Some similarities and differences between intellectual and machine text understanding for the purpose of abstracting
Rainer Kuhlen · KOPS (University of Konstanz) · 1984
The analysis of a set of intellectual abstracting rules taken from actual practice in working systems in West-Germany leads to the conclusion that, so far, there is no way of real machine simulation of intellectual text abstracting capacity, due to the significant differences in machine and human analysis and condensation techniques. But, on the other hand, there is no evidence for the traditional belief that surface oriented text analysis procedures, mainly statistically oriented, can produce high quality abstracting. There is a need for knowledge-based text analysis and processing (inferencing and transformation). As a result of the survey of abstracting rules requirements for an automatic abstracting system are formulated.