NELArgumentIdentification yields redundant Arguments

[jgilme1]

Here is an example of NEL data that is input into the corpus.

14 17 Elizabeth_II /m/0d1_f 0.8466029 15 17 Elizabeth_II /m/0d1_f 0.846602914 16 Elizabeth_I_of_England /m/02rg_ 0.007990074

During NELArgumentIdentification each of the token spans is considered as an argument.
This will yield "Queen Elizabeth II" and "Elizabeth II" as two different arguments. To reduce redundancy we can eliminate arguments that are token sub-sequences of other arguments with the same knowledge base link.