*4.3.1. Accuracy significant networks*

In accuracy significant utilization, we define α, β and γ as 2, 1, 1; however if the data accuracy is much more important than other two, we also can define α=3 or much larger. In this research, for simplify, we discuss none other but the case that significant parameter has the significance vector of 2 and the ordinary parameters are 1. According to TOI we can get the best result in fig. 10.

**Figure 10.** Tradeoffs of accuracy significant

From Fig. 10 we find that when the event generation rate is between 0.0001 and 4.0, nonaggregation is the best comparing with full and partial aggregation. When the data generation rate is between 4 and 30, the partial aggregation is the best, and the full aggregation is the best when data generation rate is larger than 30.
