Data Collection since 2002
Since 2002, we record professional data in large quantities. Manual data entry is despite in recent years significantly improved text recognition systems continues to be a vital important process. Because a good data quality is not possible without human control in most cases.
- Capture index data from document scans
- Address registration (Mass Data / Business / Contests)
- Full text detection (for example, for pre-press / Neusatz)
- Conversion (for example, XML conversion / e-book creation)
Why manual data entry?
The detection accuracy of modern OCR systems ranges from 80% in low templates up to 95% with very good templates (for example, computer printouts with font size 10 point). Thus, there remains even with good templates for each character an element of risk of at least five per cent, to not being properly detected and recorded. In normal handwritten documents is an automated detection – as is used for example in the analysis of questionnaires – in many cases does not make sense. The programming of such systems can therefore be very complicated and expensive and provides experience shows that only like manuscripts and corresponding stored reference data, good results. Each exceptional case can lead to a false detection and thus to incorrect data.
Data Acquisition with System
Basically the first place of our processes is scanning the documents. By doing so, a combination of manual data entry and software-based recording workflow is possible. The more Automatiserung in a project is possible, the lower is the manual effort as well as increased quality – because people make mistakes, especially in data collection.
- An early analysis provides a more accurate and faster detection process
- The monetary and time costs can be reduced by assistive OCR systems
- Difficult documents can be viewed and sorted prior to collection
- human errors are identified or preventively averted through OCR support
- The continuous recording of all documents is guaranteed to 100% (electronic balance)
- Optimal quality control by linking data collected and scanned document
- Digital archiving of documents in the form of documents scans possible
- Paper documents can be disposed of (see Logistics & Shredding)
Double-keying in data acquisition
At the highest quality standards, the so-called double-detecting system is used. We developed this system in order to implement even the most demanding projects with very high accuracy requirements. In the double-capture all the texts to be entered twice, electronically compared and checked for deviations. Each record, in which only one letter different in the two detection results is verified by a third typist. By this method we achieve an accuracy of up to 99.98%. Depending on the requirements can be guaranteed by triple detection accuracy to 99.999%. Even with handwritten originals, an old German font, Sütterlin and documents until the 16th century, it is imperative to make use of our experience in many cases to manual recording methods.