Nurzat Rakhmanberdieva

Title: Nurzat Rakhmanberdieva: Patent Innovation in Document Data Extraction

Introduction

Nurzat Rakhmanberdieva is an inventor associated with Berlin, Germany, and SAP SE. His patent work addresses how machine-learning systems can assess the reliability of information extracted from documents.

Latest Patents

Rakhmanberdieva is associated with 1 patent, titled “Model-Independent Data Extractor Confidence Value Prediction Machine Learned Model.” The invention describes a separate machine-learning model that predicts a confidence score for a label assigned to information extracted from a document. Its Sliding-Window method can use convolutional neural network classification and considers the extracted text, its predicted class label, its coordinates in the document, and the document text for context. The resulting score helps determine whether the predicted label is correct.

Career Highlights

Rakhmanberdieva’s listed patent work connects him with SAP SE and focuses on improving confidence assessment in document data extraction. The approach uses a model distinct from the model that initially predicts the extracted information’s label.

Collaborations

His listed coworkers include Alexey Streltsov and Christian Reisswig.

Conclusion

Nurzat Rakhmanberdieva’s patent contribution centers on using contextual, machine-learned confidence scoring to evaluate labels assigned to information extracted from documents.