William Godwin is annotated 5 times including one on the image in Anarchism article
but the actual article text will have "William Godwin" only 4 times because articletext doesn't have image content. This will lead to discrepancies in the surface form counts with 5 annotated count and 4 total count.
Currently, these are included in links element in the JSON output with span (0,0)
Should come up with an approach to actually eliminate these.
Example
William Godwin is annotated 5 times including one on the image in Anarchism article but the actual article text will have "William Godwin" only 4 times because articletext doesn't have image content. This will lead to discrepancies in the surface form counts with 5 annotated count and 4 total count.
Currently, these are included in links element in the JSON output with span (0,0)
Should come up with an approach to actually eliminate these.