[CTAKES-133] constituency-parser: escaped types confused for function tags

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Component/s: ctakes-constituency-parser
• Assignee: Tim Miller
• Reporter: Tim Miller
• Due: 22/Jan/13
• Created: 22/Jan/13 16:17
• Updated: 22/Jan/13 19:19
• Resolved: 22/Jan/13 19:19

Description

There is code to separate constituent labels from function tags - e.g., NP-SBJ for a noun phrase (NP) that is a subject (SBJ). There are other tags that are special cases to escape things that screw up tree formatting – a paren in the text "(" is rewritten as LRB to not be mistaken for a tree delimiter. But the code looking for the dash in a label-function tag will find the dash in the LRB. The fix is to check for a dash at the start of the label – escaped tags will not have function elements.