Context-Aware K-Value Learning Model in Location K-Anonymitys

With the extensive application of Location-Based Service (LBS) and the enhancement of people's awareness of privacy, location k-anonymity is more and more widely used. However, the use of location k-anonymity, whether unified k-value or personalized k-value, requires users to define k-value or specified user privacy profile, such as maximum privacy leakage, minimum quality loss, etc. Privacy is a feeling, and it is hard to quantify and directly specified by the user, and the specified value may not satisfy the user's actual demand. So, this paper firstly analyzes the factor that affects privacy and then designs a questionnaire survey according to the query theory, which is taken as the input to the context-aware k-value learning model proposed in this paper. In this model, the user's privacy profile in the questionnaire is used to obtain the user
$\mathrm{P}_{\max}$
and QoS requirements, etc., Subsequently, the user privacy profile is constructed and verified, the theoretical k-value is calculated by optimal the k-selection algorithm, and a feedback mechanism is formed to make the model more adaptive. The model proposed in this paper provides a technical approach for quantifying users' privacy feelings and provides a reference for users to specify their privacy preferences based on evidence. At last, the effectiveness of the proposed model is verified by privacy profile validator.