Research Projects

2016 ~ 2018: Research on Eliciting Users’ Implicit Feedback for Constructing Their Multi-Attribute Preferences in Complex Decision Support (GRF, principal investigator)

2013 ~ 2015: Research on Incorporating Feature-level Opinion Mining Outcomes into Modeling/Eliciting Potential Customers’ Multi-criteria Preferences and Generating High-value Products’ Recommendation in E-commerce (GRF/ECS, principal investigator)

2012 ~ 2013: Inference Learning of Users’ Self-Comparison Process in Recommender Systems (FRG, principal investigator)

2011 ~ 2012: Studying the Role of Social Opinions in Building Explanatory Mechanism for Utility-based Recommender Systems (FRG, principal investigator)

2009 ~ 2011: Cross-Cultural Usability Evaluation and Interface Design in Recommender Systems (FRG, principal investigator)

2003 ~ 2008: Modeling and Elicitation of Decision Parameters in Personal Information Agents (Swiss National Science Foundation, co-investigator)

Main Results

Increasing User Trust in Online Recommender Websites with Explanation Interfaces. I have developed a trust model specialized for product recommender systems. I have revealed the relationship between users’ competence perception and their trusting intentions such as intention to purchase and intention to return, and exposed the positive effect of explanation interfaces on trust promotion. I have further proposed a new approach, the preference-based organization interface, for generating recommendation explanations, and demonstrated its significantly positive impact on building users’ trust and their behavior intentions. In addition, the preference-based organization interface was found not only performing as a more effective explanation technique, but also more accurate algorithm for computing tradeoff suggestions that can guide users to more efficiently locate their desired choices with a relatively lower level of effort compared to related methods.


Improving User Decision Accuracy via Critiquing-based Recommender Agents. I have designed and developed intelligent tradeoff supports, called example critiquing aids. By means of real-user studies, I have found that users’ decision accuracy can be significantly improved after performing tradeoffs via the example critiquing support. In addition, I have compared two typical types of critiquing aids: user self-motivated critiquing and system-proposed critiques, and identified their respective pros and cons. I have further developed a hybrid critiquing-based recommender system to combine the two approaches’ strengths together, and proven its outstanding performance through a cross-cultural validation. An evaluation framework has been additionally established, involving objective/subjective measurements for decision accuracy and decision effort. ec

2002 ~ 2003: COMMIX, a content oriented massive information integration based on XML (China National Key Fundamental Research and Development 973 Plan, researcher)

Main Results: I designed and developed a selective dissemination of information system that was to dynamically and automatically return new or updated XML documents matching to users’ pre-stated preferences (e.g. online updated news or products in XML format sent to users’ email boxes or mobile phones). I employed advanced index techniques on XPATH queries to improve the search efficiency. The new algorithm reduced time complexity from O(n3) to O(n).

2000 ~ 2001: eCOBASE, an embedded and mobile DBMS based on COBASE (China National Advanced Technology Research and Development 863 Plan, researcher)

Main Results: I was engaged in designing and developing data synchronization and replication subsystem in eCOBASE on WinCE platform, enabling it to effectively exchange data with the stable central database (e.g. SQL Server or COBASE).