There is a wealth of mathematical knowledge that could be potentially very useful in many computational applications, ical expressions from scanned images. 3
but is not available in electronic form. This knowledge comes
This paper documents our present efforts to develop in the form of mechanically typeset books and journals going such a system. Our purpose in writing this paper is back more than 100 years. Besides these older sources, there threefold: are a great many current publications, filled with useful mathematical information, which are difficult if not impossible to
โข to publicize the problems and prospects for automatic obtain in electronic form. Our work intends to encode, for recognition of mathematical notation to both the matheuse by computer algebra systems, integral tables and other matical and optical character recognition (OCR) commudocuments currently available in hardcopy only. Our strategy nities; is to extract character information from these documents, which
โข to present a two-dimensional parsing method, with is then passed to higher-level parsing routines for further extracparticular attention paid to details which have been glossed tion of mathematical content (or any other useful two-dimenover in the past; sional semantic content). This information can then be output
โข to describe this project within the overall context of as, for example, a Lisp or T E X expression. We have also develcreating a fast automatic database of mathematical foroped routines for rapid access to this information, specifically mulas.
for finding matches with formulas in a table of integrals. This paper reviews our current efforts and summarizes our results Because of its interesting structure, the mathematical forand the problems we have encountered.