User-driven system-mediated collaborative information retrieval

Abstract

Most of the previous approaches surrounding collaborative information retrieval (CIR) provide either a user-based mediation, in which the system only supports users' collaborative activities, or a system-based mediation, in which the system plays an active part in balancing user roles, reranking results, and distributing them to optimize overall retrieval performance. In this paper, we propose to combine both of these approaches by a role mining methodology that learns from users' actions about the retrieval strategy they adapt. This hybrid method aims at showing how users are di↵erent and how to use these di↵erences for suggesting roles. The core of the method is expressed as an algorithm that (1) monitors users' actions in a CIR setting; (2) discovers di↵erences among the collaborators along certain dimensions; and (3) suggests appropriate roles to make the most out of individual skills and optimize IR performance. Our approach is empirically evaluated and relies on two di↵erent laboratory studies involving 70 pairs of users. Our experiments show promising results that highlight how role mining could optimize the collaboration within a search session. The contributions of this work include a new algorithm for mining user roles in collaborative IR, an evaluation methodology, and a new approach to improve IR performance with the operationalization of user-driven system-mediated collaboration.

On the Collaboration Support in Information Retrieval

“…The consideration of the user-driven mediation is a particular setting in CIR in which the system lets the user performing his/her task and applies dynamically the best suited adapted retrieval techniques (e.g., automatically expanding the query with speci c words or suggesting query reformulations) with respect to his/her behavior. This setting has been recently introduced in CIR [118]. Assuming that other ve elds generally adopt a constant retrieval strategy, they refer to a system-based mediation.…”

“…However, other work [89,111,[118][119][120] assumes that users might act according to asymmetric roles. These roles attempts to structure and organizes the group members in the collaborative process [66].…”

“…Instead of ranking documents according to the division of labor and the sharing of knowledge concepts, another approach is to proactively help users to perform an e ective search session by making inference on the basis of their search logs. This challenge is tackled in the literature according to two main approaches: the recommendation of users willing to collaborate [36] and the dynamic identi cation of collaborators' roles [118,122]. We detail these two approaches below.…”

“…The second category of models [118,122] is a variant of the group building ones, since it assumes that collaborators already work together and mediate their search through the coordination of their roles. The proposed approaches rely on the assumptions that (a) skill or preference di erence is one of the motivations that make users collaborate, and (b) collaborators' roles might evolve throughout the session [123] (in contrast to previous work based on xed roles [89,111,119]).…”

“…The proposed approaches rely on the assumptions that (a) skill or preference di erence is one of the motivations that make users collaborate, and (b) collaborators' roles might evolve throughout the session [123] (in contrast to previous work based on xed roles [89,111,119]). With this in mind, the authors of Reference [118] propose to leverage collaborators' real-time actions to identify the most suited roles of collaborators throughout the search session. This model opens a new vision of the collaborative search by letting the collaborators structure themselves their session and considering their actions to enhance algorithmically the retrieval e ectiveness of the search session through a role-mining approach.…”