Challenges in Creating a Taxonomy for Genres of Digital Documents

Introduction: We report on one phase of a project whose aim is to discover whether and how identifying the genres of digital documents helps in a variety of information-seeking tasks (Crowston & Kwa nik, 200507). The project has three phases: I. Harvesting and identifying a test-set of webpages from journalists, teachers, and engineers, three groups that share a discourse community in which a set of identifiable tasks and genres may play a role and in which the identification of the genre of a document is likely to be important for their tasks. For each webpage we ask the respondent to identify the task, the type of Webpage (genre), the clues the respondent used to identify the genre, and the usefulness of that document to their task. II. In the second phase, presently underway, we attempt to build a faceted taxonomy of the genres identified in Phase I. This is the phase on which we focus in this paper. III. In the final phase we will test the utility of including genre information. Using the taxonomy created in Phase II, we will manipulate a simulated search environment to test the effect of genre identification on such tasks as query formulation, searching, and processing of search output.