Towards a unified framework of IR tasks and strategies

Despite huge advances in making information accessible to vast numbers of users, the effective retrieval of relevant information remains a challenge. User studies of online library catalogues, abstracting and indexing systems, and web search portals, repeatedly show that despite knowledge of basic search techniques, many users do not acquire strategies to find relevant information effectively. The phenomenon of the ineffective use of retrieval systems despite experience, is not unique; numerous user studies of other complex computer systems such as word processors, spreadsheets, and CAD systems show that despite experience with basic tools, users do not progress to a more effective use of these systems. The effective use of computer systems is essential for improving the overall productivity of information workers and has been in the forefront of our research efforts. This research has led to a framework of effective and general strategies to use complex authoring systems. The framework has been used to develop a new training approach to teach strategic knowledge that has been tested in three universities, and has affected the learning of more than 150 students. Unfortunately, such a unified framework of effective and general strategies is currently lacking in the field of Information Retrieval (IR). This paper presents our current work and a detailed proposal to build a framework of effective and general strategies based on the following approach: (1) develop a taxonomy of IR tasks base on tasks of real users; (2) develop a taxonomy of strategies based on general and effective IR strategies; (3) develop a descriptive model of expert performance; (4) develop a prescriptive model of effective performance; (5) apply the framework to design a curriculum of effective training of IR strategies. We have begun to perform the above research through a multidisciplinary effort of researchers from three critical domains: Human-Computer Interaction, Library Science and Information Retrieval. The research will consist of the analysis and categorization of reference questions from real users, experiments involving the observation and modeling of experts, and controlled experiments to evaluate the efficacy of teaching effective IR strategies to university students. The proposed work has the potential of broad impact in three areas: (1) development of an explicit user-based approach to model information retrieval; (2) an approach to train users to be effective retrievers of information; (3) a systematic method to guide designers to identify and implement functionalities that enable users to execute effective IR strategies.