Estimated Release Date
Extraction of complex metadata from raw bank documents is paramount to support intelligent data indexing, to face the challenge of sharing info in an effective way within large organisations. Banking language is very specific and rather different from common language. General purpose semantic engines may be not effective in understanding banking related concepts. This evidence raises the need to develop innovative solutions for metadata extraction.
The asset that have been developed is based on a weakly-supervised neural methodology for creating semantic metadata from bank documents. It exploits a neural pre-training method optimized against legacy semantic resources able to minimize the training effort. The method has been tested on documents from the Italian banking community.
Field of use