Insurance Document Upload and Machine Learning



You can upload a demo document that contains non-life insurance information e.g. a homeowners insurance policy declarations page here .

The extraction logic is similar to that used by Bnie Insurance, the difference being that this demo provides selected raw data from the document and an indication of what is believed to be missing for the purposes of understand the insurance situation of a household. In contrast, if you register with Bnie Insurance and upload the same document, you'll get a series of insurance ideas generated to help you based on the data that you provide to the website. The extraction logic is a good example of a complex search which with hundreds of search elements, would be complex and time consuming to configure correctly without the help of various machine learning and other algorithms. One of these, we'll call it "wrong" for short, shows the elements believed to be missing based on the classification of the document. So, as an example, most but not all auto insurance policies have one or two bodily injury limits and the items in wrong might contain this entry if absent from the list of retrieved data items.