Table 3.

Performance of NegScope, BaselineScope and the NegEx system on the Negex testing data. NegScope and BaselineScope were trained on both biological and clinical sentences

	NegScope					BaselineScope		NegEx
Features used	Words	Part of speech	Part of speech	Part of speech	Part of speech	Words	Words	–
Cue phrase identified using	–	NegCue	NegCue	BaselineCue	BaselineCue	BaselineCue	BaselineCue	–
Cue phrase replaced	–	No	Yes	No	Yes	–	–	–
Scope limited by	–	–	–	–	–	Comma and period	Period only	–
Both biological and clinical sentences used for trainining
Recall	75.3±6.86	67.89±7.30	67.11±8.34	81.13±5.25	83.83±4.93	75.94±5.79	85.24±4.77	96.27±3.08
Precision	90.48±3.57	97.21±2.87	96.48±3.77	90.54±2.68	87.16±3.51	78.9±4.88	76.22±4.39	95.43±2.02
F1-score	82.12±5.30	79.79±5.64	78.99±6.81	85.53±3.76	85.42±3.82	77.29±4.41	80.44±4.13	95.82±1.91
Accuracy	93.35±1.57	93.06±1.38	92.8±1.75	94.4±1.20	94.15±1.34	90.87±1.46	91.5±1.59	98.27±0.78
Clinical sentences used for training
Recall	36.88±5.42	58.52±6.48	70.33±5.93	59.13±6.53	72.0±5.76	63.61±6.77	80.06±5.37	96.27±3.08
Precision	94.59±5.43	96.59±2.72	95.07±3.03	93.36±2.36	91.64±2.38	80.46±6.14	81.59±3.87	95.43±2.02
F1-score	52.92±6.10	72.73±5.32	80.72±4.41	72.22±4.99	80.55±4.17	70.99±6.28	80.78±4.39	95.82±1.91
Accuracy	86.53±1.79	91.04±1.48	93.18±0.94	90.7±1.46	92.93±0.94	89.35±2.06	92.21±1.45	98.27±0.78

Value after ‘±’ indicates SD. Values in bold represent the best performance.