Reading Unstructured Data

[Anonymous]

Hi On my system (Chrome based application), I have a free-text field which is user populated. The issue I have is that I need to extract data from that field, but there is no consistency / standards as to how the data is entered. It is simply a narrative which can vary depending on how the person entering the data submits it. To pick out the key data I need, is the best way to do this simply to search for specific keywords and then pull the data near this? Or is this highly prone to error and my best option is to see if I can build a standard template which is applied to this free-text field and then do the above? Thanks  

[Denis__Dennehy]

RPA requires structured data - there is no workaround for it and as you say trying to implement made up extraction rules could be seen as high risk (there may be a future freetext note that does not fit your rules).   You have two options here: 1. Design an assisted automation with handoffs between human and virtual teams.  The Vitual team cannot understand unstructured data so this bit of the process can be done by humans and the rest of the work done by robots.  2. Your account manager or the consultancy resource enabling your organisation can discuss with you Cognative options - using a 3rd party tool such as IBM Watson or Microsoft Cognative etc.  Those tools need training in your freetext notes and lots of past examples - and to use them it often requires sending your data into the cloud which your risk/security teams will need to investigate the hows of.