Collective Information Seeking: Supporting Search with Low-Cost Activity Sharing and Collaboration Scott Bateman, Carl Gutwin, Gordon McCalla University of Saskatchewan Saskatoon, Saskatchewan, Canada <firstname.lastname>@usask.ca Ryen W. White Microsoft Research Redmond, Washington, USA ryenw@microsoft.com ABSTRACT When searching for information, people often seek help from others. However, while people can benefit from communicating with others, they can usually satisfy their information needs, to some degree, without help. Because establishing explicit collaborations is often seen as onerous, there are many missed opportunities where shared experiences could save time and effort. In this work, we propose Collective Information Seeking, which integrates theory of group work with previous research in Social Navigation and Collaborative Information Seeking. We propose that by sharing and visualizing activities within groups of information seekers a low-cost form of collaboration is made possible. Shared activities provide both a form of implicit communication that can provide guidance, and provide cues and context facilitating entry into direct collaboration at a later time. Categories and Subject Descriptors H.5.3 Group and Organization Interfaces: CSCW. General Terms Design, Human Factors, Theory. Keywords Collective information seeking, social navigation, collaborative information seeking. 1. INTRODUCTION Current information seeking tools have primarily been designed for single users working alone [11]. However, even with single- user tools, information seekers explicitly communicate with others: before searching, to help describe and focus what is being looked for; during search, to receive advice on what avenues to pursue; and after search, to share what has been found [2]. Collaboration in many information seeking situations begins with an informal form of collaboration [1]. Informal collaborations are characterized by being opportunistic and discretionary [8]. Opportunistic collaborations arise as opportunities present themselves and are grounded in being aware about the activities of other people and what they might be able to provide guidance on. Discretionary collaborations are those that do not need to occur at all, as the task can be completed, to some extent, individually. Consider the situation where a user, John, wants to find a hotel in a safe neighborhood in Vancouver. He is unaware of the fact that several of his colleagues have recently travelled to Vancouver and had good experiences with their accommodations. John‟s colleagues are not around for him to quickly ask if they had any relevant Vancouver travel advice to share. John wants to get “booking a hotel” off his to-do list, so he picks the first one he finds that seems good enough. In this example, John has decided the overhead of emailing or tracking down his colleagues, on the off chance they might have something to recommend, is too high. Situations such as this lead to missed opportunities and lead users to be less efficient. In such situations users could benefit from the knowledge and guidance of others, but without the right tools, acquiring and sharing information is often too costly. In this position paper we highlight that for guidance and experiences to be widely shared between individual information seekers, systems must be designed that take into account how communication and collaboration naturally occur in group work. In particular, most communication in groups is informal; it is both unplanned and discretionary in nature. In informal collaborations, people opportunistically make use of cues in work environments, such as the documents someone has on their desk or overheard conversations, to provide an awareness of their past and current activities. Being aware of the activities others are undertaking is an implicit form of collaboration, because it can inform decision making. However, being aware of what others have done can also be used as means to enter into more direct forms of explicit collaboration [10]. This paper proposes the design of collective information seeking (CIS) systems that provide opportunities to gather detailed information on the activities of other members in a group of familiar users. The benefits to having an appropriate awareness of others‟ activities are threefold.  First, having access to others‟ activities (e.g., the searches people have recently performed) allow users to update their knowledge about what other people are currently doing and can provide advice on. For example, had a system informed John that he had a colleague who recently issued a “Vancouver hotel” query he would have known they might have advice.  Second, being aware of the activities of others may allow users to collaborate implicitly, relieving the need to enter into a higher-cost explicit collaboration. For example, a system could display the trail John‟s colleague followed after issuing his “Vancouver hotel” query. John sees that one of his colleagues had last clicked on a link for the Pan Pacific Hotel and then went to the room booking page. John decides to book the hotel, knowing that this particular colleague has a good attention to detail and that this hotel will likely meet his needs too.  Third, an up-to-date and contextual awareness of others‟ activities can provide cues and opportunities to enter into direct collaboration. For example, John may be working on other tasks, but a system updates him that another of his colleagues, Mary, is currently uploading pictures to her “Vancouver 2011” photo album. Learning that Mary has recently been to Vancouver and that she is not currently busy with work, he requests a browser sharing session with Mary. Mary agrees and