Personalised Privacy by Default Preferences - Experiment and Analysis

Abstract

In this paper, we present a novel mechanism that provides individuals with personalised privacy by default setting when they register into a new system or service. The proposed approach consists of an intelligent mechanism that learns users' context and preferences to generate personalised default privacy settings. To achieve this, we used a machine learning approach that requires a minimal number of questions at the registration phase, and, based on users' responses, sets up privacy settings associated to users' privacy preferences for a particular service. This is the first attempt to predict general privacy preferences from a minimal number of questions. We propose two approaches. The first scheme is based on the sole use of SVM to predict users' personalised settings. The second scheme implemented an additional layer that includes clustering. The accuracy of proposed approaches is evaluated by comparing the guessed answers against the answers from a questionnaire administered to 10,000 participants. Results show that, the SVM based scheme is able to guess the the full set of personalised privacy settings with an accuracy of 85%, by using a limited input of only 5 answers from the user.

Explainable Machine Learning for Default Privacy Setting Prediction

“…The aim of this chapter is to line out open research questions of the preceding papers of Nakamura et al [31,32,33] that form the starting point of the current paper.…”

“…The first requirement is to find a method to optimize the selection of the best combination of n out of the 80 features for questioning the user. Table 5 for Model 1 and Table 6 for Model 2 show that the selection is done randomly and not user-specific in the previous approaches [31,32,33].…”

“…In general, all papers aim to provide a solution to reduce the burden of privacy setting for web-based services according to the user's privacy preferences. Therefore, the first papers [31,32] propose a machine learning-based approach for default privacy setting prediction that is further improved in the third paper [33]. In the fourth paper [34] the model is tested and evaluated in an user experiment.…”

“…A tool fulfilling these requirements besides providing explainability was already proposed by Nakamura et al [31,32,33]. Their two models can automatically predict 75 unknown privacy features from 5 features received from the user with an accuracy of 82 % and 85 % and are trained and tested based on a survey with 10,000 Japanese participants.…”

“…In Section III we explain the methodology of the paper. In Section IV, we compare the previous work of Nakamura et al [31,32,33], and in Section III-C we elicit the requirements investigated in this paper. In Section V we present our approach including the requirements of Section III-C.…”