Evaluating the synergic effect of collaboration in information seeking

It is typically expected that when people work together, they can often accomplish goals that are difficult or even impossible for individuals. We consider this notion of the group achieving more than the sum of all individuals' achievements to be the synergic effect in collaboration. Similar expectation exists for people working in collaboration for information seeking tasks. We, however, lack a methodology and appropriate evaluation metrics for studying and measuring the synergic effect. In this paper we demonstrate how to evaluate this effect and discuss what it means to various collaborative information seeking (CIS) situations. We present a user study with four different conditions: single user, pair of users at the same computer, pair of users at different computers and co-located, and pair of users remotely located. Each of these individuals or pairs was given the same task of information seeking and usage for the same amount of time. We then combined the outputs of single independent users to form artificial pairs, and compared against the real pairs. Not surprisingly, participants using different computers (co-located or remotely located) were able to cover more information sources than those using a single computer (single user or a pair). But more interestingly, we found that real pairs with their own computers (co-located or remotely located) were able to cover more unique and useful information than that of the artificially created pairs. This indicates that those working in collaboration achieved something greater and better than what could be achieved by adding independent users, thus, demonstrating the synergic effect. Remotely located real teams were also able to formulate a wider range of queries than those pairs that were co-located or artificially created. This shows that the collaborators working remotely were able to achieve synergy while still being able to think and work independently. Through the experiments and measurements presented here, we have also contributed a unique methodology and an evaluation metric for CIS.