Information Technology Reference
3.47GHz/12M Cache, 6 cores, 2 CPUs, 128GB DDR3 1066MHz ECC Reg
RAM, 2TB 7.2K RPM SATA Hard Drive.
- For software, the client is equiped with Windows 7 Professional 64bit OS
and the cluster installs HANA Studio and SUSE Enterprise Linux Server 11
SP3 with SAP HANA Platform SP07.
Fig. 5. The overview of experiment context
4.3 Experimental Evaluation
Two groups of experiments are conducted to measure the veracity and time
consuming of HelpMe. In the first group, we compare HelpMe with ALPR in five
aspects, sensitivity, specificity, precision, accuracy and ROC curve to evaluate
the veracity of HelpMe. And in the second group, we explore the time consuming
of HelpMe. And in all experiments, the value of k of Nb k ( ca i ) is set to 2, 3, 4
respectively and ʸ =3allthetime.
Accuracy Evaluation. Metrics of sensitivity, specificity, precision, accuracy
and ROC curve mainly indicate the veracity of HelpMe.
Comparison of HelpMe and ALPR in sensitivity, specificity, precision and accu-
racy. Sensitivity, specificity, precision and accuracy are four statistical measures
of the performance of a binary classification test. The higher sensitivity, speci-
ficity, precision or accuracy represents the more accurate correction rate.
The sensitivity and specificity values of ALPR comparing with HelpMe with
k value equals 2,3,4 are shown in Fig. 6. It could be found that the sensitivity
and specificity values of HelpMe with k equals 2,3,4 are much higher than ALPR
(e.g., the sensitivity values of HelpMe with k equals 2,3,4 are respectively 0.1943,