| "Relevance" is the key abstract concept in information retrieval and information service. Because the purpose of information retrieval and service is to retrieve and provide documents "relevant" to the user, the abstract concept is implied or explicated in an information retrieval and service. In this study, users' problem descriptions and relevance judgments in 24 cases and 240 documents are observed in order to find several dimensions of relevance and variables which are affecting dimensions of relevance.; Building on the inferential communication model, the notion of inferential relevance is introduced. Through observations on the problem description and relevance judgment, it is found that the meaning of a document, the relevance, and the use of a document are inferred by the user in the context of the user's situation and problem. As a result, existing information retrieval models, which implicitly assume that the consistency of the linguistic message implies the consistency of relevance, need to be reconsidered in the framework which recognizes that both the meaning and the relevance of a document are inferred by the user.; The findings from this study are (1) the inferential nature of relevance, (2) dimensions and orientations of relevance, (3) 9 traits of relevance, (4) problem variables, (5) document variables, (6) document use variables, (7) interactions among the document variable, the problem variable, and the use variable, and (8) complex but systematic effects of these variables on relevance dimensions.; Also the model of inferential information retrieval is built, based on these findings. According to these findings and the model, (1) the model of inferential information retrieval is a satisfactory framework for studying the problem of relevance and human information behavior. It is further proposed that (2) the model of inferential information retrieval is the correct framework for information retrieval and service system design. It follows that (3) the model of inferential information retrieval is the correct framework for the evaluation of an information service. This work also suggests that (4) the document variables, problem variables, and document use variables found in this study should be applied to the representation of both documents and problems, and that (5) a matching algorithm should be built to take into account the relevance dimensions and orientations found in this study. |
