Parascript’s Intelligent Document Processing solution, FormXtra.AI 8.4, introduces significant improvements when automating the extraction of complex data.
January 30, 2023, Longmont, CO – Parascript, a software company with more than 25 years of experience in data extraction, has released a new version of its Intelligent Document Processing (IDP) solution, FormXtra.AI 8.4. This new version adds multi-language support for context-free handwriting recognition in French, German, Spanish, and Portuguese; nested tables support for better recognition and extraction of data in complex tables; and significant improvements to full-page intelligent character recognition (ICR), trainable fields based on natural language processing (NLP) algorithms, and dynamic signature field location for unstructured documents.
“Version 8.4 is huge leap forward for FormXtra.AI. Our R&D team has implemented numerous improvements to increase accuracy and efficiency when automating data extraction from traditionally difficult documents to automate. We’re excited to see the levels of accuracy achievable with documents that include a combination of handwriting and machine print alongside complex tables.” – Dawn Wood, VP Sales.
FormXtra.AI is Parascript’s premier IDP software that builds upon the original FormXtra software by enhancing it with the most innovative trends and approaches that exist in deep learning and natural language processing. FormXtra.AI has high universality, which allows the software to classify and extract data in any type of document despite format, style of text, and complexity.
Multi-language Support
FormXtra.AI 8.4 now provides context-free reading of any type of handwritten text for a broader range of Latin and Cyrillic-based languages than ever before. Vocabulary-free data extraction is now available for documents written in English, French, German, Russian, Spanish, and Portuguese.
Full-Page ICR Improvements
Automating the extraction of a full page of handwritten and machine-printed data at high accuracy levels is complicated, but now, this complex automation task is more accessible than ever before. In FormXtra.AI 8.4, the Parascript development team has improved its proprietary ICR technology to achieve better performance when recognizing and extracting full-page data. This full-page extraction capability is complemented by artificial intelligence for improvements to accuracy over time.
Nested Tables Support
This innovative solution allows FormXtra.AI 8.4 to fully automate data extraction from complex tables, including tables with irregular structure and tables nested within tables. It is ready to use on any type of document (structured, semi-structured, and unstructured), and its wide implementation capabilities make this feature a universal capability for use in many applications and many industries.
Trainable Fields
FormXtra.AI 8.4 includes enhanced capability when locating and extracting paragraphs of text in documents with meaning similar to the meaning of paragraphs used for training. This capability implements a proprietary approach to NLP technology that can be efficiently trained on a limited set of samples (3-50). Using this NLP technology, the software can efficiently solve challenging tasks such as extracting key contractual terms in a legal document, extracting named entities in unstructured documents, locating paragraphs related to the legal description of property within contracts, provide sentiment analysis, and much more.
Parascript
Parascript’s software saves companies over $1 billion annually. Parascript state-of-the-art software incorporates proprietary AI technologies to provide robust data capture solutions that bring the highest levels of accuracy when processing documents. With over three decades of experience applying AI to solve complex problems, Parascript can automate document-oriented processes in structured, semi-structured, and unstructured formats, drastically reducing human involvement. Parascript has automated the postal industry, mortgage, payment processing, and hundreds of other processes.