English Language AI NLU Enablement (ELAINE)
A Symbolic Artificial Intelligence System for Natural Language
Qualitative Analysis powered by English Language AI NLU Enablement
Artificial Intelligence (AI) that is derived from the principles of symbolic logic is called "Symbolic AI". Symbolic logic is a derivative of formal logic such as Propositional logic and First-order logic. Symbolic AI is different from AI derived from statistics. Machine Learning or other variation of Deep Learning matches problems with previously known problems by means of statistical analysis, whereas Symbolic AI is based on symbolized reasoning to discover scenarios and solves problems by formulating rules.
Applications of Symbolic AI from the past require pre-defined rules. Example of these applications can be found among Expert Systems.
Modern day Symbolic AI discovers new rules with symbolized reasoning autonomously without human in the loop. In the process, it can learn on its own by accretion. Newer generations of Symbolic AI maintain an accretive knowledge base that enables it to resolve new and unknown problems.
English Language AI NLU Enablement (ELAINE) is an implementation of Symbolic AI for Natural Language applications developed by SiteFocus Inc. It is capable of detecting fallacy, contradiction and consistency found in Natural Language documents. Using symbolized reasoning, ELAINE transforms knowledge it encounters into rules automatically, thereby, eliminating the need for human experts to enter the rules. The following is an example of how it works:
From the FOMC meeting notes, Chairman Powell said:
"... people believe that inflation will come back to the target or around the target so it does not go down as much inflation - does not go down as much in a downturn and does not go up as much when the you know, when the economy strong"
ELAINE would process the above into sentential logic similar to the following Venn Diagram:
How to Read the Report
Business intelligence, emails, enterprise communications, social networks and many other unstructured textual assets are essential resources for strategic decision-making. These documents are complex and convey a lot of information. Getting insights can be a challenge. Like an analyst, ELAINE analyzes textual data for what?s important and why. Its findings consist of relevant subjects of importance (the what) and supporting context (the why).
ELAINE builds a report - a hierarchical knowledge graph - that works like an interactive Table of Contents(like a book!) ordered by relevance. Navigate the subjects and high level abstracts to get the Big Picture, identify Points of Interest. Documents & contexts may appear more than once. A document - even within the context of a paragraph - may cover several subjects from different Points-of-View. ELAINE uses Multi-Aspect NLU to capture that spectrum and map the entity-relationships for deeper, strategic insight discovery. Get
- Knowledge Graph - a link to a separate report on Knowledge Graph enable direct navigation on knowledge graphs related to each and all top level subjects.
- Relevancy Ranking - whenever possible, a link to Complex Event Analysis Report is include at the beginning of the Context Discriminant Report.
- Get the Big Picture by reviewing the high level topics! Click to get a detailed breakdown of high level abstracts.
- Discover Points of Interest by navigating high level abstractions that relate to your objective.
- Get Details & Validation with supporting context from the original source, auto indexed under each sub-topic.
- View text being analyzed - click on the hyperlink that appears before the excerpt (Reae more). This will open a new window with a microanalysis of the original source. To jump to the proximity of the text referenced in the excerpt, click on "Go to semantic neighborhood context". This is useful when the text analyzed is from a large document.
- View Original Source Document - is a link located at the bottom of the "Text Being Analyzed". Depending on how the document was ingested, there may be 1-2 links.
- Source - refers to the webpage for the source document