Skip to main content

Table 1 Accuracy of multiple tokenizers when tested on the chemical entities of the test set Word Cosine Similarity.

From: Chemical entity extraction using CRF and an ensemble of extractors

Measure ChemSpot OSCAR4 ChemXSeer
Correct 17149 20491 17869
Split Correct 2379 1744 3190
Total Correct 19528 22235 21059
Incorrect 5823 3116 4292
Accuracy Percentage 77.03% 87.7% 83.06%