Optical character recognition errors and their effects on natural language processing Export

@inproceedings{citeulike:3226232, abstract = {Errors are unavoidable in advanced computer vision applications such as optical character recognition, and the noise induced by these errors presents a serious challenge to down-stream processes that attempt to make use of such data. In this paper, we apply a new paradigm we have proposed for measuring the impact of recognition errors on the stages of a standard text analysis pipeline: sentence boundary detection, tokenization, and part-of-speech tagging. Our methodology formulates error classification as an optimization problem solvable using a hierarchical dynamic programming approach. Errors and their cascading effects are isolated and analyzed as they travel through the pipeline. We present experimental results based on a large collection of scanned pages to study the varying impact depending on the nature of the error and the character(s) involved. The problem of identifying tabular structures that should not be parsed as sentential text is also discussed.}, address = {New York, NY, USA}, author = {Lopresti, Daniel}, booktitle = {AND '08: Proceedings of the second workshop on Analytics for noisy unstructured text data}, citeulike-article-id = {3226232}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1390749.1390753}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/1390749.1390753}, doi = {10.1145/1390749.1390753}, isbn = {978-1-60558-196-5}, keywords = {ocr}, location = {Singapore}, pages = {9--16}, posted-at = {2009-09-22 08:26:24}, priority = {2}, publisher = {ACM}, title = {Optical character recognition errors and their effects on natural language processing}, url = {http://dx.doi.org/10.1145/1390749.1390753}, year = {2008} }

TY - CONF ID - citeulike:3226232 L3 - citeulike-article-id:3226232 N2 - Errors are unavoidable in advanced computer vision applications such as optical character recognition, and the noise induced by these errors presents a serious challenge to down-stream processes that attempt to make use of such data. In this paper, we apply a new paradigm we have proposed for measuring the impact of recognition errors on the stages of a standard text analysis pipeline: sentence boundary detection, tokenization, and part-of-speech tagging. Our methodology formulates error classification as an optimization problem solvable using a hierarchical dynamic programming approach. Errors and their cascading effects are isolated and analyzed as they travel through the pipeline. We present experimental results based on a large collection of scanned pages to study the varying impact depending on the nature of the error and the character(s) involved. The problem of identifying tabular structures that should not be parsed as sentential text is also discussed. CY - Singapore SN - 978-1-60558-196-5 AD - New York, NY, USA EP - 16 TI - Optical character recognition errors and their effects on natural language processing PB - ACM T2 - AND '08: Proceedings of the second workshop on Analytics for noisy unstructured text data SP - 9 KW - ocr AU - Lopresti, Daniel PY - 2008/// UR - http://dx.doi.org/10.1145/1390749.1390753 ER -