Potential for Advanced Computerized Aids for Comprehensible Writing of Technical Documents

摘要

It is widely agreed that technical documents for equipment are poorly written and hard to comprehend. This has been a long-standing problem because the information-processing demands of editing and revision are so high that many comprehensibility problems go undetected. However, many of these problems can be detected by competurized systems that scan a document and point out where the writing can be improved. Existing systems of this type are based on conventional writing customs, rather than on the research literature on comprehension, and give poor advice or miss important problems. They also do not process the input document to any depth. An approach to advanced writing aids is described; such a system would base its criticisms on what is known about the cognition psychology of comprehension, and would make use of techniques from Artifical Intelligence for processing the language. Some examples of the relevant research results are presented, and a demonstration system of this type is briefly described.