outline Institute for Software Research motivation A Software - PDF document

Institute for Software Research outline Institute for Software Research � motivation A Software Product Line Approach � software product line for Handling Privacy Constraints � our privacy-enabling user modeling architecture in Web Personalization � an example Yang Wang and Alfred Kobsa � conclusion University of California, Irvine, USA UM05 Workshop on Privacy-Enhanced Personalization Edinburgh, Scotland A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 2 Privacy Constraints in Web Personalization Institute for Software Research Institute for Software Research motivation aims and key characteristics � web personalization benefits � provide optimal personalization while respecting both customers and vendors privacy laws, regulations and users’ personal privacy preferences � privacy concerns counteract � leverage the flexibility of software product line to the personalization benefits address combinatorial complexity of privacy � how to dynamically balance constraints personalization and privacy? � apply state-of-the-art industry practice for managing software variants at run time A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 3 PEP05 Yang Wang and Alfred Kobsa 4 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization Institute for Software Research Institute for Software Research UniversalFriends.com privacy concerns � bridge physical complex combinations of: distances � users’ personal privacy preferences � foster universal � international privacy laws (even trans-border) friendship � privacy regulations � recommend personalized list of likely friends A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 5 PEP05 Yang Wang and Alfred Kobsa 6 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization � 1

Institute for Software Research Institute for Software Research users’ privacy preferences privacy laws � EU data protection directive � Asia-Pacific Economic Cooperation (APEC) privacy framework “do not store � Organisation for Economic Co-operation and “only collect my my true name !” Development (OECD) privacy guidelines information if I give � over 30 countries already have their own explicit consent !” privacy laws and more countries are coming � “do not track me !” A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 7 PEP05 Yang Wang and Alfred Kobsa 8 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization Institute for Software Research Institute for Software Research data protection laws – year 2003 privacy regulations � sector standards (e.g., in USA) � medical: HIPPA � children: COPPA � finance: Gramm-Leach-Bliley Act � self-regulation policies � TRUST e � Network Advertising Initiative � Chinese E-Commerce Trust Consortium A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 9 PEP05 Yang Wang and Alfred Kobsa 10 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization Institute for Software Research Institute for Software Research the research question revisit the research question seeking a mechanism to dynamically select user modeling components how can personalized web-based systems that comply with maximize the personalization benefits the currently prevailing privacy constraints while at the same time being compliant with the privacy constraints that are currently in effect? A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 11 PEP05 Yang Wang and Alfred Kobsa 12 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization � 2

Institute for Software Research Institute for Software Research user modeling component pool product line architecture � “The common architecture for a set of related products or user modeling data used systems developed by an organization.” [Bosch, 2000] component methods used demographic user-supplied visited data data pages UMC 1 clustering X � a PLA includes rule-based reasoning UMC 2 X UMC 3 fuzzy reasoning with uncertainty X � Stable core: basic functionalities rule-based reasoning UMC 4 X X � Options: optional features/qualities X X UMC 5 fuzzy reasoning with uncertainty � Variants: alternative features/qualities X X incremental machine learning UMC 6 X XXX UMC 7 one-time machine learning � A particular architecture instance is selected from the X XXX UMC 8 one-time machine learning + fuzzy reasoning with uncertainty product-line architecture A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 13 PEP05 Yang Wang and Alfred Kobsa 14 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization Institute for Software Research Institute for Software Research our approach an example A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 15 PEP05 Yang Wang and Alfred Kobsa 16 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization Institute for Software Research Institute for Software Research the privacy constraints A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 17 PEP05 Yang Wang and Alfred Kobsa 18 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization � 3

Institute for Software Research Institute for Software Research conclusions project status and future work � provide optimal personalization while respecting � currently prototyping the system privacy laws, regulations and users’ personal � systematically express privacy constraints privacy preferences � a decentralized version � leverage the flexibility of software product line to � privacy kernel address combinatory complexity of privacy constraints � apply state-of-the-art industry practice for managing software variants at run time A Software Product Line Approach for Handling A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 19 PEP05 Yang Wang and Alfred Kobsa 20 Privacy Constraints in Web Personalization Privacy Constraints in Web Personalization Institute for Software Research acknowledgements � Alfred Kobsa � André van der Hoek � Eric Dashofy � Yun Huang � Norman Su � Colleagues at CommerceNet A Software Product Line Approach for Handling PEP05 Yang Wang and Alfred Kobsa 21 Privacy Constraints in Web Personalization � 4