PPCF


Also found in: Acronyms.
References in periodicals archive ?
We proposed the two-fold PPCF solutions for two different CF approaches and then proposed schemes to make two parties collaborate on partitioned data with rating overlaps.
In Table 2, we display the outcomes related to user-based schemes and corresponding gains obtained by PPCF schemes with respect to single evaluation of CF in percentages where Gain(X) = 100 x ([MAE.sub.Singie] - MAEX)/[MAE.sub.Single] and MAEX stands for the obtained MAE value from method X.
To eliminate privacy risks posed by online vendors utilizing CF schemes, such as price discrimination, profiling, and unsolicited marketing, researchers have developed PPCF techniques [9][22].
We also propose a new attack design mechanism, which is more suitable to randomized perturbation-based PPCF schemes, and evaluate two varieties of well-known neighborhood-based prediction algorithms.
The prediction algorithm proposed by [16], which is also the algorithm used in PPCF frameworks, employs a variance weighting onto neighbors via z-score normalization.
To achieve confidentiality in PPCF schemes, confidential data are masked using a variety of data disguising methods.
PPCF applications collect disguised data from users to protect customers' privacy.
We conducted real data-based experiments to evaluate the effectiveness of our modified shilling attack models on two memory-based PPCF algorithms.
5.3.1 Evaluating the Effects of Push Attack Models in PPCF Schemes