Structured Information Extraction from Unstructured Documents

The constant development and use of information technology to digitize and digitalize business processes leads to an increasing amount of data available in various formats. This data comes in two main forms, namely structured and unstructured. Nowadays, around 80% of data in organisations is unstructured (Grimes, 2008). Insurance policy documents are a good example of this kind of data, with a lot of text and domain specific language. It is difficult to leverage this data as there is no clear structure and the massive amount of data makes it too time consuming to analyse it manually. Over the past years, the need to handle unstructured data has arisen and that’s why many researchers are trying to tackle this challenge.Currently, it is not possible to automatically extract structured information from insurance policy documents. Insurance brokers analyse every policy by hand and search for relevant terms. This information is then used to compare different insurance quotes and to offer the best combination to the customer. Instead of spending their time in consultancy and helping their clients find the best insurance solution, the brokers lose a lot of time in extracting relevant information....