IR system by incorporating users' document selection and real users of a bibliographic retrieval system. It reports use behavior. on Part 1 of a longitudinal study of decision making on Document selection is the endpoint of a bibliographic document use by academics during an actual research search. From there, the selected document may proceed project. (Part 2 followed up the same users on how the to the next stage: document use (reading and citing).
selected documents were actually used in subsequent stages.) The participants are 25 self-selected faculty and Figure 1 depicts three major decision points of document graduate students in Agricultural Economics. After a refuse along the stages of document seeking and use during erence interview, the researcher conducted a search of a research project. The solid box indicates the process in DIALOG databases and prepared a printout. The users which the decision about a retrieved document is made. selected documents from this printout; they were asked to read and think aloud while selecting documents. Their The dotted box depicts the decisions about reading and verbal reports were recorded and analyzed from a utilityciting the retrieved document in later stages; reading and theoretic perspective. The following model of the deciciting interweave. This research aims at the decision maksion-making in the selection process emerged: docuing processes across these different stages in a real life ment information elements (DIEs) in document records situation. provide the information for judging documents on 11 criteria (including topicality, orientation, quality, novelty, Previous studies that have examined real users on and authority); the criteria judgments are combined in either document selection or citation motivations are, an assessment of document value along five dimensions
among others, Barry (1993), Bonzi andSnyder (1991), (epistemic, functional, conditional, social, and emotional Brooks (1986), Liu (1993), and Park (1992). In this values), leading to the use decision. This model accounts for the use of personal knowledge and decision study, data from real users in real situations were collected strategies applied in the selection process. The model at both the outset of their projects and the conclusion of has implications for the design of an intelligent docuthe same projects. The resulting longitudinal data reflect ment selection assistant.