For the second set of experiments, the dishonest recommenders are assumed to give a high recommendation value between 0.8 to 1.0, thus launching a ballot stuffing attack.
The recommendation value of the TTL is the average delivery latency of the single-copy algorithm.
A certain percentage of recommenders providing the recommendation to the service provider are dishonest and gives a high recommendation value between 0.8 to 1.0, thus launching a BS attack.
In Equation 1, the square of absolute deviation from the median is taken to signify the impact of extremes, i.e., the farther the recommendation value x i is from the median, the larger the squared deviation is.
Although, in the test phase, one can obtain the recommendation value (real-value prediction from Eq. 2) of each prop for a player, and then the rank of all labels, the problem of choosing how many recommendations for the player is left untouched.
The Kaiser-Meyer-Oiklin value was 0.727, exceeding the recommendation value of 0.6 and the Bartlett's Test of Sphericity reached statistical significance (p < 0.001).
However, Ni concentrations (0.134 0.676 mg/kg) were considerably higher than the related recommendation values.
Since smart attackers can give recommendations with little bias to go undetected, we divide the range of possible recommendation values into b intervals (or bins).
It is assumed that the recommendation values that lie outside the interval defined by LCL and UCL are malicious, therefore discarded from the set of valid recommendations.
In their work, however, generated recommendation values are neither used for automatic service composition, nor do they evolve by learning from history.
