Table 7.

Dictionary lookup performance.

This table shows the speed and accuracy of dictionary lookup tasks using the human gene/protein dictionary and gene/protein name snippets. F-score is the harmonic mean of precision and recall. The values in the parentheses are the threshold values in soft string matching.

Method	Precision	Recall	F-score	Average lookup time (microsecond)
Bigram similariy (0.97)	0.758	0.587	0.661	6.7 × 105
Bigram similariy (0.95)	0.691	0.592	0.638	6.8 × 105
Bigram similariy (0.93)	0.612	0.610	0.611	6.8 × 105
No normalization	0.809	0.502	0.619	7
Case normalization	0.782	0.582	0.666	8
Heuristic normalization [18]	0.730	0.657	0.692	8
Automatic normalization	0.767	0.633	0.694	29