Art. 05 – Vol. 26 – No. 4 – 2016
Mihnea Horia Vrejoiu
National Institute for Research & Development in Informatics, ICI Bucharest
Abstract: Manually inputting data from paper support represents, on one hand, a necessary and important activity in the context of the widely spread informatization in most areas, but, on the other hand, so human, and time, resources consuming, and also a potential source of typing errors, especially when large volumes of data are inputted. In this context, techniques and tools by which this activity can be automated are extremely useful. However, often, due to critical requirements on the accuracy of the inputted data, it is absolutely necessary at least a check and validation of these data by the human operator in the absence of other criteria and/or automated validation possibilities. In this context, it has been proposed and experimentally implemented a semi-automatic software solution based on OCR/ICR techniques, intended to assist the work of inputting the data filled in standardized forms, with fixed format, for improving the efficiency, productivity and accuracy. Testing experiments were conducted, there were made observations on the functioning and results, there were summarized some conclusions and possible further improvements and optimizations.
Keywords: data inputting, standardized forms, machine learning, supervised learning, OCR/ICR, regular expressions.