Jihie Kim's Home Page

Education

Research

• PedDiscourse (Pedagogical Discourse Analysis). Assessing student activities in threaded on-line discussions and providing answers to student queries. (from the National Science Foundation, CCLI grant [Award #0618859]) See more details in these papers:
  • An Intelligent Discussion-Bot for Answering Queries in Threaded Discussions (IUI 2006).
  • Learning to Detect Conversation Focus of Threaded Discussions (HLT-NAACL 2006).
  • Towards Modeling Threaded Discussions through Ontology-based Analysis (AAAI 2006).

• Semantic Approaches to Workflows:
  • CAT (Composition Analysis Tool). CAT facilitates interactive construction of workflows (computation pathways) where users select and connect existing workflow components, and the system interactively gernerates assistance in completing a correctly formulated pathway. See more details in this paper(IUI 2004) and this site.

  • Wings (Workflow INstance Generation and Selection). We present an approach to workflow creation and validation that uses semantic representations to describe complex scientific applications in a data-independent manner, then automatically generates workflows of computations for given data sets, and finally maps them to available computing resources. We have used this to create workflows of thousands of computations, which are submitted to the Pegasus mapping system for execution over grid computing environments. See more details in this paper(ISWC 2006).

• ECHO/ SLICK (Skills for Learning to Interactively Capture Knowledge). Developing acquisition interfaces that are proactive learners, able to reason about learning activities and with initiative in participating in the process accordingly. See more details in this article (IJHCS 2007), this paper(AIEd 2003) and these sites: Slick and Echo.

• KANAL (Knowledge ANALysis). A Tool for checking process models entered by users. By relating different pieces of information in process models among themselves and to the existing KB, it performs a variety of verification and validation checks and propose useful fixes. See more details in this paper (IJCAI 2001) and this site.

I have also worked on

• EMeD(Expect Method Developer). A tool to guide knowledge base creation based on interdependency analysis. See more details in this paper(AAAI 1999) and this site.

• KA (Knowledge Acquisition) Evaluation Methodology. See more details in this paper (JETAI article 2001).

• Active Catalogs that provides on-line catalogs augmented by behavioral models and their consumption environment, significantly enhancing the engineering and design support that is desirable but beyond the reach of current engineering environments. See more details in this paper (AI EDAM article 2004).

• Utility Problem in Soar (Expensive Chunks). My dissertation research centers on application of machine learning techniques to speed up problem solving. Many learning systems suffer from the utility problem; that is, that time after learning is greater than time before learning. Discovering how to assure that learned knowledge will in fact speed up system performance has been a focus of research in explanation-based learning (EBL). One way of finding a solution which can guarantee such cost boundness is to analyze all the sources of cost increase in the learning process and then eliminate these sources. I began on this task by decomposing the learning process into a sequence of transformations that go from a problem solving episode, through a sequence of intermediate problem solving/rule hybrids, to a learned rule. This transformational analysis itself is important to understand the characteristics of the learning system, including cost changes through learning. Such an analysis has been performed for Soar (a problem solving system with a variant of EBL). By analyzing these transformations, I have identified a set of sources which can make the output rule expensive. Also, I have implemented modifications of the learning system based on the analysis. The work is summarized in this paper (AI Journal article)

• the YODA robot project. A mobile agent that can autonomously learn from its environment based on its own actions, percepts, and missions. (article in USC Chronicle) Also see this paper (AI magazine article).

Contact Information

Email: (my first name)@isi.edu, (my first name)@usc.edu

ISI:            University of Southern California
                Information Sciences Institute	
                4676 Admiralty Way			
                Marina del Rey, CA 90292		
                (310) 448-8769