Nodes of Topicality: Modeling User Notions of On Topic Documents Howard Greisdorf Texas Center for Digital Knowledge, University of North Texas, P.O. Box 31168, Denton, TX 76203-1068. E-mail: hfg0001@comcast.net Brian O’Connor Texas Center for Digital Knowledge, University of North Texas, P.O. Box 31168, Denton, TX 76203-1068. E-mail: boconnor@lis.admin.unt.edu Topicality, while demonstrably an empirically manage- able variable of investigation, engenders aspects of cog- nitive complexity that may, or may not, be easily man- aged during user interactions with IR systems. If an item retrieved from an IR system is considered to be on topic by a user, the meaning of that judgment may imply other underlying criteria. What makes an item on topic for users is the subject of this investigation. Although topi- cality has served to generate a great deal of attention in the body of information science literature, the meaning of topicality to IR system users has suffered from a lack of full understanding in designing more effective ap- proaches to information search and retrieval. This inves- tigation takes an inductive approach to the deductive extraction of characteristics that describe and explain how items retrieved from interactions with IR systems can be considered as on topic. Introduction In the field of information science the concept of topi- cality has remained a concern and variable of investigation due to its conceived manifestation of binding a user’s need for information with an IR system that may have the poten- tial for resolving that need. A difficulty that continues to plague the nature of that binding operation, however, is a user’s cognitive engagement with a retrieved item that oc- curs after the mechanical engagement of the system has been completed. Although these engagements have been the subject of numerous investigations, the underlying concep- tual frameworks have mostly been focused on a more ade- quate understanding of relevance as a key evaluative mea- sure of retrieval success. The nature, manifestations, and effects of topicality as a mediating variable in the search, retrieval, and evaluation process have been continuously addressed, but a model of topicality from user perspectives has never emerged. This investigation attempts to uncover some of the aspects of topicality that can serve to model what on topic means to a user during a search for informa- tion. Although other variables may contribute to stricter definitions of topicality, the approach engendered by this investigation is to tie user subjectivity to some of the ob- jective components of formal logic that have been used to structure retrieval systems. What does on topic mean? Cooper (1971) was among the first to identify topicality as one of the “stickiest points” to deal with in connection with information retrieval. From his perspective, a topic merely indicates a subject area of in- terest without actually asking any specific questions about it; and, the appropriateness of that topic has to do with whether or not a piece of information is on a subject that has some topical bearing on the information need in question. However, from a semantic perspective the concepts of “sub- ject of interest” and “topical bearing” imply some form of functional gap that exists (and must be filled) between the information needed and the item(s) retrieved to complete the interaction. In their discussion of relevance, Robertson, Maron, and Cooper (1982) posed that when users interact with an IR system for the purpose of retrieving information concerning a problem at hand, three nodes of engagement exist during that interaction. The first node of engagement is the formu- lation of a query that in some way represents the problem under investigation. The second node of engagement is how the system interprets that formulated query to return items of information to the user; and the third node represents how the user cognitively interprets the items of information returned by the system in relation to the problem at hand. Received November 14, 2002; revised May 1, 2003; accepted May 1, 2003 © 2003 Wiley Periodicals, Inc. JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY, 54(14):1296 –1304, 2003