Mona Deshmukh, Shruti Maheshwari
Abstract: Information extraction is concerned with applying natural language processing to automatically extract required information from free form based text documents. Several machine learning techniques have been applied in order to facilitate the portability of the information extraction systems. The challenge is not just to extract data from scanned documents but also to extract it accurately. This paper describes a general method for building an information extraction system using properties such as tokenization, POS tagging, entity detection and dependency parsing along with supervised learning algorithms. In this method, the extraction decisions are lead by a set of classifiers instead of sophisticated linguistic analyses. A major problem incurred by many businesses today is insufficiency to leverage data from scanned documents and images. Whenever a business makes use of data which is to be captured from paper documents, manually entering data can impact the efficiency, system vulnerability and speed of carrying out of business. In such business cases, we need data entry automation that helps to extract data from scanned documents and automate document based business processes.
Keywords: spaCy, POS tagging, tokenization, OCR engine, open NLP