This paper addresses an important and interesting issue: Can we use problem solving heuristics to make a generalized mechanism that infers generalizations? Unfortunately, no good, useful solutions are given. It is proposed that we combine empirical and analytical techniques. Only the empirical ones are running on a computer, with a program called LEX. No computerization of the analytical techniques are even inferred.
The subject matter is very narrow: Symbolic Integration. No evidence is given to help generalize the concepts to any other fields. There are too many unsupported assumptions. The one good thing about the papers is the 300-plus item bibliography.