andreas@flamingo.philosophie.uni-stuttgart.de (Andreas Eisele) (11/01/90)
Please let me know if this is the wrong newsgroup for my question. We are interested in using a scanner and appropriate OCR-software to collect natural language text from books, newspapers etc. in order to do linguistic investigations on the material. However, we don't know if there is OCR-software currently available for an affordable price (< $5K) which could be used for that purpose. The system should be capable to recognize french accents, german umlaute, at a reasonable robustness (<<0.5% error rate) for typical book or newspaper texts, and it should run on a Suns 3, a MacIntosh or on an MS-DOS (386) machine. Does such a thing exist (yet) ? Did anyone try to read large amounts of book text with one of the new systems that have appeared on the marked recently? Did anyone compare different systems? Is is at all reasonable to try it? Any help will be highly appreciated. -- +--------------------------------------------------------------------------+ |Andreas Eisele Institut fuer maschinelle Sprachverarbeitung| |Tel.: +49-711-121-3128 Universitaet Stuttgart, Keplerstrasse 17 | |Fax.: +49-711-121-3141 D-7000 Stuttgart 1, Fed. Rep. of Germany | |Internet: andreas@adler.philosophie.uni-stuttgart.dbp.de | +--------------------------------------------------------------------------+