Some of the clinical files have a date format like "23 June 2010" that we are unable to capture. It would be nice to figure out a rule-based approach to identify any type of formatted date, even ones that are separated by stop words like "the 23rd of June in 2010".
Some of the clinical files have a date format like "23 June 2010" that we are unable to capture. It would be nice to figure out a rule-based approach to identify any type of formatted date, even ones that are separated by stop words like "the 23rd of June in 2010".