Visual Knowledge Engineering

Abstract

Artificial intelligence software has, for many years, made extensive use of graphics facilities. However, knowledge engineers have not had access to visual programming tools which assist them during the critical early phases of knowledge acquisition. Moreover, during later phases of knowledge base debugging, knowledge engineers have had to work with program tracing tools (whether graphical or textual) which are inherently incapable of scaling up to the monitoring demands imposed by large, heterogeneous knowledge bases. To address these deficiencies, and to satisfy the needs of knowledge engineers throughout the software design, development, and debugging cycle, we have developed several novel visual programming and program visualization techniques aimed at knowledge engineers. Foremost among these are 1) a hypertext transcript analyzer from which conceptual models can be generated, 2) a “direct graph manipulation” sketchpad which allows the knowledge engineer to sketch out objects and relations (including control flow and rule dependencies) from which code can be generated, and 3) “dependency viewers” which allow the knowledge engineer to examine and manipulate temporal and logical rule dependencies at different levels of granularity. The paper describes how these facilities are incorporated into KEATS, The Knowledge Engineer’s Assistant, and what key themes emerge from our approach to visual knowledge engineering. © 1990 IEEE