Enhancing Enterprise Business Processes Through AI Based Approach for Entity Extraction – An Overview of an Application

While Industries are growing strong with their digital transformation, advanced analytics are making them stronger through data driven decisions. At the same time traditional automation is getting matured and emerging as cognitive automation. In the era of Industry 4.0, handshake of business process automation, advance analytics and cognitive services have laid down a strong platform for ‘Cognitive Bots’. Enterprises can leverage the advent of powerful technologies and approaches, anticipating ultimate goal of the business through more adaptive, self-learning, and contextual applications. This paper explains one of such cognitive bot for finance department where invoices are of utmost importance for the function. The mentioned bot is intended for amount detection and verification; additionally, it can also extract various entities like organization name, location and date which contributes to perform analytics to a great extent. The application reward business in reducing turnaround time and human errors. The accuracy of specially customized trained neural model has achieved state of the art results on the current set of learning data. The proposed framework makes use of Optical Character Recognition and PDFMiner for text extraction from scanned invoices. A Quality Classifier that will reject hand written invoices for any further processing. A spaCy’s Name-Entity-Recognition predicts the amount, date, organization name and location from extracted unstructured text.