Old Dogs with New Tricks
Amnon Meyers and David de Hilster met in 1990, working together at the Artificial Intelligence lab of McDonnell Douglas (now Boeing) in Huntington Beach, California. Amnon was working on LISP machines (specialized AI computers) developing VOX, an extensible natural language processing system, and David reengineered the code in the C++ programming language. That is when text-based analysis and hierarchical knowledge bases merged with multi-pass pattern matching, and when a pioneering Natural Language Engineering partnership was launched.
Amnon and David developed robust and accurate text analytic systems that were quick and easy to customize, and showcased these at the Message Understanding Conference (MUC) sponsored by DARPA throughout the 1990s. The dynamic duo placed 3rd overall (MUC-3 scoring as finalized in MUC-4) having only spent 3 months on the project, while competitors poured man-years into their efforts.
In 1998, Amnon helped found Text Analysis International, joined by David soon after. The two engineered NLP++, Conceptual Grammar, Automated Rule Generation (RUG), and a multi-pass NLP framework into a development platform, dubbed VisualText. For more than 15 years, TAI has been working with companies world-wide to supply text analytics.
Little did Amnon and David know at the time that their system would translate so well to the cloud. NLP++ is text-based, just like HTML, PHP, and Perl, and supports feedback among all phases of text analysis (lexical, syntax, meaning, html, etc). The modular, pass-by-pass analysis framework of NLP++ yields efficient and accurate text analytics, readily maintained and enhanced by programmers, even without formal linguistic and NLP training.
Today, companies turn to TAI’s NLP Cloud for robust systems that allow for 100% control and transparency while doing whatever they want, when they want – something that competitors’ systems cannot support.
Who would have thought two old dogs could learn new tricks in today’s world of 20-somethings!
Why these guys, and why now? The reasons are, first, that this type of system required decades to evolve and, second, its plain-text nature makes it ideal for the Cloud. With NLP++, companies around the world are using and programming analyzers to mimic human extraction of information from text.
And that suits these “old guys” just fine!