Gary G. Hendrix
It is one thing to write research papers or build laboratory software addressing some tough natural-language phenomenon. It is quite another to build natural language capabilities into a product that has to support real people doing real work. This presentation is about the latter. A hard look is taken both at the kinds of applications today’s NL technology is really able to support and at some of the not-so-technical problems that can arise when NL technology is employed. The capabilities of current NL products are evaluated, and insights gained from basic research are reviewed as a guide to forecasting future engineering developments and to identifying major technical obstacles blocking the advancement of applications.