USAF HR Initiative Considers Business Rules Framework

A Business Rules Management System (BRMS) may be in the cards for the US Air Force -- at least for some Human Resources applications.  According to an Air Force Request for Information, sources are being invited to "investigate the viability of such a tool in the current Business and IT structure."  The Air Force clearly wants an off-the-shelf solution, not a custom, military-only beast.  They're looking at rules such as those involving policy, eligibility, and vocational speciality-specific requirements. The goal is to reduce application maintenance, facilitate "consistent business policy implementation across applications," and be usable by non-technical staff.  One other thing: the project being contemplated by the 554th Electronic Systems Group must work with IBM's WebSphere.

◦

USAF HR Initiative Considers Business Rules Framework

Feigenbaum on Domain-Specific Knowledge and Pattern Recognition

The June 2010 issue of Communications of the ACM features an interview with AI pioneer Edward A. Feigenbaum. In addition to providing his view the successes of DENDRAL and Meta-DENDRAL, he observes that "I believe that AI is mostly a qualitative science, not a quantitative science.  You are looking for places where heuristics and inexact knowledge can come into play" (p. 43).  He recaps his experience as Chief Scientist for the Secretary of the Air Force, in which he recounts his "one big report" called It's A Software-First World.

Especially noteworthy was Feigenbaum's observation about the domain-specific nature of AI:

I don’t believe there is a general pattern recognition problem. I believe that pattern  recognition, like most of human reasoning, is domain specific. Cognitive acts are surrounded by knowledge of the domain, and that includes acts of inductive behavior. So I don’t really put much hope in “general anything” for AI. In that sense I have been very much aligned with Marvin Minsky’s view of a “society of mind.” I’m very much oriented toward a knowledge-based model of mind. 

◦

Feigenbaum on Domain-Specific Knowledge and Pattern Recognition