In other words, the .find() method of the RegexNameFinder returns an
array of Spans where each span has identical boundaries (startIndex ==
endIndex)!
e.g
<Span [2..2)>
<Span [8..8)>
Jim
On 08/11/12 17:29, Jim - FooBar(); wrote:
Hi guys,
I think I've discovered another issue in the name-find package, this
time with the regex name-finder.
I am instantiating it and passing 10 Pattern objects but while
evaluating it on a small paragraph the tags it outputs are empty and
immediately before the recognised token...sort of like this:
<START> <END> fluvoxamine
where fluvoxamine should be inside the tag!
Also, even if it did output the term inside the tag, i don't see any
obvious way to specify what kind of tag we want (e.g person or drug or
whatever). This is problematic because it stops the regex name-finder
from being able to be combined with the other 2 name-finders that do
support this and thus their predictions can be easily combined.
any thoughts?
Jim
