Haoran Xu

IMG_0647_副本1 Position Visiting Scholar/PhD student
Email xu863@purdue.edu
Office LWSN 2142U#01
Website https://www.researchgate.net/profile/Haoran_Xu2

Research Interests & Expertise

My current research interests focus on data mining and privacy protection in social media. My works include user modeling, user preference analysis, privacy policy recommendation, user varients identification and privacy protection mechanism.

Select projects

Privacy in Mobile Computing

We tackle the location privacy problem in a predication way by recommending a privacy-preserving path for a requester. We consider the popular navigation application, where users may continuously query different location based servers during their movements. Based on a set of metrics on privacy, distance and the quality of services that a LBS requester often desires, a secure path is computed for each request according to user’s preference, and can be dynamically adjusted when the situation is changed.

Incentive Mechanism in Crowdsourcing

In open crowdsourcing applications, workers are often with diverse capacities, various initiatives and without restrictions, which highly influence the process of a problem being solved. So, an incentive mechanism is desired so as to promote the correct results of problems. We propose an incentive mechanism to reward those active and excellent users, which takes into account the quality and quantity of answers a user provided, as well as the problems a user solved. The performance of this incentive is evaluated on tagging systems, which also can be applied to other crowdsourcing systems.

User Preference Analysis and Personalized Privacy Protection

Social network services have been integrated into people’s daily lives not only for entertainment and leisure but also for communication and work. Therefore, quite a lot of user profiles are stored on the social service platform, as well as a large amount of user action records and their uploaded files. Privacy setting becomes an important issue to protect these data. To help users better manage their private data, we propose a user preference based privacy policy recommendation approach for the current popular privacy setting modes. We investigate user privacy preferences from their own current privacy policies and recommend similar settings to them when a new friend is added or a new item is uploaded.