A Digital Twin is used to simulate spatial and behavioral aspects of the Physical Twin under given conditions. Digital Twin construction and maintenance is typically a daunting challenge. With REPS, the Digital Twin is automatically created and maintained by watching the Physical Twin.
Given a set of experiences (i.e. situations followed in time by other situations):
A prediction is an ordered set of possible future situations, considering the current situation and previous known experiences. A prediction explanation lists the most relevant previous experiences for the given situation. The explanation also states how each relevant experience supports the prediction. In addition, the explanation may identify frequently occurring fact combinations that tend to precede facts in the predicted situation.
A prescription is an ordered set of changes to the given situation that if applied, would maximize a set of desired outcomes, based on the available Experiences and Desired Outcome priorities. A prescription explanation lists how the effects of each ordered set of changes to the given situation, contribute to Desired Outcome.
Natural langauge research starts with a natural language query (may be multiple sentences), reads a [large] corpus of documents (may be any source of text), finds semantically relevant content (may not resemble the original query text in any way) and renders the findings (as data or in a constructed natural language report with bibliography).
A fact expressed in Natural Language generally is represented by a concept cluster involving a relationship between 2 or more concepts. Typically, the relationship is represented by a verb, preposition, adjective, infinitive, or gerund. Further, a fact may also be a single concept like a noun phrase. A fact may also convey the understanding of "exists" or "does not exist" within this situation (which may be expressed or implied).
A single fact can be expressed in multiple different ways in natural language. Therefore, the conversion of natural language to a Fact must be able to reduce all the ways of saying the same thing in natural language (i.e. same meaning) to the same Fact.
Thus, a fact is a token representing a single concept cluster that may be expressed in many different ways (e.g. using different grammatical structure, synonyms, and specializations) in the Natural Language Text.
Third Wave AI
Natural Language Understanding