I am interested in personal and social aspects of information use and management with the emphasis on the combination of algorithms and HCI.  Specific projects include personalized/social information access, social media/network analysis, and adaptive exploratory search.  All in all, they focus on the relationships: relationships between systems and people (personalization) or among a group of people (social).  My research questions below are answered by empirically studying large pool of data including textual contents, social interaction of people, and user interaction with systems, using statistical or visual analysis methods.

Research Questions

  1. I.What are the important features and techniques for visually analyzing static or dynamic social networks?

  2. II.How can we analyze and visualize large social networks and their evolutions?

  3. III.How can we combine user interface (visualization) with personalization or recommendation that emphasizes the personal or social context?

  4. IV.How to use visual or exploratory methods for personal or social information access to large textual data?

  5. V.How to represent user profiles and help users to interact/control them?

Dynamic Social Network Analysis and Visualization for Social Media

I am creating tools for analyzing the evolution of dynamic social networks with the help of visual analytic techniques.  I first built a task taxonomy of the temporal network analysis in order to define what features need to be implemented within the tools.


Collaborating with the Biotracker and the Encyclopedia of Life teams, I was able to analyze dynamic networks of citizen scientists, extracted from a large set of data objects and communications (1.2 billion records or 75 gigabytes).

A tool named TempoVis was implemented using Adobe AIR that supports temporal navigation of network evolution.  I also contributed to the design of a NodeXL extension called NetEvVis that supports temporal analysis of network changes.

Information Retrieval in Personal or Social Context

I designed and implemented a personalized and social search system for educational resources, which supports the representation of personal progress and the social recommendation of educational resources.

  1. 1.KnowledgeSea Search

  2. 2.Adaptive navigation support for KnowledgeSea

Adaptive News Filtering by Open or Interactive User Modeling

During the GALE program, I worked on exploratory search/filtering engines that could process a large amount of web news texts. I have implemented three systems and conducted user studies to learn the benefits and shortcomings of the proposed ideas.  These systems crawls the live web news from multiple sources or used a news topic tracking evaluation corpus (TDT4, Topic Detection and Tracking - Phase 4).

  1. 1.YourNews - RSS feed based personalized news filtering with "Open User Models" (published in WWW’07)

  2. 2.TaskSieve - News filtering system supports the mediation between user model and user queries (published in WWW’08)

  3. 3.NameSieve - News filtering system using semantic named-entity visualization/clustering (published at IP&M in 2010)

  4. 4.Adaptive VIBE - News search system using an adaptive visualization technique for open user models (Ph.D. dissertation study and published at IP&M in 2013).


Adaptive Visualization

I have been working on implementing adaptive visualization tools and apply them to various domains. The system is called ADVISE (Adaptive Document Visualization for Education) that includes three systems: ADVISE VIBE - POI (Point of Interest) based visualization, ADVISE 2D -- a force directed visualization, and ADVISE 3D - three dimensional visualization. These tools are being applied to generic text filtering as well as educational resources.

  1. 1.QuizVIBE - Adaptive VIBE visualization for E-learning support

  2. 2.ADVISE2D - Adaptive Force directed visualization for E-learning support

  3. 3.VIBEFusion - VIBE visualization supporting mediation between user model and user queries

  4. 4.VIBEjs - VIBE in JavaScript supporting mobile devices (introduced at HCIR 2009).