Jintae Lee is the Chair of the Management and Entrepreneurship Division, Leeds School of Business, at the University of Colorado in Boulder. His research interest is in knowledge representation and management, text mining, and process modeling. His articles have been published in Management Science, MIS Quarterly, Human Computer Interactions, among others. He was also a guest editor for two special issues on ontology engineering, including one in Communications of the ACM.
My research interest is in knowledge management, in particular on knowledge sharing processes within and across companies. My Ph.D. work was on design rationale management, i.e. capture and use of design knowledge. Since then I have worked on the Process Interchange Format project, whose goal was to develop a common process ontology for sharing process knowledge; the Process Handbook project, which has been creating an electronic clearinghouse of process models for the purpose of reusing and sharing them; and the Design Space of Knowledge Sharing Processes whose goal is to compile, represent, and organize knowledge sharing processes.
My recent research focus is on the use of text mining to study the role of concept in knowledge management--how a given concept differs across organizational and national cultures, how the similarities and differences among them can be identified and represented, how a given concept evolve over time, and how these differences affect knowledge sharing.
For science to proceed incrementally, researchers must be able to find and deploy all research related to their own. At present, however, scientists cannot possibly meet this requirement, because the research in their fields is overwhelming, and no existing approach to combing through it can identify the full range of related papers. In the biomedical sciences, funding agencies, including the NSF, have recently takennote and have funded large projects for the collection and automatic analysis of textual data. In this proposal, we seek funding from the NSF for a project that will collect and automatically analyze textual data in the behavioral sciences.
By focusing on the most atomic level of behavioral research—namely, the constructs measured in questionnaire research—we propose to transform the way behavioral scientists locate relevant research during the literature review process. As defined by Cronbach and Meehl, a construct is “an intellectual device by means of which one construes events. It is a means of organizing experience into categories” (1955, p. 464). Unfortunately, because different researchers use different words to identify similar constructs, finding related constructs and theories eventually becomes a labor?intensive process requiring specialized knowledge, when it is even possible. We propose a solution to this problem.
We propose to collect the questionnaire items that represent constructs and discover the overlapping network among them by using four different approaches, including Latent Semantic Indexing (LSI). Weplan not only to make clear what the fundamental items (or constructs) actually are, but also to specify the interlocking systems of theoretical constructs that exist across theories (or nomological networks) byconstructing what we call an inter?nomological network (INN).
As Larsen (2003) has demonstrated, when behavioral scientists are unable to deal with the enormous stores of current research, the integrity of their fields is threatened. By producing maps and overviews, behavioral scientists will be able to establish new relationships among constructs, which will increase our understanding of the individual constructs. Moreover, by using these maps in association with an INN of the sort that we will create, researchers will be able to extend theories in the behavioral sciences well beyond what is currently possible. Thus, we believe that this project, if funded, will produce the following benefits, elaborated further in the rest of the proposal