ELAINE Symbolic AI offers community tool to cure “TL;DR” Syndrome
The idea of applying Natural Language tool to advance human intelligence is not new. Examples can be found among popular search engines and chat-bots. These applications generally require Machine Learning ahead of lengthy preparations with “human in the loop” training datasets. These pre-requisites are cost intensive in terms of time, labor, infrastructure and skill.
A few years back, we were looking for a solution to cure the “Too Long; Didn’t Read” (TL;DR) syndrome. TL;DR is a common problem in today’s information overloaded world. We started researching for an effective solution to eliminate the aforementioned pre-requisites. Our objective is to enable a user to upload textual documents to a natural language AI engine to read and analyze over a WEB session, and get back results in real-time. By fulfilling the goal of “Tell me something I don’t know”, the TL;DR syndrome is resolved.
The project was code named ELAINE, an acronym that stands for “English Language AI NLU Enablement”.
ELAINE is a derivative of Symbolic Logic. Symbolic Logic is a philosophy that uses symbols to represent propositions. In a textual document, these symbols are keywords. ELAINE uses heuristics to discover related keywords. High Level Abstraction (HLA) formed from keywords identifies unique semantics.
Accessing document content by semantics has the advantage of direct access to semantics of interest to the audience.
Fast forward to today, we have successfully launched ELAINE Community Service. Offered under a public license as a complimentary service, it is a tool that anyone can use without registration. Content processed by ELAINE are temporary stored only for the duration of analysis while privacy and anonymity are strictly enforced. While the community version of ELAINE is limited by the processing time, it can handle documents of substantial size in the neighbor of 100 pages of text.
Elaine analyzes textual documents independent of problem domain. It handles documents from all categories, i.e., medical, transcript, research etc. I am planning to post a series of discussions on different use cases in coming weeks.
Credit: Data Science Central By: Sing Koo
