A comment on the recognition of hand-printed two-dimensional mathematical expressions

摘要

The conference paper "Computer Input/Output of Mathematical Expressions" by Professor Martin is an excellent summary of the characteristics of 2D math expressions and of algorithms for their recognition and display. Professor Martin argues that for efficiency a recognition system must be carefully programmed to allow a rather complete notation and still take advantage of constraints in the notation. In practice, this usually means "hand-tailoring" a recognition algorithm to a particular notation. Unfortunately, a limitation of most "hand-tailored" programs is that they are rather inflexible and difficult to change in any significant way. I would like to focus my comments on the pair of questions, "(1) Are 'hand-tailored' 2D recognition algorithms necessary for efficiency?, and (2) Is there an alternative?"